{"utterance_id": "3081-166546-0085", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0085.flac", "task_type": "lexical_detection", "query": "vigour", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"vigour\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 1.07, "vigour"]], "response_mode": "offline", "answer": "[{\"word\":\"vigour\",\"start\":0.64,\"end\":1.07}]"}
{"utterance_id": "8297-275155-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0000.flac", "task_type": "lexical_detection", "query": "arranged", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"arranged\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.13, 3.56, "arranged"]], "response_mode": "offline", "answer": "[{\"word\":\"arranged\",\"start\":3.13,\"end\":3.56}]"}
{"utterance_id": "1988-147956-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0018.flac", "task_type": "lexical_detection", "query": "noticed", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"noticed\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 0.9, "noticed"]], "response_mode": "offline", "answer": "[{\"word\":\"noticed\",\"start\":0.48,\"end\":0.9}]"}
{"utterance_id": "5694-64038-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[7.44, 7.67, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":7.44,\"end\":7.67}]"}
{"utterance_id": "652-130737-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 3.31, "eighteen ninety three"]], "response_mode": "offline", "answer": "[{\"word\":\"eighteen ninety three\",\"start\":2.13,\"end\":3.31}]"}
{"utterance_id": "8842-302203-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0008.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.3, 0.59, "second"], [2.92, 3.2, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.3,\"end\":0.59},{\"word\":\"first\",\"start\":2.92,\"end\":3.2}]"}
{"utterance_id": "6345-93306-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0025.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.74, 9.05, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":8.74,\"end\":9.05}]"}
{"utterance_id": "2428-83705-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0035.flac", "task_type": "lexical_detection", "query": "given", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"given\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.21, 3.5, "given"]], "response_mode": "offline", "answer": "[{\"word\":\"given\",\"start\":3.21,\"end\":3.5}]"}
{"utterance_id": "652-130737-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0002.flac", "task_type": "lexical_detection", "query": "clarets", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"clarets\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.19, "clarets"]], "response_mode": "offline", "answer": "[{\"word\":\"clarets\",\"start\":1.72,\"end\":2.19}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.36, "brain"], [7.67, 8.16, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"brain\",\"start\":1.9,\"end\":2.36},{\"word\":\"hands\",\"start\":7.67,\"end\":8.16}]"}
{"utterance_id": "6313-76958-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.66, 3.93, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":3.66,\"end\":3.93}]"}
{"utterance_id": "7976-105575-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0013.flac", "task_type": "lexical_detection", "query": "center", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"center\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.79, 7.2, "center"]], "response_mode": "offline", "answer": "[{\"word\":\"center\",\"start\":6.79,\"end\":7.2}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "lexical_detection", "query": "volume", "prompt": "After the audio ends, extract all occurrences of the word \"volume\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.01, 7.57, "volume"]], "response_mode": "offline", "answer": "[{\"word\":\"volume\",\"start\":7.01,\"end\":7.57}]"}
{"utterance_id": "6319-57405-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.34, 2.64, "son"]], "response_mode": "offline", "answer": "[{\"word\":\"son\",\"start\":2.34,\"end\":2.64}]"}
{"utterance_id": "174-168635-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.64, "six"]], "response_mode": "offline", "answer": "[{\"word\":\"six\",\"start\":1.31,\"end\":1.64}]"}
{"utterance_id": "2035-147960-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.34, 1.14, "otto fuchs"]], "response_mode": "offline", "answer": "[{\"word\":\"otto fuchs\",\"start\":0.34,\"end\":1.14}]"}
{"utterance_id": "2803-154328-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0023.flac", "task_type": "lexical_detection", "query": "kill", "prompt": "After listening to the whole recording, output every instance of the word \"kill\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.46, 5.85, "kill"]], "response_mode": "offline", "answer": "[{\"word\":\"kill\",\"start\":5.46,\"end\":5.85}]"}
{"utterance_id": "5895-34615-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0016.flac", "task_type": "lexical_counting", "query": "called", "prompt": "Process the full audio offline. For every occurrence of the word \"called\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.42, 10.75, "one"], [13.03, 13.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.42,\"end\":10.75},{\"word\":\"two\",\"start\":13.03,\"end\":13.38}]"}
{"utterance_id": "8297-275154-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.88, 6.42, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":5.88,\"end\":6.42}]"}
{"utterance_id": "6319-64726-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0010.flac", "task_type": "lexical_detection", "query": "awe", "prompt": "When the utterance is complete, report all instances of the word \"awe\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.22, 4.6, "awe"]], "response_mode": "offline", "answer": "[{\"word\":\"awe\",\"start\":4.22,\"end\":4.6}]"}
{"utterance_id": "3000-15664-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0020.flac", "task_type": "lexical_detection", "query": "strangely", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"strangely\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.94, 12.8, "strangely"]], "response_mode": "offline", "answer": "[{\"word\":\"strangely\",\"start\":11.94,\"end\":12.8}]"}
{"utterance_id": "6313-66125-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0024.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.81, 2.3, "picnic"]], "response_mode": "offline", "answer": "[{\"word\":\"picnic\",\"start\":1.81,\"end\":2.3}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a body part: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[10.0, 10.38, "one"], [13.59, 14.05, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.0,\"end\":10.38},{\"word\":\"two\",\"start\":13.59,\"end\":14.05}]"}
{"utterance_id": "3170-137482-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0040.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.87, 6.37, "devil"]], "response_mode": "offline", "answer": "[{\"word\":\"devil\",\"start\":5.87,\"end\":6.37}]"}
{"utterance_id": "2803-154320-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.98, 11.5, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":10.98,\"end\":11.5}]"}
{"utterance_id": "1988-147956-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0020.flac", "task_type": "lexical_detection", "query": "blown", "prompt": "After listening to the whole recording, output every instance of the word \"blown\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.18, 5.54, "blown"]], "response_mode": "offline", "answer": "[{\"word\":\"blown\",\"start\":5.18,\"end\":5.54}]"}
{"utterance_id": "84-121550-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0015.flac", "task_type": "lexical_detection", "query": "came", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"came\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.6, 6.88, "came"]], "response_mode": "offline", "answer": "[{\"word\":\"came\",\"start\":6.6,\"end\":6.88}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[4.77, 5.2, "one"], [7.41, 7.77, "two"], [10.67, 11.09, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.77,\"end\":5.2},{\"word\":\"two\",\"start\":7.41,\"end\":7.77},{\"word\":\"three\",\"start\":10.67,\"end\":11.09}]"}
{"utterance_id": "8842-304647-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0009.flac", "task_type": "lexical_detection", "query": "organ", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"organ\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.84, "organ"]], "response_mode": "offline", "answer": "[{\"word\":\"organ\",\"start\":0.53,\"end\":0.84}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.54, 8.66, "ferdinand armine"], [11.69, 12.3, "mirabel"]], "response_mode": "offline", "answer": "[{\"word\":\"ferdinand armine\",\"start\":7.54,\"end\":8.66},{\"word\":\"mirabel\",\"start\":11.69,\"end\":12.3}]"}
{"utterance_id": "3170-137482-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0033.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 0.85, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.65,\"end\":0.85}]"}
{"utterance_id": "1462-170145-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0010.flac", "task_type": "lexical_detection", "query": "slipped", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"slipped\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.51, "slipped"]], "response_mode": "offline", "answer": "[{\"word\":\"slipped\",\"start\":1.19,\"end\":1.51}]"}
{"utterance_id": "2035-152373-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0018.flac", "task_type": "lexical_detection", "query": "period", "prompt": "Once the audio has finished, identify every instance of the word \"period\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.0, 5.54, "period"]], "response_mode": "offline", "answer": "[{\"word\":\"period\",\"start\":5.0,\"end\":5.54}]"}
{"utterance_id": "2086-149220-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0023.flac", "task_type": "lexical_detection", "query": "fancy", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"fancy\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.61, 5.15, "fancy"]], "response_mode": "offline", "answer": "[{\"word\":\"fancy\",\"start\":4.61,\"end\":5.15}]"}
{"utterance_id": "84-121123-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.89, 1.47, "assassin"]], "response_mode": "offline", "answer": "[{\"word\":\"assassin\",\"start\":0.89,\"end\":1.47}]"}
{"utterance_id": "6241-61943-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0004.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.19, "ghost"]], "response_mode": "offline", "answer": "[{\"word\":\"ghost\",\"start\":0.81,\"end\":1.19}]"}
{"utterance_id": "1993-147149-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0008.flac", "task_type": "lexical_detection", "query": "chance", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"chance\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.63, "chance"]], "response_mode": "offline", "answer": "[{\"word\":\"chance\",\"start\":1.21,\"end\":1.63}]"}
{"utterance_id": "422-122949-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[3.9, 4.1, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.9,\"end\":4.1}]"}
{"utterance_id": "5895-34622-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0013.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a person's name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.72, 2.23, "one"], [3.47, 3.95, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.72,\"end\":2.23},{\"word\":\"two\",\"start\":3.47,\"end\":3.95}]"}
{"utterance_id": "652-129742-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0018.flac", "task_type": "lexical_detection", "query": "ice", "prompt": "Once the audio has finished, identify every instance of the word \"ice\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.13, 2.32, "ice"]], "response_mode": "offline", "answer": "[{\"word\":\"ice\",\"start\":2.13,\"end\":2.32}]"}
{"utterance_id": "2428-83699-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0034.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 3.99, "several"]], "response_mode": "offline", "answer": "[{\"word\":\"several\",\"start\":3.59,\"end\":3.99}]"}
{"utterance_id": "1988-147956-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.34, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.81,\"end\":3.34}]"}
{"utterance_id": "2428-83705-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0022.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 0.8, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":0.37,\"end\":0.8}]"}
{"utterance_id": "8297-275154-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0007.flac", "task_type": "lexical_detection", "query": "visit", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"visit\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.88, 3.28, "visit"]], "response_mode": "offline", "answer": "[{\"word\":\"visit\",\"start\":2.88,\"end\":3.28}]"}
{"utterance_id": "6241-61943-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 0.79, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":0.53,\"end\":0.79}]"}
{"utterance_id": "6295-244435-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.26, 0.98, "colonel kenton"]], "response_mode": "offline", "answer": "[{\"word\":\"colonel kenton\",\"start\":0.26,\"end\":0.98}]"}
{"utterance_id": "6313-76958-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 1.97, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":1.61,\"end\":1.97}]"}
{"utterance_id": "2086-149220-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0015.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.28, 7.7, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":7.28,\"end\":7.7}]"}
{"utterance_id": "1919-142785-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0016.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.28, "lemon"]], "response_mode": "offline", "answer": "[{\"word\":\"lemon\",\"start\":1.93,\"end\":2.28}]"}
{"utterance_id": "251-118436-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0014.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.72, 2.14, "golden"]], "response_mode": "offline", "answer": "[{\"word\":\"golden\",\"start\":1.72,\"end\":2.14}]"}
{"utterance_id": "777-126732-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0034.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.06, 2.7, "michaelis"]], "response_mode": "offline", "answer": "[{\"word\":\"michaelis\",\"start\":2.06,\"end\":2.7}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "semantic_counting", "query": "date", "prompt": "Listen to the full audio. Count every occurrence of a date. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.73, "one"], [3.3, 4.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.26,\"end\":1.73},{\"word\":\"two\",\"start\":3.3,\"end\":4.06}]"}
{"utterance_id": "174-84280-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0013.flac", "task_type": "lexical_detection", "query": "freely", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"freely\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.72, 10.23, "freely"]], "response_mode": "offline", "answer": "[{\"word\":\"freely\",\"start\":9.72,\"end\":10.23}]"}
{"utterance_id": "8842-302196-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0012.flac", "task_type": "lexical_detection", "query": "vita", "prompt": "After the audio ends, extract all occurrences of the word \"vita\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.21, "vita"]], "response_mode": "offline", "answer": "[{\"word\":\"vita\",\"start\":1.92,\"end\":2.21}]"}
{"utterance_id": "5694-64038-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0010.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.41, 2.88, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":2.41,\"end\":2.88}]"}
{"utterance_id": "2277-149897-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.22, 6.67, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":6.22,\"end\":6.67}]"}
{"utterance_id": "777-126732-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0006.flac", "task_type": "lexical_detection", "query": "added", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"added\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.27, 3.45, "added"]], "response_mode": "offline", "answer": "[{\"word\":\"added\",\"start\":3.27,\"end\":3.45}]"}
{"utterance_id": "2277-149874-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0019.flac", "task_type": "lexical_detection", "query": "intervals", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"intervals\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.15, 5.76, "intervals"]], "response_mode": "offline", "answer": "[{\"word\":\"intervals\",\"start\":5.15,\"end\":5.76}]"}
{"utterance_id": "3081-166546-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0028.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.95, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.67,\"end\":0.95}]"}
{"utterance_id": "2086-149220-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.63, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.2,\"end\":2.63}]"}
{"utterance_id": "1673-143396-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0004.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[4.56, 5.16, "one"], [12.13, 12.64, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.56,\"end\":5.16},{\"word\":\"two\",\"start\":12.13,\"end\":12.64}]"}
{"utterance_id": "7976-110124-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0024.flac", "task_type": "lexical_detection", "query": "vain", "prompt": "After the audio ends, extract all occurrences of the word \"vain\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.29, "vain"]], "response_mode": "offline", "answer": "[{\"word\":\"vain\",\"start\":0.99,\"end\":1.29}]"}
{"utterance_id": "3000-15664-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0028.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 2.02, "woods"]], "response_mode": "offline", "answer": "[{\"word\":\"woods\",\"start\":1.48,\"end\":2.02}]"}
{"utterance_id": "6295-64301-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0027.flac", "task_type": "lexical_detection", "query": "confusion", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"confusion\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.85, 2.45, "confusion"]], "response_mode": "offline", "answer": "[{\"word\":\"confusion\",\"start\":1.85,\"end\":2.45}]"}
{"utterance_id": "1462-170142-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0020.flac", "task_type": "lexical_detection", "query": "bit", "prompt": "After listening to the whole recording, output every instance of the word \"bit\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.63, "bit"]], "response_mode": "offline", "answer": "[{\"word\":\"bit\",\"start\":0.43,\"end\":0.63}]"}
{"utterance_id": "3536-8226-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0010.flac", "task_type": "lexical_detection", "query": "followers", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"followers\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.5, 2.12, "followers"]], "response_mode": "offline", "answer": "[{\"word\":\"followers\",\"start\":1.5,\"end\":2.12}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "semantic_counting", "query": "city", "prompt": "Listen to the full audio. Count every occurrence of a city name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.71, "one"], [5.72, 6.13, "two"], [16.28, 16.82, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.26,\"end\":1.71},{\"word\":\"two\",\"start\":5.72,\"end\":6.13},{\"word\":\"three\",\"start\":16.28,\"end\":16.82}]"}
{"utterance_id": "777-126732-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0041.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.33, 0.77, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":0.33,\"end\":0.77}]"}
{"utterance_id": "2412-153954-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.8, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.62,\"end\":1.8}]"}
{"utterance_id": "3170-137482-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0041.flac", "task_type": "lexical_detection", "query": "scruples", "prompt": "When the utterance is complete, report all instances of the word \"scruples\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.66, "scruples"]], "response_mode": "offline", "answer": "[{\"word\":\"scruples\",\"start\":5.9,\"end\":6.66}]"}
{"utterance_id": "6319-64726-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[4.22, 4.55, "heads"]], "response_mode": "offline", "answer": "[{\"word\":\"heads\",\"start\":4.22,\"end\":4.55}]"}
{"utterance_id": "3081-166546-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0045.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.2, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.99,\"end\":5.2}]"}
{"utterance_id": "1919-142785-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0003.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.17, "long pepper"], [2.46, 2.86, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"long pepper\",\"start\":0.5,\"end\":1.17},{\"word\":\"black\",\"start\":2.46,\"end\":2.86}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.36, 10.77, "sister"]], "response_mode": "offline", "answer": "[{\"word\":\"sister\",\"start\":10.36,\"end\":10.77}]"}
{"utterance_id": "1988-147956-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.78, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.44,\"end\":2.78}]"}
{"utterance_id": "2277-149897-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0004.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Process the full audio offline. Find every instance of a disease name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.34, 4.51, "ill"]], "response_mode": "offline", "answer": "[{\"word\":\"ill\",\"start\":4.34,\"end\":4.51}]"}
{"utterance_id": "8297-275155-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.99, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":1.44,\"end\":1.99}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.97, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":0.43,\"end\":0.97}]"}
{"utterance_id": "6313-66129-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0013.flac", "task_type": "lexical_detection", "query": "dog", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"dog\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.42, "dog"]], "response_mode": "offline", "answer": "[{\"word\":\"dog\",\"start\":2.05,\"end\":2.42}]"}
{"utterance_id": "5895-34622-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0016.flac", "task_type": "lexical_detection", "query": "ursus", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"ursus\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.21, 0.61, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":0.21,\"end\":0.61}]"}
{"utterance_id": "251-137823-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0005.flac", "task_type": "lexical_detection", "query": "faber", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"faber\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.27, "faber"]], "response_mode": "offline", "answer": "[{\"word\":\"faber\",\"start\":3.81,\"end\":4.27}]"}
{"utterance_id": "2035-152373-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0004.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 2.03, "sparta"]], "response_mode": "offline", "answer": "[{\"word\":\"sparta\",\"start\":1.48,\"end\":2.03}]"}
{"utterance_id": "6241-61943-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0002.flac", "task_type": "lexical_detection", "query": "wind", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wind\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.57, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":2.14,\"end\":2.57}]"}
{"utterance_id": "5694-64038-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.06, 1.47, "pulse"], [3.7, 4.1, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"pulse\",\"start\":1.06,\"end\":1.47},{\"word\":\"heart\",\"start\":3.7,\"end\":4.1}]"}
{"utterance_id": "3853-163249-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0050.flac", "task_type": "lexical_counting", "query": "ready", "prompt": "After the audio ends, tally every instance of the word \"ready\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.17, 2.46, "one"], [3.6, 3.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.17,\"end\":2.46},{\"word\":\"two\",\"start\":3.6,\"end\":3.97}]"}
{"utterance_id": "5536-43363-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0008.flac", "task_type": "lexical_detection", "query": "relatives", "prompt": "Process the full audio offline. Find every instance of the word \"relatives\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.29, "relatives"]], "response_mode": "offline", "answer": "[{\"word\":\"relatives\",\"start\":2.84,\"end\":3.29}]"}
{"utterance_id": "3576-138058-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0022.flac", "task_type": "lexical_counting", "query": "thee", "prompt": "After listening to the whole recording, output timed count words for the word \"thee\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[18.71, 18.89, "one"], [24.13, 24.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":18.71,\"end\":18.89},{\"word\":\"two\",\"start\":24.13,\"end\":24.45}]"}
{"utterance_id": "174-50561-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0000.flac", "task_type": "lexical_detection", "query": "captive", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"captive\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.52, "captive"]], "response_mode": "offline", "answer": "[{\"word\":\"captive\",\"start\":2.95,\"end\":3.52}]"}
{"utterance_id": "2277-149874-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.47, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":2.07,\"end\":2.47}]"}
{"utterance_id": "1462-170138-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.33, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":3.86,\"end\":4.33}]"}
{"utterance_id": "2035-147961-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0019.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.49, 0.77, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.49,\"end\":0.77}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.51, 1.11, "salad"], [2.87, 3.34, "apples"], [5.4, 5.89, "celery"], [8.39, 8.99, "walnuts"]], "response_mode": "offline", "answer": "[{\"word\":\"salad\",\"start\":0.51,\"end\":1.11},{\"word\":\"apples\",\"start\":2.87,\"end\":3.34},{\"word\":\"celery\",\"start\":5.4,\"end\":5.89},{\"word\":\"walnuts\",\"start\":8.39,\"end\":8.99}]"}
{"utterance_id": "5694-64029-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0022.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "You will hear a complete utterance. When it ends, list every instance of a weather term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.45, "hail storm"]], "response_mode": "offline", "answer": "[{\"word\":\"hail storm\",\"start\":0.67,\"end\":1.45}]"}
{"utterance_id": "3752-4944-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0047.flac", "task_type": "lexical_detection", "query": "boy's", "prompt": "When the utterance is complete, report all instances of the word \"boy's\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.11, 5.49, "boy's"]], "response_mode": "offline", "answer": "[{\"word\":\"boy's\",\"start\":5.11,\"end\":5.49}]"}
{"utterance_id": "6319-275224-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0007.flac", "task_type": "lexical_detection", "query": "grieved", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"grieved\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.21, 1.67, "grieved"]], "response_mode": "offline", "answer": "[{\"word\":\"grieved\",\"start\":1.21,\"end\":1.67}]"}
{"utterance_id": "3536-8226-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0030.flac", "task_type": "lexical_detection", "query": "opinion", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"opinion\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.22, 1.76, "opinion"]], "response_mode": "offline", "answer": "[{\"word\":\"opinion\",\"start\":1.22,\"end\":1.76}]"}
{"utterance_id": "2428-83705-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0022.flac", "task_type": "lexical_detection", "query": "bird", "prompt": "Once the audio has finished, identify every instance of the word \"bird\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.37, 0.8, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":0.37,\"end\":0.8}]"}
{"utterance_id": "3000-15664-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0040.flac", "task_type": "lexical_detection", "query": "wilderness", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wilderness\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.34, 3.96, "wilderness"]], "response_mode": "offline", "answer": "[{\"word\":\"wilderness\",\"start\":3.34,\"end\":3.96}]"}
{"utterance_id": "6313-66125-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.88,\"end\":2.12}]"}
{"utterance_id": "6345-93302-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0025.flac", "task_type": "lexical_detection", "query": "really", "prompt": "Once the audio has finished, identify every instance of the word \"really\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.44, 0.65, "really"]], "response_mode": "offline", "answer": "[{\"word\":\"really\",\"start\":0.44,\"end\":0.65}]"}
{"utterance_id": "3081-166546-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0055.flac", "task_type": "lexical_detection", "query": "germans", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"germans\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.33, "germans"]], "response_mode": "offline", "answer": "[{\"word\":\"germans\",\"start\":0.78,\"end\":1.33}]"}
{"utterance_id": "84-121123-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0019.flac", "task_type": "lexical_detection", "query": "sign", "prompt": "When the utterance is complete, report all instances of the word \"sign\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.39, "sign"]], "response_mode": "offline", "answer": "[{\"word\":\"sign\",\"start\":1.1,\"end\":1.39}]"}
{"utterance_id": "6345-93306-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0021.flac", "task_type": "lexical_detection", "query": "key", "prompt": "Process the full audio offline. Find every instance of the word \"key\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.32, 3.62, "key"]], "response_mode": "offline", "answer": "[{\"word\":\"key\",\"start\":3.32,\"end\":3.62}]"}
{"utterance_id": "2277-149897-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0002.flac", "task_type": "lexical_detection", "query": "agreed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"agreed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.78, 5.1, "agreed"]], "response_mode": "offline", "answer": "[{\"word\":\"agreed\",\"start\":4.78,\"end\":5.1}]"}
{"utterance_id": "5694-64038-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.68, 8.53, "joe johnston"]], "response_mode": "offline", "answer": "[{\"word\":\"joe johnston\",\"start\":7.68,\"end\":8.53}]"}
{"utterance_id": "251-136532-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 1.64, "gloria standish"]], "response_mode": "offline", "answer": "[{\"word\":\"gloria standish\",\"start\":0.41,\"end\":1.64}]"}
{"utterance_id": "1673-143396-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0006.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.01, 3.4, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.01,\"end\":3.4}]"}
{"utterance_id": "6241-61943-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0009.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Process the full audio offline. Find every instance of an organization name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.08, "valkyrie"]], "response_mode": "offline", "answer": "[{\"word\":\"valkyrie\",\"start\":0.59,\"end\":1.08}]"}
{"utterance_id": "84-121550-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0034.flac", "task_type": "lexical_detection", "query": "desires", "prompt": "After listening to the whole recording, output every instance of the word \"desires\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.75, 3.33, "desires"]], "response_mode": "offline", "answer": "[{\"word\":\"desires\",\"start\":2.75,\"end\":3.33}]"}
{"utterance_id": "2428-83699-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0039.flac", "task_type": "lexical_detection", "query": "bell", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"bell\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.29, "bell"]], "response_mode": "offline", "answer": "[{\"word\":\"bell\",\"start\":1.01,\"end\":1.29}]"}
{"utterance_id": "7850-286674-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0016.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.26, 8.64, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":8.26,\"end\":8.64}]"}
{"utterance_id": "3000-15664-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0014.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 2.29, "strawberry valley"]], "response_mode": "offline", "answer": "[{\"word\":\"strawberry valley\",\"start\":1.27,\"end\":2.29}]"}
{"utterance_id": "652-130726-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0019.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.68, "viands"]], "response_mode": "offline", "answer": "[{\"word\":\"viands\",\"start\":2.26,\"end\":2.68}]"}
{"utterance_id": "777-126732-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.83, 2.17, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.83,\"end\":2.17}]"}
{"utterance_id": "2277-149897-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0002.flac", "task_type": "semantic_counting", "query": "clock_time", "prompt": "Listen to the full audio. Count every occurrence of a time of day. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.31, "one"], [5.81, 6.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.86,\"end\":4.31},{\"word\":\"two\",\"start\":5.81,\"end\":6.27}]"}
{"utterance_id": "6313-76958-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.38, 1.15, "lumpy bates"]], "response_mode": "offline", "answer": "[{\"word\":\"lumpy bates\",\"start\":0.38,\"end\":1.15}]"}
{"utterance_id": "2035-147961-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0002.flac", "task_type": "lexical_detection", "query": "bag", "prompt": "When the utterance is complete, report all instances of the word \"bag\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.46, 4.73, "bag"]], "response_mode": "offline", "answer": "[{\"word\":\"bag\",\"start\":4.46,\"end\":4.73}]"}
{"utterance_id": "5694-64025-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.93, 4.56, "gladden's"]], "response_mode": "offline", "answer": "[{\"word\":\"gladden's\",\"start\":3.93,\"end\":4.56}]"}
{"utterance_id": "1988-147956-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.94, "fuchs"]], "response_mode": "offline", "answer": "[{\"word\":\"fuchs\",\"start\":0.53,\"end\":0.94}]"}
{"utterance_id": "6313-66129-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0014.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 0.72, "supper"]], "response_mode": "offline", "answer": "[{\"word\":\"supper\",\"start\":0.35,\"end\":0.72}]"}
{"utterance_id": "2078-142845-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.76, "next"]], "response_mode": "offline", "answer": "[{\"word\":\"next\",\"start\":0.42,\"end\":0.76}]"}
{"utterance_id": "7850-281318-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0005.flac", "task_type": "lexical_detection", "query": "indeed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"indeed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 1.09, "indeed"]], "response_mode": "offline", "answer": "[{\"word\":\"indeed\",\"start\":0.68,\"end\":1.09}]"}
{"utterance_id": "7850-281318-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0022.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.48, "wood pigeon"]], "response_mode": "offline", "answer": "[{\"word\":\"wood pigeon\",\"start\":0.67,\"end\":1.48}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.97, "missus wilson"], [17.5, 17.99, "alice"]], "response_mode": "offline", "answer": "[{\"word\":\"missus wilson\",\"start\":3.31,\"end\":3.97},{\"word\":\"alice\",\"start\":17.5,\"end\":17.99}]"}
{"utterance_id": "251-136532-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0009.flac", "task_type": "lexical_detection", "query": "they're", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"they're\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.6, "they're"]], "response_mode": "offline", "answer": "[{\"word\":\"they're\",\"start\":1.35,\"end\":1.6}]"}
{"utterance_id": "1462-170142-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0021.flac", "task_type": "lexical_detection", "query": "quietly", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"quietly\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.48, "quietly"]], "response_mode": "offline", "answer": "[{\"word\":\"quietly\",\"start\":2.98,\"end\":3.48}]"}
{"utterance_id": "174-50561-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0006.flac", "task_type": "lexical_detection", "query": "harm", "prompt": "When the utterance is complete, report all instances of the word \"harm\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.77, "harm"]], "response_mode": "offline", "answer": "[{\"word\":\"harm\",\"start\":3.4,\"end\":3.77}]"}
{"utterance_id": "3853-163249-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0047.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.83, 10.19, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":9.83,\"end\":10.19}]"}
{"utterance_id": "3752-4944-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.21, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":3.58,\"end\":4.21}]"}
{"utterance_id": "1462-170138-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0006.flac", "task_type": "semantic_detection", "query": "country", "prompt": "When the utterance is complete, report all instances of a country name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.56, 8.12, "ireland"]], "response_mode": "offline", "answer": "[{\"word\":\"ireland\",\"start\":7.56,\"end\":8.12}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "lexical_detection", "query": "regards", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"regards\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.01, "regards"], [9.64, 9.98, "regards"]], "response_mode": "offline", "answer": "[{\"word\":\"regards\",\"start\":3.7,\"end\":4.01},{\"word\":\"regards\",\"start\":9.64,\"end\":9.98}]"}
{"utterance_id": "7850-73752-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0016.flac", "task_type": "lexical_detection", "query": "enchanted", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"enchanted\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.85, 1.56, "enchanted"]], "response_mode": "offline", "answer": "[{\"word\":\"enchanted\",\"start\":0.85,\"end\":1.56}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "lexical_detection", "query": "fifteen", "prompt": "When the utterance is complete, report all instances of the word \"fifteen\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 1.07, "fifteen"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen\",\"start\":0.47,\"end\":1.07}]"}
{"utterance_id": "251-118436-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0002.flac", "task_type": "lexical_detection", "query": "guarded", "prompt": "Once the audio has finished, identify every instance of the word \"guarded\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.98, 2.39, "guarded"]], "response_mode": "offline", "answer": "[{\"word\":\"guarded\",\"start\":1.98,\"end\":2.39}]"}
{"utterance_id": "6295-244435-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0018.flac", "task_type": "lexical_detection", "query": "intended", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"intended\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 0.71, "intended"]], "response_mode": "offline", "answer": "[{\"word\":\"intended\",\"start\":0.31,\"end\":0.71}]"}
{"utterance_id": "5694-64029-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0026.flac", "task_type": "lexical_detection", "query": "hospital", "prompt": "After listening to the whole recording, output every instance of the word \"hospital\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.51, "hospital"]], "response_mode": "offline", "answer": "[{\"word\":\"hospital\",\"start\":1.69,\"end\":2.51}]"}
{"utterance_id": "5895-34622-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0009.flac", "task_type": "lexical_counting", "query": "worked", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"worked\" with word/start/end. If none, reply [].", "labels": [[1.34, 1.64, "one"], [4.54, 4.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.34,\"end\":1.64},{\"word\":\"two\",\"start\":4.54,\"end\":4.85}]"}
{"utterance_id": "3536-8226-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0000.flac", "task_type": "lexical_detection", "query": "communication", "prompt": "Once the audio has finished, identify every instance of the word \"communication\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.83, 1.58, "communication"]], "response_mode": "offline", "answer": "[{\"word\":\"communication\",\"start\":0.83,\"end\":1.58}]"}
{"utterance_id": "174-50561-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0001.flac", "task_type": "lexical_detection", "query": "presumed", "prompt": "Once the audio has finished, identify every instance of the word \"presumed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.33, 2.96, "presumed"]], "response_mode": "offline", "answer": "[{\"word\":\"presumed\",\"start\":2.33,\"end\":2.96}]"}
{"utterance_id": "6313-76958-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0011.flac", "task_type": "lexical_detection", "query": "using", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"using\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.07, "using"]], "response_mode": "offline", "answer": "[{\"word\":\"using\",\"start\":1.73,\"end\":2.07}]"}
{"utterance_id": "1993-147149-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0021.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[2.68, 3.03, "dawn"]], "response_mode": "offline", "answer": "[{\"word\":\"dawn\",\"start\":2.68,\"end\":3.03}]"}
{"utterance_id": "2035-147960-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.94, 1.4, "tony"]], "response_mode": "offline", "answer": "[{\"word\":\"tony\",\"start\":0.94,\"end\":1.4}]"}
{"utterance_id": "6241-66616-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0023.flac", "task_type": "lexical_detection", "query": "later", "prompt": "When the utterance is complete, report all instances of the word \"later\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.23, "later"]], "response_mode": "offline", "answer": "[{\"word\":\"later\",\"start\":0.99,\"end\":1.23}]"}
{"utterance_id": "3752-4943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[3.98, 4.28, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":3.98,\"end\":4.28}]"}
{"utterance_id": "6319-57405-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0005.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "After the audio ends, extract all occurrences of a weather term. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.62, 6.08, "flood"]], "response_mode": "offline", "answer": "[{\"word\":\"flood\",\"start\":5.62,\"end\":6.08}]"}
{"utterance_id": "1272-141231-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0026.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.81, 2.27, "brion"]], "response_mode": "offline", "answer": "[{\"word\":\"brion\",\"start\":1.81,\"end\":2.27}]"}
{"utterance_id": "2277-149897-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0016.flac", "task_type": "lexical_detection", "query": "meanwhile", "prompt": "After the audio ends, extract all occurrences of the word \"meanwhile\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.41, 4.89, "meanwhile"]], "response_mode": "offline", "answer": "[{\"word\":\"meanwhile\",\"start\":4.41,\"end\":4.89}]"}
{"utterance_id": "7976-110523-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0013.flac", "task_type": "lexical_detection", "query": "shut", "prompt": "After the audio ends, extract all occurrences of the word \"shut\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 3.27, "shut"]], "response_mode": "offline", "answer": "[{\"word\":\"shut\",\"start\":3.03,\"end\":3.27}]"}
{"utterance_id": "3000-15664-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0030.flac", "task_type": "lexical_detection", "query": "darkness", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"darkness\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.07, 1.77, "darkness"]], "response_mode": "offline", "answer": "[{\"word\":\"darkness\",\"start\":1.07,\"end\":1.77}]"}
{"utterance_id": "5694-64038-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0015.flac", "task_type": "lexical_detection", "query": "ourselves", "prompt": "Process the full audio offline. Find every instance of the word \"ourselves\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.3, 3.98, "ourselves"]], "response_mode": "offline", "answer": "[{\"word\":\"ourselves\",\"start\":3.3,\"end\":3.98}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.8, "new york times"]], "response_mode": "offline", "answer": "[{\"word\":\"new york times\",\"start\":2.65,\"end\":3.8}]"}
{"utterance_id": "5895-34629-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0014.flac", "task_type": "lexical_detection", "query": "ate", "prompt": "After listening to the whole recording, output every instance of the word \"ate\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.2, "ate"]], "response_mode": "offline", "answer": "[{\"word\":\"ate\",\"start\":0.84,\"end\":1.2}]"}
{"utterance_id": "6241-66616-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.81, "one"], [2.16, 2.47, "score"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.81},{\"word\":\"score\",\"start\":2.16,\"end\":2.47}]"}
{"utterance_id": "652-130726-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.27, "chefs"]], "response_mode": "offline", "answer": "[{\"word\":\"chefs\",\"start\":1.96,\"end\":2.27}]"}
{"utterance_id": "777-126732-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0077.flac", "task_type": "lexical_detection", "query": "face", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"face\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.91, 1.22, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":0.91,\"end\":1.22}]"}
{"utterance_id": "2412-153947-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0000.flac", "task_type": "lexical_detection", "query": "preface", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"preface\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.97, "preface"]], "response_mode": "offline", "answer": "[{\"word\":\"preface\",\"start\":0.53,\"end\":0.97}]"}
{"utterance_id": "2035-147961-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.6, "pavel"]], "response_mode": "offline", "answer": "[{\"word\":\"pavel\",\"start\":1.22,\"end\":1.6}]"}
{"utterance_id": "1462-170145-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0003.flac", "task_type": "lexical_detection", "query": "consideration", "prompt": "Once the audio has finished, identify every instance of the word \"consideration\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.55, 6.49, "consideration"]], "response_mode": "offline", "answer": "[{\"word\":\"consideration\",\"start\":5.55,\"end\":6.49}]"}
{"utterance_id": "2803-154320-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.73, 4.12, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":3.73,\"end\":4.12}]"}
{"utterance_id": "2902-9008-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.57, 3.08, "porphyry"]], "response_mode": "offline", "answer": "[{\"word\":\"porphyry\",\"start\":2.57,\"end\":3.08}]"}
{"utterance_id": "6313-66129-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0021.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.22, 12.62, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":12.22,\"end\":12.62}]"}
{"utterance_id": "777-126732-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0021.flac", "task_type": "lexical_detection", "query": "prisoner", "prompt": "Process the full audio offline. Find every instance of the word \"prisoner\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.16, "prisoner"]], "response_mode": "offline", "answer": "[{\"word\":\"prisoner\",\"start\":1.61,\"end\":2.16}]"}
{"utterance_id": "3170-137482-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[18.35, 18.95, "twenty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty\",\"start\":18.35,\"end\":18.95}]"}
{"utterance_id": "6313-76958-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.82, "leg"]], "response_mode": "offline", "answer": "[{\"word\":\"leg\",\"start\":1.54,\"end\":1.82}]"}
{"utterance_id": "3752-4943-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0019.flac", "task_type": "lexical_detection", "query": "does", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"does\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.27, 0.4, "does"]], "response_mode": "offline", "answer": "[{\"word\":\"does\",\"start\":0.27,\"end\":0.4}]"}
{"utterance_id": "6313-76958-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0024.flac", "task_type": "lexical_detection", "query": "foreman", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"foreman\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.83, "foreman"]], "response_mode": "offline", "answer": "[{\"word\":\"foreman\",\"start\":2.38,\"end\":2.83}]"}
{"utterance_id": "5694-64038-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0000.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "You will hear a complete utterance. When it ends, list every instance of a state or province name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 2.16, "tennessee"]], "response_mode": "offline", "answer": "[{\"word\":\"tennessee\",\"start\":1.4,\"end\":2.16}]"}
{"utterance_id": "5338-24615-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.07, 4.27, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":4.07,\"end\":4.27}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.2, 9.55, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":9.2,\"end\":9.55}]"}
{"utterance_id": "6313-66129-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0020.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.73, 1.05, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":0.73,\"end\":1.05}]"}
{"utterance_id": "84-121123-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0021.flac", "task_type": "lexical_detection", "query": "landing", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"landing\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.59, "landing"]], "response_mode": "offline", "answer": "[{\"word\":\"landing\",\"start\":1.11,\"end\":1.59}]"}
{"utterance_id": "8297-275155-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.2, 5.55, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":5.2,\"end\":5.55}]"}
{"utterance_id": "2428-83705-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.39, 4.01, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":3.39,\"end\":4.01}]"}
{"utterance_id": "2078-142845-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0029.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.57, 4.44, "seventeen twenty six"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen twenty six\",\"start\":2.57,\"end\":4.44}]"}
{"utterance_id": "3081-166546-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0079.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.62, 3.97, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":3.62,\"end\":3.97}]"}
{"utterance_id": "3536-23268-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0002.flac", "task_type": "lexical_detection", "query": "sigh", "prompt": "Once the audio has finished, identify every instance of the word \"sigh\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.11, 3.54, "sigh"]], "response_mode": "offline", "answer": "[{\"word\":\"sigh\",\"start\":3.11,\"end\":3.54}]"}
{"utterance_id": "6313-66125-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.14, 7.52, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":7.14,\"end\":7.52}]"}
{"utterance_id": "6295-244435-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0021.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Process the full audio offline. Find every instance of a date and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.89, "april"]], "response_mode": "offline", "answer": "[{\"word\":\"april\",\"start\":2.38,\"end\":2.89}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.17, "wolves"]], "response_mode": "offline", "answer": "[{\"word\":\"wolves\",\"start\":0.73,\"end\":1.17}]"}
{"utterance_id": "5694-64025-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0017.flac", "task_type": "semantic_detection", "query": "date", "prompt": "When the utterance is complete, report all instances of a date including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 0.94, "monday"]], "response_mode": "offline", "answer": "[{\"word\":\"monday\",\"start\":0.66,\"end\":0.94}]"}
{"utterance_id": "3752-4943-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0023.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Process the full audio offline. Find every instance of a city name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 1.16, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":0.57,\"end\":1.16}]"}
{"utterance_id": "2902-9008-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0015.flac", "task_type": "lexical_detection", "query": "excellency", "prompt": "Once the audio has finished, identify every instance of the word \"excellency\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.42, 1.14, "excellency"]], "response_mode": "offline", "answer": "[{\"word\":\"excellency\",\"start\":0.42,\"end\":1.14}]"}
{"utterance_id": "2035-147960-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0000.flac", "task_type": "lexical_detection", "query": "older", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"older\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.13, 1.39, "older"]], "response_mode": "offline", "answer": "[{\"word\":\"older\",\"start\":1.13,\"end\":1.39}]"}
{"utterance_id": "7850-111771-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.13, 1.37, "lee"]], "response_mode": "offline", "answer": "[{\"word\":\"lee\",\"start\":1.13,\"end\":1.37}]"}
{"utterance_id": "2086-149220-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0025.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.78, 7.21, "painter"]], "response_mode": "offline", "answer": "[{\"word\":\"painter\",\"start\":6.78,\"end\":7.21}]"}
{"utterance_id": "1988-24833-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.24, 0.45, "tom"], [1.85, 2.26, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.24,\"end\":0.45},{\"word\":\"hilda\",\"start\":1.85,\"end\":2.26}]"}
{"utterance_id": "1673-143396-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0006.flac", "task_type": "lexical_detection", "query": "logos", "prompt": "When the utterance is complete, report all instances of the word \"logos\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.64, 5.3, "logos"]], "response_mode": "offline", "answer": "[{\"word\":\"logos\",\"start\":4.64,\"end\":5.3}]"}
{"utterance_id": "84-121550-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0032.flac", "task_type": "lexical_detection", "query": "low", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"low\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.25, "low"]], "response_mode": "offline", "answer": "[{\"word\":\"low\",\"start\":0.89,\"end\":1.25}]"}
{"utterance_id": "2428-83699-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0010.flac", "task_type": "lexical_detection", "query": "expect", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"expect\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.65, "expect"]], "response_mode": "offline", "answer": "[{\"word\":\"expect\",\"start\":1.14,\"end\":1.65}]"}
{"utterance_id": "3536-23268-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 2.36, "mister dorriforth"]], "response_mode": "offline", "answer": "[{\"word\":\"mister dorriforth\",\"start\":1.57,\"end\":2.36}]"}
{"utterance_id": "2035-147961-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0020.flac", "task_type": "lexical_detection", "query": "hill", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"hill\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.85, 2.19, "hill"]], "response_mode": "offline", "answer": "[{\"word\":\"hill\",\"start\":1.85,\"end\":2.19}]"}
{"utterance_id": "1919-142785-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0033.flac", "task_type": "lexical_detection", "query": "omitted", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"omitted\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.79, 3.23, "omitted"]], "response_mode": "offline", "answer": "[{\"word\":\"omitted\",\"start\":2.79,\"end\":3.23}]"}
{"utterance_id": "2412-153948-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0015.flac", "task_type": "lexical_detection", "query": "range", "prompt": "When the utterance is complete, report all instances of the word \"range\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.85, "range"]], "response_mode": "offline", "answer": "[{\"word\":\"range\",\"start\":1.46,\"end\":1.85}]"}
{"utterance_id": "7976-110124-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0009.flac", "task_type": "lexical_detection", "query": "youngest", "prompt": "After the audio ends, extract all occurrences of the word \"youngest\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.47, 3.86, "youngest"]], "response_mode": "offline", "answer": "[{\"word\":\"youngest\",\"start\":3.47,\"end\":3.86}]"}
{"utterance_id": "1462-170138-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0009.flac", "task_type": "lexical_detection", "query": "delightful", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"delightful\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 0.94, "delightful"]], "response_mode": "offline", "answer": "[{\"word\":\"delightful\",\"start\":0.48,\"end\":0.94}]"}
{"utterance_id": "1993-147965-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0004.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen carefully to the entire clip, then report every instance of a time of day as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.23, "afternoon"]], "response_mode": "offline", "answer": "[{\"word\":\"afternoon\",\"start\":0.72,\"end\":1.23}]"}
{"utterance_id": "3752-4944-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0005.flac", "task_type": "lexical_detection", "query": "yourself", "prompt": "Process the full audio offline. Find every instance of the word \"yourself\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.41, "yourself"]], "response_mode": "offline", "answer": "[{\"word\":\"yourself\",\"start\":1.05,\"end\":1.41}]"}
{"utterance_id": "251-136532-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0015.flac", "task_type": "lexical_detection", "query": "drinking", "prompt": "After listening to the whole recording, output every instance of the word \"drinking\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.71, 5.1, "drinking"]], "response_mode": "offline", "answer": "[{\"word\":\"drinking\",\"start\":4.71,\"end\":5.1}]"}
{"utterance_id": "3752-4944-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0061.flac", "task_type": "lexical_detection", "query": "look", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"look\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.22, "look"]], "response_mode": "offline", "answer": "[{\"word\":\"look\",\"start\":0.95,\"end\":1.22}]"}
{"utterance_id": "84-121550-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.31, "four"], [4.39, 4.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.86,\"end\":2.31},{\"word\":\"two\",\"start\":4.39,\"end\":4.67}]"}
{"utterance_id": "84-121123-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a disease name with fields word, start, end (seconds). If none, reply [].", "labels": [[9.78, 10.39, "epilepsy"]], "response_mode": "offline", "answer": "[{\"word\":\"epilepsy\",\"start\":9.78,\"end\":10.39}]"}
{"utterance_id": "5895-34615-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0008.flac", "task_type": "lexical_detection", "query": "depend", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"depend\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.88, "depend"]], "response_mode": "offline", "answer": "[{\"word\":\"depend\",\"start\":1.4,\"end\":1.88}]"}
{"utterance_id": "3081-166546-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[2.5, 2.85, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":2.5,\"end\":2.85}]"}
{"utterance_id": "652-129742-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0007.flac", "task_type": "lexical_detection", "query": "garnish", "prompt": "Once the audio has finished, identify every instance of the word \"garnish\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.99, 1.46, "garnish"]], "response_mode": "offline", "answer": "[{\"word\":\"garnish\",\"start\":0.99,\"end\":1.46}]"}
{"utterance_id": "2086-149220-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0020.flac", "task_type": "lexical_detection", "query": "turn", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"turn\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.61, 0.85, "turn"]], "response_mode": "offline", "answer": "[{\"word\":\"turn\",\"start\":0.61,\"end\":0.85}]"}
{"utterance_id": "2803-161169-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0008.flac", "task_type": "lexical_counting", "query": "nut", "prompt": "Listen to the entire clip, then report the count sequence for the word \"nut\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.14, "one"], [9.6, 10.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.75,\"end\":4.14},{\"word\":\"two\",\"start\":9.6,\"end\":10.06}]"}
{"utterance_id": "5694-64029-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0021.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "When the utterance is complete, report all instances of a state or province name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.47, "tennessee"]], "response_mode": "offline", "answer": "[{\"word\":\"tennessee\",\"start\":2.06,\"end\":2.47}]"}
{"utterance_id": "5694-64029-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0024.flac", "task_type": "lexical_detection", "query": "glory", "prompt": "When the utterance is complete, report all instances of the word \"glory\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.87, "glory"]], "response_mode": "offline", "answer": "[{\"word\":\"glory\",\"start\":2.3,\"end\":2.87}]"}
{"utterance_id": "6295-244435-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.22, 0.76, "beauregard"]], "response_mode": "offline", "answer": "[{\"word\":\"beauregard\",\"start\":0.22,\"end\":0.76}]"}
{"utterance_id": "1272-135031-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0003.flac", "task_type": "lexical_detection", "query": "asleep", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"asleep\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.83, "asleep"]], "response_mode": "offline", "answer": "[{\"word\":\"asleep\",\"start\":1.26,\"end\":1.83}]"}
{"utterance_id": "2277-149896-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0026.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "You will hear a complete utterance. When it ends, list every instance of a weather term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 0.93, "drizzle"]], "response_mode": "offline", "answer": "[{\"word\":\"drizzle\",\"start\":0.58,\"end\":0.93}]"}
{"utterance_id": "2902-9006-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0006.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.46, "emperors"]], "response_mode": "offline", "answer": "[{\"word\":\"emperors\",\"start\":0.94,\"end\":1.46}]"}
{"utterance_id": "6313-66125-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.02, 3.47, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":3.02,\"end\":3.47}]"}
{"utterance_id": "5338-24640-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0007.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Process the full audio offline. Find every instance of an organization name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[16.62, 17.37, "inquisition"]], "response_mode": "offline", "answer": "[{\"word\":\"inquisition\",\"start\":16.62,\"end\":17.37}]"}
{"utterance_id": "2412-153947-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0010.flac", "task_type": "semantic_detection", "query": "date", "prompt": "When the utterance is complete, report all instances of a date including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 6.24, "may first eighteen seventy one"], [15.69, 16.72, "eighteen seventy two"]], "response_mode": "offline", "answer": "[{\"word\":\"may first eighteen seventy one\",\"start\":4.44,\"end\":6.24},{\"word\":\"eighteen seventy two\",\"start\":15.69,\"end\":16.72}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "lexical_detection", "query": "removed", "prompt": "Once the audio has finished, identify every instance of the word \"removed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.64, 2.24, "removed"]], "response_mode": "offline", "answer": "[{\"word\":\"removed\",\"start\":1.64,\"end\":2.24}]"}
{"utterance_id": "5536-43363-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0018.flac", "task_type": "lexical_detection", "query": "former", "prompt": "After listening to the whole recording, output every instance of the word \"former\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.4, 6.81, "former"]], "response_mode": "offline", "answer": "[{\"word\":\"former\",\"start\":6.4,\"end\":6.81}]"}
{"utterance_id": "3853-163249-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0054.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.82, 2.26, "breath"], [11.36, 11.81, "hearts"], [16.53, 16.84, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"breath\",\"start\":1.82,\"end\":2.26},{\"word\":\"hearts\",\"start\":11.36,\"end\":11.81},{\"word\":\"blood\",\"start\":16.53,\"end\":16.84}]"}
{"utterance_id": "2078-142845-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0026.flac", "task_type": "lexical_detection", "query": "toast", "prompt": "After the audio ends, extract all occurrences of the word \"toast\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.2, "toast"]], "response_mode": "offline", "answer": "[{\"word\":\"toast\",\"start\":1.44,\"end\":2.2}]"}
{"utterance_id": "2803-154320-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0012.flac", "task_type": "lexical_detection", "query": "brute", "prompt": "Once the audio has finished, identify every instance of the word \"brute\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.82, 2.55, "brute"]], "response_mode": "offline", "answer": "[{\"word\":\"brute\",\"start\":1.82,\"end\":2.55}]"}
{"utterance_id": "8842-304647-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0000.flac", "task_type": "lexical_detection", "query": "limb", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"limb\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.14, 3.49, "limb"]], "response_mode": "offline", "answer": "[{\"word\":\"limb\",\"start\":3.14,\"end\":3.49}]"}
{"utterance_id": "3081-166546-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 1.83, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":1.48,\"end\":1.83}]"}
{"utterance_id": "3752-4943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.16, 0.63, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":0.16,\"end\":0.63}]"}
{"utterance_id": "3081-166546-0067", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0067.flac", "task_type": "lexical_detection", "query": "glimpse", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"glimpse\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.5, 6.98, "glimpse"]], "response_mode": "offline", "answer": "[{\"word\":\"glimpse\",\"start\":6.5,\"end\":6.98}]"}
{"utterance_id": "1272-141231-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0030.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.68, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":3.26,\"end\":3.68}]"}
{"utterance_id": "3081-166546-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0053.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 2.18, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.59,\"end\":2.18}]"}
{"utterance_id": "1272-135031-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0000.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.89, 10.38, "dove"]], "response_mode": "offline", "answer": "[{\"word\":\"dove\",\"start\":9.89,\"end\":10.38}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[3.31, 3.97, "one"], [17.5, 17.99, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.31,\"end\":3.97},{\"word\":\"two\",\"start\":17.5,\"end\":17.99}]"}
{"utterance_id": "3853-163249-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0026.flac", "task_type": "lexical_detection", "query": "gifts", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"gifts\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.05, 3.39, "gifts"]], "response_mode": "offline", "answer": "[{\"word\":\"gifts\",\"start\":3.05,\"end\":3.39}]"}
{"utterance_id": "1462-170142-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0012.flac", "task_type": "lexical_detection", "query": "shadows", "prompt": "After listening to the whole recording, output every instance of the word \"shadows\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.43, 4.92, "shadows"]], "response_mode": "offline", "answer": "[{\"word\":\"shadows\",\"start\":4.43,\"end\":4.92}]"}
{"utterance_id": "6345-93306-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0023.flac", "task_type": "lexical_detection", "query": "certainly", "prompt": "Once the audio has finished, identify every instance of the word \"certainly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.46, 1.91, "certainly"]], "response_mode": "offline", "answer": "[{\"word\":\"certainly\",\"start\":1.46,\"end\":1.91}]"}
{"utterance_id": "3170-137482-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0007.flac", "task_type": "lexical_detection", "query": "fun", "prompt": "Process the full audio offline. Find every instance of the word \"fun\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 4.72, "fun"]], "response_mode": "offline", "answer": "[{\"word\":\"fun\",\"start\":4.44,\"end\":4.72}]"}
{"utterance_id": "7976-110523-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0015.flac", "task_type": "lexical_detection", "query": "grethel", "prompt": "After listening to the whole recording, output every instance of the word \"grethel\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.75, "grethel"]], "response_mode": "offline", "answer": "[{\"word\":\"grethel\",\"start\":0.4,\"end\":0.75}]"}
{"utterance_id": "6295-64301-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.21, 0.54, "letty"]], "response_mode": "offline", "answer": "[{\"word\":\"letty\",\"start\":0.21,\"end\":0.54}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.68, 7.37, "sancho"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho\",\"start\":6.68,\"end\":7.37}]"}
{"utterance_id": "5694-64025-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0005.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.24, 5.52, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":5.24,\"end\":5.52}]"}
{"utterance_id": "3081-166546-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0052.flac", "task_type": "lexical_detection", "query": "musicians", "prompt": "When the utterance is complete, report all instances of the word \"musicians\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 2.07, "musicians"]], "response_mode": "offline", "answer": "[{\"word\":\"musicians\",\"start\":1.4,\"end\":2.07}]"}
{"utterance_id": "6313-66129-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.52, 3.99, "ponies"]], "response_mode": "offline", "answer": "[{\"word\":\"ponies\",\"start\":3.52,\"end\":3.99}]"}
{"utterance_id": "2428-83699-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0005.flac", "task_type": "lexical_counting", "query": "nothing", "prompt": "Process the full audio offline. For every occurrence of the word \"nothing\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.76, "one"], [3.94, 4.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.45,\"end\":0.76},{\"word\":\"two\",\"start\":3.94,\"end\":4.2}]"}
{"utterance_id": "6313-66129-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0010.flac", "task_type": "lexical_detection", "query": "reckon", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"reckon\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.6, 0.93, "reckon"]], "response_mode": "offline", "answer": "[{\"word\":\"reckon\",\"start\":0.6,\"end\":0.93}]"}
{"utterance_id": "3752-4944-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0047.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.97, 3.33, "face"], [6.33, 6.72, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":2.97,\"end\":3.33},{\"word\":\"head\",\"start\":6.33,\"end\":6.72}]"}
{"utterance_id": "1919-142785-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0003.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Once the audio has finished, list each occurrence of a plant name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.5, 1.17, "one"], [2.46, 2.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.5,\"end\":1.17},{\"word\":\"two\",\"start\":2.46,\"end\":2.86}]"}
{"utterance_id": "8842-304647-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.32, 2.68, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.32,\"end\":2.68}]"}
{"utterance_id": "2078-142845-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0049.flac", "task_type": "lexical_detection", "query": "rusks", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"rusks\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.37, "rusks"]], "response_mode": "offline", "answer": "[{\"word\":\"rusks\",\"start\":1.62,\"end\":2.37}]"}
{"utterance_id": "2803-154328-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0014.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.66, "major"]], "response_mode": "offline", "answer": "[{\"word\":\"major\",\"start\":2.18,\"end\":2.66}]"}
{"utterance_id": "3752-4943-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0009.flac", "task_type": "lexical_detection", "query": "order", "prompt": "After listening to the whole recording, output every instance of the word \"order\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.74, "order"]], "response_mode": "offline", "answer": "[{\"word\":\"order\",\"start\":0.45,\"end\":0.74}]"}
{"utterance_id": "2086-149220-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0037.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.27, 1.58, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":1.27,\"end\":1.58}]"}
{"utterance_id": "2086-149220-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0013.flac", "task_type": "lexical_detection", "query": "antique", "prompt": "Process the full audio offline. Find every instance of the word \"antique\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.47, "antique"]], "response_mode": "offline", "answer": "[{\"word\":\"antique\",\"start\":1.72,\"end\":2.47}]"}
{"utterance_id": "6313-66125-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0001.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.7, "bushes"]], "response_mode": "offline", "answer": "[{\"word\":\"bushes\",\"start\":1.27,\"end\":1.7}]"}
{"utterance_id": "5536-43358-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0001.flac", "task_type": "lexical_detection", "query": "eternal", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"eternal\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.92, 3.57, "eternal"]], "response_mode": "offline", "answer": "[{\"word\":\"eternal\",\"start\":2.92,\"end\":3.57}]"}
{"utterance_id": "422-122949-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.0, 13.37, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":13.0,\"end\":13.37}]"}
{"utterance_id": "6319-64726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.45, 6.93, "feast"]], "response_mode": "offline", "answer": "[{\"word\":\"feast\",\"start\":6.45,\"end\":6.93}]"}
{"utterance_id": "3853-163249-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0046.flac", "task_type": "lexical_detection", "query": "fails", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"fails\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.21, 10.71, "fails"]], "response_mode": "offline", "answer": "[{\"word\":\"fails\",\"start\":10.21,\"end\":10.71}]"}
{"utterance_id": "7850-281318-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0008.flac", "task_type": "lexical_detection", "query": "birds", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"birds\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.06, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":1.61,\"end\":2.06}]"}
{"utterance_id": "652-129742-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0005.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.56, 3.04, "cabbage"]], "response_mode": "offline", "answer": "[{\"word\":\"cabbage\",\"start\":2.56,\"end\":3.04}]"}
{"utterance_id": "3576-138058-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.94, 6.25, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":5.94,\"end\":6.25}]"}
{"utterance_id": "3081-166546-0082", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0082.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "When the utterance is complete, report how a number accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.52, "one"], [3.68, 4.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.52},{\"word\":\"two\",\"start\":3.68,\"end\":4.29}]"}
{"utterance_id": "5895-34629-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.35, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":0.89,\"end\":1.35}]"}
{"utterance_id": "7850-286674-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0013.flac", "task_type": "lexical_detection", "query": "comes", "prompt": "After the audio ends, extract all occurrences of the word \"comes\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.64, "comes"]], "response_mode": "offline", "answer": "[{\"word\":\"comes\",\"start\":0.36,\"end\":0.64}]"}
{"utterance_id": "8297-275154-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0014.flac", "task_type": "lexical_detection", "query": "dressed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"dressed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.25, 6.88, "dressed"]], "response_mode": "offline", "answer": "[{\"word\":\"dressed\",\"start\":6.25,\"end\":6.88}]"}
{"utterance_id": "1673-143397-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0010.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Listen to the full audio. Count every occurrence of a profession. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.57, 4.91, "one"], [5.92, 6.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.57,\"end\":4.91},{\"word\":\"two\",\"start\":5.92,\"end\":6.32}]"}
{"utterance_id": "2086-149220-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.05, 4.52, "egg"]], "response_mode": "offline", "answer": "[{\"word\":\"egg\",\"start\":4.05,\"end\":4.52}]"}
{"utterance_id": "6319-57405-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.48, 2.9, "pyrrha"]], "response_mode": "offline", "answer": "[{\"word\":\"pyrrha\",\"start\":2.48,\"end\":2.9}]"}
{"utterance_id": "1673-143396-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0001.flac", "task_type": "lexical_detection", "query": "disappeared", "prompt": "After the audio ends, extract all occurrences of the word \"disappeared\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 2.0, "disappeared"]], "response_mode": "offline", "answer": "[{\"word\":\"disappeared\",\"start\":1.28,\"end\":2.0}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.88, 4.6, "forefathers"], [8.89, 9.3, "mother"], [16.58, 17.03, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"forefathers\",\"start\":3.88,\"end\":4.6},{\"word\":\"mother\",\"start\":8.89,\"end\":9.3},{\"word\":\"father's\",\"start\":16.58,\"end\":17.03}]"}
{"utterance_id": "8297-275155-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.15, 6.86, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":6.15,\"end\":6.86}]"}
{"utterance_id": "1272-135031-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.39, "ruggedo"]], "response_mode": "offline", "answer": "[{\"word\":\"ruggedo\",\"start\":0.9,\"end\":1.39}]"}
{"utterance_id": "777-126732-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0056.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.14, 3.43, "white"], [6.79, 7.04, "dark"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":3.14,\"end\":3.43},{\"word\":\"dark\",\"start\":6.79,\"end\":7.04}]"}
{"utterance_id": "2035-152373-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0010.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Process the full audio offline. Find every instance of a country name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.13, 1.81, "northumbria"]], "response_mode": "offline", "answer": "[{\"word\":\"northumbria\",\"start\":1.13,\"end\":1.81}]"}
{"utterance_id": "5694-64029-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0031.flac", "task_type": "lexical_detection", "query": "fired", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"fired\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.21, 5.61, "fired"]], "response_mode": "offline", "answer": "[{\"word\":\"fired\",\"start\":5.21,\"end\":5.61}]"}
{"utterance_id": "8842-304647-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.78, "bestia"]], "response_mode": "offline", "answer": "[{\"word\":\"bestia\",\"start\":1.22,\"end\":1.78}]"}
{"utterance_id": "1272-141231-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.08, "two"], [6.79, 7.47, "thousands"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.83,\"end\":4.08},{\"word\":\"thousands\",\"start\":6.79,\"end\":7.47}]"}
{"utterance_id": "2902-9006-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0002.flac", "task_type": "lexical_detection", "query": "centuries", "prompt": "Once the audio has finished, identify every instance of the word \"centuries\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.76, 11.49, "centuries"]], "response_mode": "offline", "answer": "[{\"word\":\"centuries\",\"start\":10.76,\"end\":11.49}]"}
{"utterance_id": "5536-43358-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0017.flac", "task_type": "lexical_counting", "query": "mystery", "prompt": "After listening to the whole recording, output timed count words for the word \"mystery\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.2, 1.62, "one"], [6.21, 6.64, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.2,\"end\":1.62},{\"word\":\"two\",\"start\":6.21,\"end\":6.64}]"}
{"utterance_id": "2277-149874-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0011.flac", "task_type": "lexical_detection", "query": "thinking", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"thinking\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.88, 1.22, "thinking"]], "response_mode": "offline", "answer": "[{\"word\":\"thinking\",\"start\":0.88,\"end\":1.22}]"}
{"utterance_id": "652-130737-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0005.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 1.02, "clarets"]], "response_mode": "offline", "answer": "[{\"word\":\"clarets\",\"start\":0.49,\"end\":1.02}]"}
{"utterance_id": "1919-142785-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0059.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.9, 2.16, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.9,\"end\":2.16}]"}
{"utterance_id": "251-118436-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0017.flac", "task_type": "lexical_detection", "query": "fingers", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fingers\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 1.07, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":0.48,\"end\":1.07}]"}
{"utterance_id": "2803-154328-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.7, 6.45, "eyelids"], [8.28, 8.8, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"eyelids\",\"start\":5.7,\"end\":6.45},{\"word\":\"arms\",\"start\":8.28,\"end\":8.8}]"}
{"utterance_id": "2803-154328-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.81, "robert"]], "response_mode": "offline", "answer": "[{\"word\":\"robert\",\"start\":2.43,\"end\":2.81}]"}
{"utterance_id": "1919-142785-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0062.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.23, 3.63, "salmon"]], "response_mode": "offline", "answer": "[{\"word\":\"salmon\",\"start\":3.23,\"end\":3.63}]"}
{"utterance_id": "777-126732-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0027.flac", "task_type": "lexical_detection", "query": "skin", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"skin\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.0, "skin"]], "response_mode": "offline", "answer": "[{\"word\":\"skin\",\"start\":0.66,\"end\":1.0}]"}
{"utterance_id": "174-168635-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.58, "jean valjean"]], "response_mode": "offline", "answer": "[{\"word\":\"jean valjean\",\"start\":1.62,\"end\":2.58}]"}
{"utterance_id": "251-136532-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0018.flac", "task_type": "lexical_detection", "query": "question", "prompt": "Once the audio has finished, identify every instance of the word \"question\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.72, 1.38, "question"]], "response_mode": "offline", "answer": "[{\"word\":\"question\",\"start\":0.72,\"end\":1.38}]"}
{"utterance_id": "2902-9008-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0005.flac", "task_type": "lexical_detection", "query": "cast", "prompt": "Process the full audio offline. Find every instance of the word \"cast\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.29, "cast"], [3.51, 3.86, "cast"]], "response_mode": "offline", "answer": "[{\"word\":\"cast\",\"start\":0.86,\"end\":1.29},{\"word\":\"cast\",\"start\":3.51,\"end\":3.86}]"}
{"utterance_id": "2086-149220-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0005.flac", "task_type": "lexical_detection", "query": "bees", "prompt": "When the utterance is complete, report all instances of the word \"bees\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.84, "bees"]], "response_mode": "offline", "answer": "[{\"word\":\"bees\",\"start\":0.5,\"end\":0.84}]"}
{"utterance_id": "1993-147149-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0009.flac", "task_type": "lexical_detection", "query": "upstairs", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"upstairs\" with fields word, start, end (seconds). If none, reply [].", "labels": [[14.03, 14.66, "upstairs"]], "response_mode": "offline", "answer": "[{\"word\":\"upstairs\",\"start\":14.03,\"end\":14.66}]"}
{"utterance_id": "2803-161169-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.38, 10.61, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.38,\"end\":10.61}]"}
{"utterance_id": "3536-8226-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.21, "judge"], [11.31, 11.85, "lawyers"]], "response_mode": "offline", "answer": "[{\"word\":\"judge\",\"start\":1.9,\"end\":2.21},{\"word\":\"lawyers\",\"start\":11.31,\"end\":11.85}]"}
{"utterance_id": "6241-66616-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0005.flac", "task_type": "lexical_detection", "query": "generation", "prompt": "After listening to the whole recording, output every instance of the word \"generation\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.76, 10.51, "generation"]], "response_mode": "offline", "answer": "[{\"word\":\"generation\",\"start\":9.76,\"end\":10.51}]"}
{"utterance_id": "3000-15664-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0016.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.68, "emerald"]], "response_mode": "offline", "answer": "[{\"word\":\"emerald\",\"start\":1.28,\"end\":1.68}]"}
{"utterance_id": "2412-153954-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0003.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 0.9, "vine"]], "response_mode": "offline", "answer": "[{\"word\":\"vine\",\"start\":0.56,\"end\":0.9}]"}
{"utterance_id": "8297-275156-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0003.flac", "task_type": "lexical_counting", "query": "gossip", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"gossip\" (word, start, end). If none, reply [].", "labels": [[1.75, 2.22, "one"], [5.38, 5.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.75,\"end\":2.22},{\"word\":\"two\",\"start\":5.38,\"end\":5.85}]"}
{"utterance_id": "1993-147964-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 1.71, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":1.38,\"end\":1.71}]"}
{"utterance_id": "2035-147961-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0005.flac", "task_type": "lexical_detection", "query": "armies", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"armies\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.39, "armies"]], "response_mode": "offline", "answer": "[{\"word\":\"armies\",\"start\":1.87,\"end\":2.39}]"}
{"utterance_id": "777-126732-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.18, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":3.79,\"end\":4.18}]"}
{"utterance_id": "652-129742-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0009.flac", "task_type": "lexical_counting", "query": "cups", "prompt": "After listening to the whole recording, output timed count words for the word \"cups\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[5.02, 5.53, "one"], [10.07, 10.51, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.02,\"end\":5.53},{\"word\":\"two\",\"start\":10.07,\"end\":10.51}]"}
{"utterance_id": "3853-163249-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.0, "ear"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":1.77,\"end\":2.0}]"}
{"utterance_id": "5338-284437-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.74, 7.2, "servant"]], "response_mode": "offline", "answer": "[{\"word\":\"servant\",\"start\":6.74,\"end\":7.2}]"}
{"utterance_id": "1988-24833-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0015.flac", "task_type": "lexical_detection", "query": "upset", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"upset\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.85, "upset"]], "response_mode": "offline", "answer": "[{\"word\":\"upset\",\"start\":1.47,\"end\":1.85}]"}
{"utterance_id": "6295-244435-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0016.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.01, 6.55, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":6.01,\"end\":6.55}]"}
{"utterance_id": "3853-163249-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0023.flac", "task_type": "lexical_detection", "query": "you've", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"you've\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.24, 0.43, "you've"]], "response_mode": "offline", "answer": "[{\"word\":\"you've\",\"start\":0.24,\"end\":0.43}]"}
{"utterance_id": "6295-244435-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.74, 9.28, "soldier"]], "response_mode": "offline", "answer": "[{\"word\":\"soldier\",\"start\":8.74,\"end\":9.28}]"}
{"utterance_id": "6319-57405-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0001.flac", "task_type": "lexical_detection", "query": "jupiter", "prompt": "After listening to the whole recording, output every instance of the word \"jupiter\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.4, "jupiter"]], "response_mode": "offline", "answer": "[{\"word\":\"jupiter\",\"start\":0.91,\"end\":1.4}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "lexical_counting", "query": "nobleness", "prompt": "Listen to the full audio. Count every occurrence of the word \"nobleness\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.56, "one"], [10.08, 10.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.13,\"end\":4.56},{\"word\":\"two\",\"start\":10.08,\"end\":10.48}]"}
{"utterance_id": "1988-148538-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0006.flac", "task_type": "lexical_counting", "query": "objects", "prompt": "After listening to the whole recording, output timed count words for the word \"objects\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[12.48, 12.86, "one"], [14.43, 14.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":12.48,\"end\":12.86},{\"word\":\"two\",\"start\":14.43,\"end\":14.9}]"}
{"utterance_id": "1272-135031-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.16, "shaggy"]], "response_mode": "offline", "answer": "[{\"word\":\"shaggy\",\"start\":2.74,\"end\":3.16}]"}
{"utterance_id": "2086-149220-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0046.flac", "task_type": "lexical_detection", "query": "need", "prompt": "Process the full audio offline. Find every instance of the word \"need\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.36, 2.57, "need"]], "response_mode": "offline", "answer": "[{\"word\":\"need\",\"start\":2.36,\"end\":2.57}]"}
{"utterance_id": "3752-4944-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0005.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 2.06, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":1.54,\"end\":2.06}]"}
{"utterance_id": "84-121123-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0002.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.39, 8.83, "purple"]], "response_mode": "offline", "answer": "[{\"word\":\"purple\",\"start\":8.39,\"end\":8.83}]"}
{"utterance_id": "2428-83705-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0006.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After the audio ends, extract all occurrences of a date. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 2.9, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":2.54,\"end\":2.9}]"}
{"utterance_id": "2035-147960-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.85, "snake"]], "response_mode": "offline", "answer": "[{\"word\":\"snake\",\"start\":4.45,\"end\":4.85}]"}
{"utterance_id": "2412-153948-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0012.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After listening to the whole recording, output every instance of a time of day with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.34, 5.92, "nightfall"]], "response_mode": "offline", "answer": "[{\"word\":\"nightfall\",\"start\":5.34,\"end\":5.92}]"}
{"utterance_id": "2277-149874-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0013.flac", "task_type": "lexical_detection", "query": "easy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"easy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.31, "easy"]], "response_mode": "offline", "answer": "[{\"word\":\"easy\",\"start\":1.01,\"end\":1.31}]"}
{"utterance_id": "1988-147956-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.15, "otto"]], "response_mode": "offline", "answer": "[{\"word\":\"otto\",\"start\":1.74,\"end\":2.15}]"}
{"utterance_id": "6241-66616-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.74, 3.45, "minnetaki"]], "response_mode": "offline", "answer": "[{\"word\":\"minnetaki\",\"start\":2.74,\"end\":3.45}]"}
{"utterance_id": "5895-34629-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0021.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.93, 5.43, "drugs"]], "response_mode": "offline", "answer": "[{\"word\":\"drugs\",\"start\":4.93,\"end\":5.43}]"}
{"utterance_id": "7850-286674-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0016.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.87, "one"], [10.36, 10.82, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.43,\"end\":1.87},{\"word\":\"two\",\"start\":10.36,\"end\":10.82}]"}
{"utterance_id": "1919-142785-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0044.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.18, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":0.81,\"end\":1.18}]"}
{"utterance_id": "1993-147149-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.28, 13.77, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":13.28,\"end\":13.77}]"}
{"utterance_id": "8842-304647-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0007.flac", "task_type": "lexical_detection", "query": "wonderful", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wonderful\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.9, 1.61, "wonderful"]], "response_mode": "offline", "answer": "[{\"word\":\"wonderful\",\"start\":0.9,\"end\":1.61}]"}
{"utterance_id": "2803-154320-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.6, "john mangles"]], "response_mode": "offline", "answer": "[{\"word\":\"john mangles\",\"start\":0.53,\"end\":1.6}]"}
{"utterance_id": "1993-147149-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.81, 3.17, "brothers"]], "response_mode": "offline", "answer": "[{\"word\":\"brothers\",\"start\":2.81,\"end\":3.17}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Once the audio has finished, list each occurrence of a color name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.88, 2.44, "one"], [4.98, 5.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.88,\"end\":2.44},{\"word\":\"two\",\"start\":4.98,\"end\":5.35}]"}
{"utterance_id": "6313-76958-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.31, 2.84, "cowboys"]], "response_mode": "offline", "answer": "[{\"word\":\"cowboys\",\"start\":2.31,\"end\":2.84}]"}
{"utterance_id": "3081-166546-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.2, 2.6, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.2,\"end\":2.6}]"}
{"utterance_id": "2086-149220-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0026.flac", "task_type": "lexical_detection", "query": "common", "prompt": "When the utterance is complete, report all instances of the word \"common\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.2, "common"]], "response_mode": "offline", "answer": "[{\"word\":\"common\",\"start\":1.82,\"end\":2.2}]"}
{"utterance_id": "3576-138058-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.53, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.01,\"end\":2.53}]"}
{"utterance_id": "6345-93302-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0029.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Do not answer until the audio ends. Then list each instance of a country name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.79, 2.63, "south africa"]], "response_mode": "offline", "answer": "[{\"word\":\"south africa\",\"start\":1.79,\"end\":2.63}]"}
{"utterance_id": "2428-83705-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0042.flac", "task_type": "lexical_counting", "query": "free", "prompt": "When the utterance is complete, report how the word \"free\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.3, "one"], [7.16, 7.4, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.03,\"end\":1.3},{\"word\":\"two\",\"start\":7.16,\"end\":7.4}]"}
{"utterance_id": "1462-170138-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0003.flac", "task_type": "lexical_detection", "query": "exclaimed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"exclaimed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.59, "exclaimed"]], "response_mode": "offline", "answer": "[{\"word\":\"exclaimed\",\"start\":0.88,\"end\":1.59}]"}
{"utterance_id": "3576-138058-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0033.flac", "task_type": "lexical_detection", "query": "roads", "prompt": "When the utterance is complete, report all instances of the word \"roads\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.62, 6.1, "roads"]], "response_mode": "offline", "answer": "[{\"word\":\"roads\",\"start\":5.62,\"end\":6.1}]"}
{"utterance_id": "3081-166546-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0020.flac", "task_type": "lexical_detection", "query": "police", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"police\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.61, 4.0, "police"]], "response_mode": "offline", "answer": "[{\"word\":\"police\",\"start\":3.61,\"end\":4.0}]"}
{"utterance_id": "652-130737-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0011.flac", "task_type": "lexical_detection", "query": "quality", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"quality\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.58, 2.99, "quality"]], "response_mode": "offline", "answer": "[{\"word\":\"quality\",\"start\":2.58,\"end\":2.99}]"}
{"utterance_id": "1993-147964-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.33, 2.76, "horses"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":2.33,\"end\":2.76}]"}
{"utterance_id": "6319-275224-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.79, 6.29, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":5.79,\"end\":6.29}]"}
{"utterance_id": "8297-275155-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0023.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.19, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.83,\"end\":2.19}]"}
{"utterance_id": "1988-24833-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0012.flac", "task_type": "lexical_detection", "query": "bone", "prompt": "Process the full audio offline. Find every instance of the word \"bone\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.26, "bone"]], "response_mode": "offline", "answer": "[{\"word\":\"bone\",\"start\":1.01,\"end\":1.26}]"}
{"utterance_id": "5694-64025-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0016.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After listening to the whole recording, output every instance of an organization name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.38, 9.92, "confederate"]], "response_mode": "offline", "answer": "[{\"word\":\"confederate\",\"start\":9.38,\"end\":9.92}]"}
{"utterance_id": "3853-163249-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0008.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.2, 0.48, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":0.2,\"end\":0.48}]"}
{"utterance_id": "2277-149896-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 0.79, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.61,\"end\":0.79}]"}
{"utterance_id": "174-84280-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0002.flac", "task_type": "lexical_detection", "query": "aged", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"aged\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.1, 9.58, "aged"]], "response_mode": "offline", "answer": "[{\"word\":\"aged\",\"start\":9.1,\"end\":9.58}]"}
{"utterance_id": "777-126732-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0055.flac", "task_type": "lexical_detection", "query": "cage", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cage\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.51, "cage"]], "response_mode": "offline", "answer": "[{\"word\":\"cage\",\"start\":2.93,\"end\":3.51}]"}
{"utterance_id": "6241-61943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 2.05, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":1.71,\"end\":2.05}]"}
{"utterance_id": "251-137823-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0020.flac", "task_type": "lexical_detection", "query": "laughed", "prompt": "Process the full audio offline. Find every instance of the word \"laughed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.6, 5.1, "laughed"]], "response_mode": "offline", "answer": "[{\"word\":\"laughed\",\"start\":4.6,\"end\":5.1}]"}
{"utterance_id": "3000-15664-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.39, 12.9, "snake"]], "response_mode": "offline", "answer": "[{\"word\":\"snake\",\"start\":12.39,\"end\":12.9}]"}
{"utterance_id": "2277-149896-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.16, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.92,\"end\":2.16}]"}
{"utterance_id": "7976-105575-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0023.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.31, "dozens"]], "response_mode": "offline", "answer": "[{\"word\":\"dozens\",\"start\":1.61,\"end\":2.31}]"}
{"utterance_id": "5536-43363-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0015.flac", "task_type": "lexical_detection", "query": "service", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"service\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.82, 2.29, "service"]], "response_mode": "offline", "answer": "[{\"word\":\"service\",\"start\":1.82,\"end\":2.29}]"}
{"utterance_id": "7976-110124-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0025.flac", "task_type": "lexical_detection", "query": "worries", "prompt": "Process the full audio offline. Find every instance of the word \"worries\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.1, "worries"]], "response_mode": "offline", "answer": "[{\"word\":\"worries\",\"start\":0.67,\"end\":1.1}]"}
{"utterance_id": "2277-149874-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0005.flac", "task_type": "lexical_detection", "query": "observation", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"observation\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.64, "observation"]], "response_mode": "offline", "answer": "[{\"word\":\"observation\",\"start\":1.72,\"end\":2.64}]"}
{"utterance_id": "5338-284437-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.11, 6.5, "skin"]], "response_mode": "offline", "answer": "[{\"word\":\"skin\",\"start\":6.11,\"end\":6.5}]"}
{"utterance_id": "3536-23268-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0027.flac", "task_type": "lexical_detection", "query": "sincerity", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sincerity\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.31, "sincerity"]], "response_mode": "offline", "answer": "[{\"word\":\"sincerity\",\"start\":2.54,\"end\":3.31}]"}
{"utterance_id": "5895-34622-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0003.flac", "task_type": "lexical_detection", "query": "eighty", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"eighty\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.21, 1.49, "eighty"]], "response_mode": "offline", "answer": "[{\"word\":\"eighty\",\"start\":1.21,\"end\":1.49}]"}
{"utterance_id": "8842-302203-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0009.flac", "task_type": "lexical_counting", "query": "name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"name\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.02, 1.2, "one"], [10.97, 11.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.02,\"end\":1.2},{\"word\":\"two\",\"start\":10.97,\"end\":11.16}]"}
{"utterance_id": "652-130737-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0009.flac", "task_type": "lexical_detection", "query": "bouquet", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"bouquet\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.2, "bouquet"]], "response_mode": "offline", "answer": "[{\"word\":\"bouquet\",\"start\":3.83,\"end\":4.2}]"}
{"utterance_id": "3752-4944-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.83, "mister meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"mister meekin\",\"start\":1.14,\"end\":1.83}]"}
{"utterance_id": "3536-23268-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0006.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.74, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.4,\"end\":3.74}]"}
{"utterance_id": "5338-284437-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0010.flac", "task_type": "semantic_detection", "query": "color", "prompt": "You will hear a complete utterance. When it ends, list every instance of a color name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.49, "pink"], [3.71, 4.1, "rosy"]], "response_mode": "offline", "answer": "[{\"word\":\"pink\",\"start\":2.06,\"end\":2.49},{\"word\":\"rosy\",\"start\":3.71,\"end\":4.1}]"}
{"utterance_id": "6313-66129-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0021.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.43, 2.86, "ponies"]], "response_mode": "offline", "answer": "[{\"word\":\"ponies\",\"start\":2.43,\"end\":2.86}]"}
{"utterance_id": "3752-4944-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0012.flac", "task_type": "lexical_counting", "query": "seeing", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"seeing\" with word/start/end. If none, reply [].", "labels": [[2.73, 3.06, "one"], [4.8, 5.11, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.73,\"end\":3.06},{\"word\":\"two\",\"start\":4.8,\"end\":5.11}]"}
{"utterance_id": "3752-4944-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0056.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 0.66, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":0.26,\"end\":0.66}]"}
{"utterance_id": "84-121123-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 0.55, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.37,\"end\":0.55}]"}
{"utterance_id": "7850-281318-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0023.flac", "task_type": "lexical_detection", "query": "shown", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"shown\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.47, 4.75, "shown"]], "response_mode": "offline", "answer": "[{\"word\":\"shown\",\"start\":4.47,\"end\":4.75}]"}
{"utterance_id": "5694-64038-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.85, "atlanta"]], "response_mode": "offline", "answer": "[{\"word\":\"atlanta\",\"start\":1.31,\"end\":1.85}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.95, "stomach"], [4.96, 5.53, "stomach"]], "response_mode": "offline", "answer": "[{\"word\":\"stomach\",\"start\":1.41,\"end\":1.95},{\"word\":\"stomach\",\"start\":4.96,\"end\":5.53}]"}
{"utterance_id": "422-122949-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0033.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.03, 3.6, "harpies"]], "response_mode": "offline", "answer": "[{\"word\":\"harpies\",\"start\":3.03,\"end\":3.6}]"}
{"utterance_id": "6313-66129-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0013.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.42, "dog"]], "response_mode": "offline", "answer": "[{\"word\":\"dog\",\"start\":2.05,\"end\":2.42}]"}
{"utterance_id": "6241-66616-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0017.flac", "task_type": "lexical_counting", "query": "thousand", "prompt": "Listen to the full audio. Count every occurrence of the word \"thousand\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.98, "one"], [2.1, 2.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.48,\"end\":0.98},{\"word\":\"two\",\"start\":2.1,\"end\":2.54}]"}
{"utterance_id": "3081-166546-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0037.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.03, 4.32, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.03,\"end\":4.32}]"}
{"utterance_id": "2078-142845-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0000.flac", "task_type": "lexical_detection", "query": "yeast", "prompt": "Once the audio has finished, identify every instance of the word \"yeast\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.29, 1.97, "yeast"]], "response_mode": "offline", "answer": "[{\"word\":\"yeast\",\"start\":1.29,\"end\":1.97}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "lexical_detection", "query": "butler", "prompt": "Process the full audio offline. Find every instance of the word \"butler\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.1, 6.44, "butler"]], "response_mode": "offline", "answer": "[{\"word\":\"butler\",\"start\":6.1,\"end\":6.44}]"}
{"utterance_id": "2428-83705-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0018.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Do not answer until the audio ends. Then list each instance of a money or currency term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.66, 2.92, "money"], [4.91, 5.34, "pounds"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":2.66,\"end\":2.92},{\"word\":\"pounds\",\"start\":4.91,\"end\":5.34}]"}
{"utterance_id": "251-118436-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.09, 2.59, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":2.09,\"end\":2.59}]"}
{"utterance_id": "2902-9006-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0020.flac", "task_type": "lexical_detection", "query": "pages", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"pages\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.15, "pages"]], "response_mode": "offline", "answer": "[{\"word\":\"pages\",\"start\":1.64,\"end\":2.15}]"}
{"utterance_id": "6313-76958-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0022.flac", "task_type": "lexical_detection", "query": "tail", "prompt": "Once the audio has finished, identify every instance of the word \"tail\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.63, 5.01, "tail"]], "response_mode": "offline", "answer": "[{\"word\":\"tail\",\"start\":4.63,\"end\":5.01}]"}
{"utterance_id": "1993-147149-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0018.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[7.08, 7.61, "one"], [8.81, 9.24, "two"], [12.65, 13.14, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.08,\"end\":7.61},{\"word\":\"two\",\"start\":8.81,\"end\":9.24},{\"word\":\"three\",\"start\":12.65,\"end\":13.14}]"}
{"utterance_id": "5694-64029-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0004.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After listening to the whole recording, output every instance of an organization name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.63, 0.96, "yankee"]], "response_mode": "offline", "answer": "[{\"word\":\"yankee\",\"start\":0.63,\"end\":0.96}]"}
{"utterance_id": "6295-244435-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[6.6, 7.06, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.6,\"end\":7.06}]"}
{"utterance_id": "1462-170145-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.26, "manager"]], "response_mode": "offline", "answer": "[{\"word\":\"manager\",\"start\":0.73,\"end\":1.26}]"}
{"utterance_id": "3170-137482-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0013.flac", "task_type": "lexical_detection", "query": "husband", "prompt": "When the utterance is complete, report all instances of the word \"husband\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.99, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":1.18,\"end\":1.99}]"}
{"utterance_id": "3081-166546-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0012.flac", "task_type": "lexical_detection", "query": "persons", "prompt": "After the audio ends, extract all occurrences of the word \"persons\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.19, 4.73, "persons"]], "response_mode": "offline", "answer": "[{\"word\":\"persons\",\"start\":4.19,\"end\":4.73}]"}
{"utterance_id": "2428-83699-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.28, 2.67, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.28,\"end\":2.67}]"}
{"utterance_id": "8842-302201-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0004.flac", "task_type": "lexical_detection", "query": "sighs", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sighs\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.72, 7.33, "sighs"]], "response_mode": "offline", "answer": "[{\"word\":\"sighs\",\"start\":6.72,\"end\":7.33}]"}
{"utterance_id": "2086-149220-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0030.flac", "task_type": "lexical_detection", "query": "dozen", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"dozen\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.8, 6.08, "dozen"]], "response_mode": "offline", "answer": "[{\"word\":\"dozen\",\"start\":5.8,\"end\":6.08}]"}
{"utterance_id": "652-130726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.84, 5.09, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":4.84,\"end\":5.09}]"}
{"utterance_id": "8842-302196-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.19, 7.42, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.19,\"end\":7.42}]"}
{"utterance_id": "5895-34622-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0022.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.05, 7.71, "preachers"]], "response_mode": "offline", "answer": "[{\"word\":\"preachers\",\"start\":7.05,\"end\":7.71}]"}
{"utterance_id": "2428-83705-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0008.flac", "task_type": "lexical_detection", "query": "gasped", "prompt": "When the utterance is complete, report all instances of the word \"gasped\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.16, "gasped"], [2.64, 3.24, "gasped"]], "response_mode": "offline", "answer": "[{\"word\":\"gasped\",\"start\":0.6,\"end\":1.16},{\"word\":\"gasped\",\"start\":2.64,\"end\":3.24}]"}
{"utterance_id": "5694-64025-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0010.flac", "task_type": "lexical_detection", "query": "battlefields", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"battlefields\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.72, 2.62, "battlefields"], [3.66, 4.52, "battlefields"]], "response_mode": "offline", "answer": "[{\"word\":\"battlefields\",\"start\":1.72,\"end\":2.62},{\"word\":\"battlefields\",\"start\":3.66,\"end\":4.52}]"}
{"utterance_id": "7976-105575-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.17, 0.75, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.17,\"end\":0.75}]"}
{"utterance_id": "5694-64038-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0024.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an organization name with word/start/end. If none, reply [].", "labels": [[0.42, 0.82, "one"], [2.82, 3.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.42,\"end\":0.82},{\"word\":\"two\",\"start\":2.82,\"end\":3.29}]"}
{"utterance_id": "2035-147961-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0020.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.74, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"black\",\"start\":0.39,\"end\":0.74}]"}
{"utterance_id": "1919-142785-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0000.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After listening to the whole recording, output every instance of a plant name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.23, "pepper"]], "response_mode": "offline", "answer": "[{\"word\":\"pepper\",\"start\":1.84,\"end\":2.23}]"}
{"utterance_id": "652-130737-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.01, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.87,\"end\":2.01}]"}
{"utterance_id": "1988-24833-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0014.flac", "task_type": "lexical_detection", "query": "channel", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"channel\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.85, 2.22, "channel"]], "response_mode": "offline", "answer": "[{\"word\":\"channel\",\"start\":1.85,\"end\":2.22}]"}
{"utterance_id": "1462-170138-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0017.flac", "task_type": "lexical_detection", "query": "bowed", "prompt": "After the audio ends, extract all occurrences of the word \"bowed\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 1.99, "bowed"]], "response_mode": "offline", "answer": "[{\"word\":\"bowed\",\"start\":1.49,\"end\":1.99}]"}
{"utterance_id": "6295-64301-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0024.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.92, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.35,\"end\":1.92}]"}
{"utterance_id": "6241-61946-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0010.flac", "task_type": "lexical_detection", "query": "stiff", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"stiff\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 3.0, "stiff"]], "response_mode": "offline", "answer": "[{\"word\":\"stiff\",\"start\":2.43,\"end\":3.0}]"}
{"utterance_id": "5895-34615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.98, 9.51, "flesh"], [13.69, 14.41, "countenance"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":8.98,\"end\":9.51},{\"word\":\"countenance\",\"start\":13.69,\"end\":14.41}]"}
{"utterance_id": "2428-83705-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.87, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":0.5,\"end\":0.87}]"}
{"utterance_id": "2277-149874-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0018.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the full audio. After it finishes, find every instance of a time of day. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.61, 3.88, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":3.61,\"end\":3.88}]"}
{"utterance_id": "1462-170145-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0006.flac", "task_type": "lexical_detection", "query": "you'd", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"you'd\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.43, 0.56, "you'd"]], "response_mode": "offline", "answer": "[{\"word\":\"you'd\",\"start\":0.43,\"end\":0.56}]"}
{"utterance_id": "1272-128104-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0003.flac", "task_type": "lexical_detection", "query": "doubts", "prompt": "Once the audio has finished, identify every instance of the word \"doubts\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.28, 1.72, "doubts"]], "response_mode": "offline", "answer": "[{\"word\":\"doubts\",\"start\":1.28,\"end\":1.72}]"}
{"utterance_id": "84-121123-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.64, 8.08, "face"], [9.56, 10.0, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":7.64,\"end\":8.08},{\"word\":\"lips\",\"start\":9.56,\"end\":10.0}]"}
{"utterance_id": "84-121550-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0012.flac", "task_type": "lexical_detection", "query": "moved", "prompt": "After the audio ends, extract all occurrences of the word \"moved\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.23, "moved"]], "response_mode": "offline", "answer": "[{\"word\":\"moved\",\"start\":3.79,\"end\":4.23}]"}
{"utterance_id": "6345-93306-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0010.flac", "task_type": "lexical_detection", "query": "aware", "prompt": "When the utterance is complete, report all instances of the word \"aware\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.33, "aware"]], "response_mode": "offline", "answer": "[{\"word\":\"aware\",\"start\":1.87,\"end\":2.33}]"}
{"utterance_id": "2902-9008-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0002.flac", "task_type": "lexical_counting", "query": "room", "prompt": "When the utterance is complete, report how the word \"room\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 3.31, "one"], [15.2, 15.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.03,\"end\":3.31},{\"word\":\"two\",\"start\":15.2,\"end\":15.47}]"}
{"utterance_id": "777-126732-0069", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0069.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Process the full audio offline. Find every instance of a time of day and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 1.82, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":1.55,\"end\":1.82}]"}
{"utterance_id": "174-84280-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.71, 1.16, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":0.71,\"end\":1.16}]"}
{"utterance_id": "6319-275224-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0020.flac", "task_type": "lexical_detection", "query": "tie", "prompt": "When the utterance is complete, report all instances of the word \"tie\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.35, 9.66, "tie"]], "response_mode": "offline", "answer": "[{\"word\":\"tie\",\"start\":9.35,\"end\":9.66}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "lexical_detection", "query": "monastery", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"monastery\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.26, "monastery"]], "response_mode": "offline", "answer": "[{\"word\":\"monastery\",\"start\":2.46,\"end\":3.26}]"}
{"utterance_id": "2428-83699-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0003.flac", "task_type": "semantic_counting", "query": "money", "prompt": "After the audio ends, tally every instance of a money or currency term. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.03, 2.54, "one"], [5.09, 5.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.03,\"end\":2.54},{\"word\":\"two\",\"start\":5.09,\"end\":5.34}]"}
{"utterance_id": "5536-43363-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0007.flac", "task_type": "lexical_counting", "query": "food", "prompt": "After the audio ends, tally every instance of the word \"food\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.95, 2.24, "one"], [9.07, 9.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.95,\"end\":2.24},{\"word\":\"two\",\"start\":9.07,\"end\":9.46}]"}
{"utterance_id": "2035-152373-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.36, 8.74, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":8.36,\"end\":8.74}]"}
{"utterance_id": "3853-163249-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0054.flac", "task_type": "lexical_detection", "query": "souls", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"souls\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.83, 3.36, "souls"]], "response_mode": "offline", "answer": "[{\"word\":\"souls\",\"start\":2.83,\"end\":3.36}]"}
{"utterance_id": "2035-152373-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0015.flac", "task_type": "lexical_detection", "query": "holy", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"holy\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.11, 1.41, "holy"]], "response_mode": "offline", "answer": "[{\"word\":\"holy\",\"start\":1.11,\"end\":1.41}]"}
{"utterance_id": "5694-64029-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0030.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.51, 2.75, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":2.51,\"end\":2.75}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "lexical_detection", "query": "reverend", "prompt": "After listening to the whole recording, output every instance of the word \"reverend\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.7, "reverend"], [3.58, 3.98, "reverend"]], "response_mode": "offline", "answer": "[{\"word\":\"reverend\",\"start\":0.33,\"end\":0.7},{\"word\":\"reverend\",\"start\":3.58,\"end\":3.98}]"}
{"utterance_id": "5895-34622-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.23, "ursus"], [3.47, 3.95, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":1.72,\"end\":2.23},{\"word\":\"gwynplaine\",\"start\":3.47,\"end\":3.95}]"}
{"utterance_id": "2078-142845-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0044.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.32, "italian"]], "response_mode": "offline", "answer": "[{\"word\":\"italian\",\"start\":0.52,\"end\":1.32}]"}
{"utterance_id": "3536-23268-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.99, "hand"], [11.31, 11.65, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":0.67,\"end\":0.99},{\"word\":\"heart\",\"start\":11.31,\"end\":11.65}]"}
{"utterance_id": "3170-137482-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.37, "profession"]], "response_mode": "offline", "answer": "[{\"word\":\"profession\",\"start\":1.76,\"end\":2.37}]"}
{"utterance_id": "3536-23268-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.62, 2.23, "miss woodley"]], "response_mode": "offline", "answer": "[{\"word\":\"miss woodley\",\"start\":1.62,\"end\":2.23}]"}
{"utterance_id": "2035-147961-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0021.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[2.85, 3.2, "driver"]], "response_mode": "offline", "answer": "[{\"word\":\"driver\",\"start\":2.85,\"end\":3.2}]"}
{"utterance_id": "5694-64029-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0020.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.14, "twenty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty\",\"start\":0.85,\"end\":1.14}]"}
{"utterance_id": "2277-149897-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0029.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Process the full audio offline. Find every instance of a city name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.36, 4.71, "ogden"]], "response_mode": "offline", "answer": "[{\"word\":\"ogden\",\"start\":4.36,\"end\":4.71}]"}
{"utterance_id": "7850-281318-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0011.flac", "task_type": "lexical_detection", "query": "class", "prompt": "Process the full audio offline. Find every instance of the word \"class\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.91, 6.37, "class"]], "response_mode": "offline", "answer": "[{\"word\":\"class\",\"start\":5.91,\"end\":6.37}]"}
{"utterance_id": "2035-147960-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0009.flac", "task_type": "lexical_detection", "query": "tony", "prompt": "After listening to the whole recording, output every instance of the word \"tony\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.4, "tony"]], "response_mode": "offline", "answer": "[{\"word\":\"tony\",\"start\":0.94,\"end\":1.4}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.31, 5.5, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":5.31,\"end\":5.5}]"}
{"utterance_id": "2428-83705-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0041.flac", "task_type": "lexical_detection", "query": "line", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"line\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.49, 3.86, "line"]], "response_mode": "offline", "answer": "[{\"word\":\"line\",\"start\":3.49,\"end\":3.86}]"}
{"utterance_id": "6295-64301-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0029.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[2.55, 2.88, "one"], [4.37, 4.82, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.55,\"end\":2.88},{\"word\":\"two\",\"start\":4.37,\"end\":4.82}]"}
{"utterance_id": "2035-147961-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0009.flac", "task_type": "lexical_detection", "query": "raged", "prompt": "After the audio ends, extract all occurrences of the word \"raged\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.68, "raged"]], "response_mode": "offline", "answer": "[{\"word\":\"raged\",\"start\":1.22,\"end\":1.68}]"}
{"utterance_id": "1673-143396-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.55, 15.09, "flesh"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":14.55,\"end\":15.09}]"}
{"utterance_id": "3536-23268-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0005.flac", "task_type": "lexical_detection", "query": "evening", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"evening\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.61, 3.95, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":3.61,\"end\":3.95}]"}
{"utterance_id": "84-121550-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.2, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":1.8,\"end\":2.2}]"}
{"utterance_id": "3170-137482-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0014.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[2.4, 3.02, "morrow"]], "response_mode": "offline", "answer": "[{\"word\":\"morrow\",\"start\":2.4,\"end\":3.02}]"}
{"utterance_id": "5694-64029-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.95, "shoulder"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulder\",\"start\":1.52,\"end\":1.95}]"}
{"utterance_id": "777-126732-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0061.flac", "task_type": "lexical_detection", "query": "pane", "prompt": "After listening to the whole recording, output every instance of the word \"pane\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.36, "pane"]], "response_mode": "offline", "answer": "[{\"word\":\"pane\",\"start\":2.07,\"end\":2.36}]"}
{"utterance_id": "6295-64301-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0023.flac", "task_type": "lexical_detection", "query": "lovest", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lovest\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.22, "lovest"]], "response_mode": "offline", "answer": "[{\"word\":\"lovest\",\"start\":0.84,\"end\":1.22}]"}
{"utterance_id": "251-118436-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0020.flac", "task_type": "lexical_detection", "query": "fast", "prompt": "After the audio ends, extract all occurrences of the word \"fast\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.71, 5.31, "fast"]], "response_mode": "offline", "answer": "[{\"word\":\"fast\",\"start\":4.71,\"end\":5.31}]"}
{"utterance_id": "777-126732-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0034.flac", "task_type": "lexical_detection", "query": "protested", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"protested\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.0, "protested"]], "response_mode": "offline", "answer": "[{\"word\":\"protested\",\"start\":1.44,\"end\":2.0}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After the audio ends, tally every instance of a food name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.9, 1.43, "one"], [4.15, 4.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.9,\"end\":1.43},{\"word\":\"two\",\"start\":4.15,\"end\":4.59}]"}
{"utterance_id": "2428-83699-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0007.flac", "task_type": "lexical_detection", "query": "boyhood", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"boyhood\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.88, 3.46, "boyhood"]], "response_mode": "offline", "answer": "[{\"word\":\"boyhood\",\"start\":2.88,\"end\":3.46}]"}
{"utterance_id": "7976-110523-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.18, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cat\",\"start\":2.72,\"end\":3.18}]"}
{"utterance_id": "652-129742-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0002.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.26, 2.64, "leaves"]], "response_mode": "offline", "answer": "[{\"word\":\"leaves\",\"start\":2.26,\"end\":2.64}]"}
{"utterance_id": "6295-244435-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0030.flac", "task_type": "lexical_detection", "query": "sides", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sides\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.08, 4.56, "sides"]], "response_mode": "offline", "answer": "[{\"word\":\"sides\",\"start\":4.08,\"end\":4.56}]"}
{"utterance_id": "5338-24615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.49, 10.8, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":10.49,\"end\":10.8}]"}
{"utterance_id": "6313-76958-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0009.flac", "task_type": "lexical_detection", "query": "cowboy", "prompt": "After listening to the whole recording, output every instance of the word \"cowboy\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.81, "cowboy"]], "response_mode": "offline", "answer": "[{\"word\":\"cowboy\",\"start\":0.39,\"end\":0.81}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.47, 10.8, "letty's"], [13.77, 14.69, "david's"]], "response_mode": "offline", "answer": "[{\"word\":\"letty's\",\"start\":10.47,\"end\":10.8},{\"word\":\"david's\",\"start\":13.77,\"end\":14.69}]"}
{"utterance_id": "1462-170138-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0000.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[8.45, 9.09, "one"], [12.02, 13.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.45,\"end\":9.09},{\"word\":\"two\",\"start\":12.02,\"end\":13.06}]"}
{"utterance_id": "8297-275154-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0026.flac", "task_type": "lexical_detection", "query": "pleaded", "prompt": "When the utterance is complete, report all instances of the word \"pleaded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.53, "pleaded"]], "response_mode": "offline", "answer": "[{\"word\":\"pleaded\",\"start\":1.97,\"end\":2.53}]"}
{"utterance_id": "6295-244435-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0000.flac", "task_type": "lexical_detection", "query": "seven", "prompt": "After listening to the whole recording, output every instance of the word \"seven\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.98, "seven"]], "response_mode": "offline", "answer": "[{\"word\":\"seven\",\"start\":0.54,\"end\":0.98}]"}
{"utterance_id": "1462-170142-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0022.flac", "task_type": "lexical_detection", "query": "heavily", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"heavily\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.82, "heavily"]], "response_mode": "offline", "answer": "[{\"word\":\"heavily\",\"start\":1.37,\"end\":1.82}]"}
{"utterance_id": "5338-284437-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0003.flac", "task_type": "lexical_detection", "query": "trot", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"trot\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.02, 1.4, "trot"]], "response_mode": "offline", "answer": "[{\"word\":\"trot\",\"start\":1.02,\"end\":1.4}]"}
{"utterance_id": "2412-153947-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0012.flac", "task_type": "lexical_detection", "query": "cause", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cause\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.89, "cause"]], "response_mode": "offline", "answer": "[{\"word\":\"cause\",\"start\":1.4,\"end\":1.89}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.01, 8.35, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":8.01,\"end\":8.35}]"}
{"utterance_id": "2086-149220-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0022.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[1.82, 2.33, "artist"]], "response_mode": "offline", "answer": "[{\"word\":\"artist\",\"start\":1.82,\"end\":2.33}]"}
{"utterance_id": "1988-147956-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0007.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[0.72, 1.31, "grandmother"]], "response_mode": "offline", "answer": "[{\"word\":\"grandmother\",\"start\":0.72,\"end\":1.31}]"}
{"utterance_id": "3000-15664-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0039.flac", "task_type": "lexical_detection", "query": "tracing", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"tracing\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.97, "tracing"]], "response_mode": "offline", "answer": "[{\"word\":\"tracing\",\"start\":0.48,\"end\":0.97}]"}
{"utterance_id": "3576-138058-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.89, 3.37, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":2.89,\"end\":3.37}]"}
{"utterance_id": "1462-170138-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 1.96, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":1.6,\"end\":1.96}]"}
{"utterance_id": "5694-64025-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0006.flac", "task_type": "lexical_detection", "query": "supporting", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"supporting\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.41, "supporting"]], "response_mode": "offline", "answer": "[{\"word\":\"supporting\",\"start\":0.77,\"end\":1.41}]"}
{"utterance_id": "3081-166546-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0037.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.09, "eyes"], [4.32, 4.74, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.55,\"end\":2.09},{\"word\":\"foot\",\"start\":4.32,\"end\":4.74}]"}
{"utterance_id": "5694-64029-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0017.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Process the full audio offline. Find every instance of an organization name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 0.93, "yankees"]], "response_mode": "offline", "answer": "[{\"word\":\"yankees\",\"start\":0.62,\"end\":0.93}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.29, 6.98, "volunteers"]], "response_mode": "offline", "answer": "[{\"word\":\"volunteers\",\"start\":6.29,\"end\":6.98}]"}
{"utterance_id": "777-126732-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0029.flac", "task_type": "lexical_detection", "query": "slight", "prompt": "Process the full audio offline. Find every instance of the word \"slight\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.27, "slight"]], "response_mode": "offline", "answer": "[{\"word\":\"slight\",\"start\":1.92,\"end\":2.27}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.57, 9.91, "food"], [14.29, 14.68, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":9.57,\"end\":9.91},{\"word\":\"food\",\"start\":14.29,\"end\":14.68}]"}
{"utterance_id": "3752-4944-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0054.flac", "task_type": "lexical_detection", "query": "who's", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"who's\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.01, 1.24, "who's"]], "response_mode": "offline", "answer": "[{\"word\":\"who's\",\"start\":1.01,\"end\":1.24}]"}
{"utterance_id": "777-126732-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0013.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.48, 7.72, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":7.48,\"end\":7.72}]"}
{"utterance_id": "422-122949-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0020.flac", "task_type": "lexical_detection", "query": "similar", "prompt": "When the utterance is complete, report all instances of the word \"similar\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.49, 11.96, "similar"]], "response_mode": "offline", "answer": "[{\"word\":\"similar\",\"start\":11.49,\"end\":11.96}]"}
{"utterance_id": "3853-163249-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0018.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.37, "one"], [7.34, 7.62, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.86,\"end\":1.37},{\"word\":\"two\",\"start\":7.34,\"end\":7.62}]"}
{"utterance_id": "8842-302201-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0009.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.74, 2.23, "angel"]], "response_mode": "offline", "answer": "[{\"word\":\"angel\",\"start\":1.74,\"end\":2.23}]"}
{"utterance_id": "3752-4944-0063", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0063.flac", "task_type": "lexical_detection", "query": "relieved", "prompt": "After the audio ends, extract all occurrences of the word \"relieved\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.94, "relieved"]], "response_mode": "offline", "answer": "[{\"word\":\"relieved\",\"start\":0.55,\"end\":0.94}]"}
{"utterance_id": "1919-142785-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0051.flac", "task_type": "lexical_detection", "query": "udder", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"udder\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.9, 1.2, "udder"]], "response_mode": "offline", "answer": "[{\"word\":\"udder\",\"start\":0.9,\"end\":1.2}]"}
{"utterance_id": "3000-15664-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0017.flac", "task_type": "lexical_counting", "query": "feet", "prompt": "Process the full audio offline. For every occurrence of the word \"feet\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.43, 8.83, "one"], [13.17, 13.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.43,\"end\":8.83},{\"word\":\"two\",\"start\":13.17,\"end\":13.43}]"}
{"utterance_id": "5895-34622-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0002.flac", "task_type": "lexical_detection", "query": "nothing", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"nothing\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.91, 2.44, "nothing"]], "response_mode": "offline", "answer": "[{\"word\":\"nothing\",\"start\":1.91,\"end\":2.44}]"}
{"utterance_id": "2803-154328-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0008.flac", "task_type": "lexical_detection", "query": "fulfill", "prompt": "Process the full audio offline. Find every instance of the word \"fulfill\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.44, 6.0, "fulfill"]], "response_mode": "offline", "answer": "[{\"word\":\"fulfill\",\"start\":5.44,\"end\":6.0}]"}
{"utterance_id": "777-126732-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0064.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen carefully to the entire clip, then report every instance of a time of day as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.55, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":2.33,\"end\":2.55}]"}
{"utterance_id": "1462-170138-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0010.flac", "task_type": "lexical_detection", "query": "showing", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"showing\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.2, 2.61, "showing"]], "response_mode": "offline", "answer": "[{\"word\":\"showing\",\"start\":2.2,\"end\":2.61}]"}
{"utterance_id": "3576-138058-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0019.flac", "task_type": "lexical_detection", "query": "knight", "prompt": "After listening to the whole recording, output every instance of the word \"knight\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.36, 4.76, "knight"]], "response_mode": "offline", "answer": "[{\"word\":\"knight\",\"start\":4.36,\"end\":4.76}]"}
{"utterance_id": "6313-66129-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0028.flac", "task_type": "lexical_detection", "query": "tad", "prompt": "After the audio ends, extract all occurrences of the word \"tad\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.2, 0.55, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":0.2,\"end\":0.55}]"}
{"utterance_id": "2412-153948-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0006.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Listen to the entire clip, then report the count sequence for an animal name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.44, "one"], [14.12, 14.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.12,\"end\":1.44},{\"word\":\"two\",\"start\":14.12,\"end\":14.65}]"}
{"utterance_id": "2428-83705-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0027.flac", "task_type": "lexical_detection", "query": "prodigious", "prompt": "Once the audio has finished, identify every instance of the word \"prodigious\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.18, 2.07, "prodigious"]], "response_mode": "offline", "answer": "[{\"word\":\"prodigious\",\"start\":1.18,\"end\":2.07}]"}
{"utterance_id": "6241-66616-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0017.flac", "task_type": "lexical_detection", "query": "adventures", "prompt": "After the audio ends, extract all occurrences of the word \"adventures\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.01, "adventures"]], "response_mode": "offline", "answer": "[{\"word\":\"adventures\",\"start\":2.54,\"end\":3.01}]"}
{"utterance_id": "6313-76958-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 1.77, "lumpy"]], "response_mode": "offline", "answer": "[{\"word\":\"lumpy\",\"start\":1.38,\"end\":1.77}]"}
{"utterance_id": "1272-128104-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.42, 1.61, "harry quilter m"]], "response_mode": "offline", "answer": "[{\"word\":\"harry quilter m\",\"start\":0.42,\"end\":1.61}]"}
{"utterance_id": "2277-149897-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0035.flac", "task_type": "lexical_detection", "query": "yours", "prompt": "When the utterance is complete, report all instances of the word \"yours\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.28, "yours"]], "response_mode": "offline", "answer": "[{\"word\":\"yours\",\"start\":0.88,\"end\":1.28}]"}
{"utterance_id": "1272-128104-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.5, 1.27, "mister quilter"]], "response_mode": "offline", "answer": "[{\"word\":\"mister quilter\",\"start\":0.5,\"end\":1.27}]"}
{"utterance_id": "3752-4943-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.26, 4.2, "rufus dawes"]], "response_mode": "offline", "answer": "[{\"word\":\"rufus dawes\",\"start\":3.26,\"end\":4.2}]"}
{"utterance_id": "1919-142785-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0047.flac", "task_type": "lexical_counting", "query": "sweet", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"sweet\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.73, 1.03, "one"], [13.99, 14.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.73,\"end\":1.03},{\"word\":\"two\",\"start\":13.99,\"end\":14.37}]"}
{"utterance_id": "6295-244435-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0016.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Once the audio has finished, identify every instance of a state or province name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.61, 1.1, "kentucky"]], "response_mode": "offline", "answer": "[{\"word\":\"kentucky\",\"start\":0.61,\"end\":1.1}]"}
{"utterance_id": "2428-83699-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0016.flac", "task_type": "lexical_detection", "query": "warm", "prompt": "Once the audio has finished, identify every instance of the word \"warm\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.06, 4.47, "warm"]], "response_mode": "offline", "answer": "[{\"word\":\"warm\",\"start\":4.06,\"end\":4.47}]"}
{"utterance_id": "5694-64029-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0025.flac", "task_type": "lexical_detection", "query": "torn", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"torn\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.04, 1.37, "torn"]], "response_mode": "offline", "answer": "[{\"word\":\"torn\",\"start\":1.04,\"end\":1.37}]"}
{"utterance_id": "5895-34615-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0000.flac", "task_type": "lexical_detection", "query": "joy", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"joy\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.85, "joy"]], "response_mode": "offline", "answer": "[{\"word\":\"joy\",\"start\":2.35,\"end\":2.85}]"}
{"utterance_id": "422-122949-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0006.flac", "task_type": "lexical_counting", "query": "evil", "prompt": "Process the full audio offline. For every occurrence of the word \"evil\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.06, 4.44, "one"], [8.06, 8.56, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.06,\"end\":4.44},{\"word\":\"two\",\"start\":8.06,\"end\":8.56}]"}
{"utterance_id": "1462-170138-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0023.flac", "task_type": "lexical_detection", "query": "smiled", "prompt": "After the audio ends, extract all occurrences of the word \"smiled\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.01, 4.5, "smiled"]], "response_mode": "offline", "answer": "[{\"word\":\"smiled\",\"start\":4.01,\"end\":4.5}]"}
{"utterance_id": "7850-111771-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0008.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.28, 5.26, "wife son"]], "response_mode": "offline", "answer": "[{\"word\":\"wife son\",\"start\":4.28,\"end\":5.26}]"}
{"utterance_id": "652-129742-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0018.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.21, "dressing"]], "response_mode": "offline", "answer": "[{\"word\":\"dressing\",\"start\":0.73,\"end\":1.21}]"}
{"utterance_id": "5338-284437-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 2.3, "button bright"]], "response_mode": "offline", "answer": "[{\"word\":\"button bright\",\"start\":1.65,\"end\":2.3}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Process the full audio offline. For every occurrence of a profession, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.25, 7.65, "one"], [10.61, 11.07, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.25,\"end\":7.65},{\"word\":\"two\",\"start\":10.61,\"end\":11.07}]"}
{"utterance_id": "2412-153954-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0007.flac", "task_type": "lexical_detection", "query": "grey", "prompt": "After listening to the whole recording, output every instance of the word \"grey\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.44, "grey"]], "response_mode": "offline", "answer": "[{\"word\":\"grey\",\"start\":3.26,\"end\":3.44}]"}
{"utterance_id": "84-121123-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0000.flac", "task_type": "lexical_detection", "query": "hear", "prompt": "After the audio ends, extract all occurrences of the word \"hear\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 1.89, "hear"]], "response_mode": "offline", "answer": "[{\"word\":\"hear\",\"start\":1.49,\"end\":1.89}]"}
{"utterance_id": "1993-147149-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.17, "jem"]], "response_mode": "offline", "answer": "[{\"word\":\"jem\",\"start\":0.84,\"end\":1.17}]"}
{"utterance_id": "3853-163249-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.89, 3.35, "letty"]], "response_mode": "offline", "answer": "[{\"word\":\"letty\",\"start\":2.89,\"end\":3.35}]"}
{"utterance_id": "2277-149874-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0000.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a number with word/start/end. If none, reply [].", "labels": [[1.52, 1.76, "one"], [13.81, 14.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.52,\"end\":1.76},{\"word\":\"two\",\"start\":13.81,\"end\":14.53}]"}
{"utterance_id": "7850-111771-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.67, 3.11, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":2.67,\"end\":3.11}]"}
{"utterance_id": "2412-153948-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0008.flac", "task_type": "lexical_detection", "query": "surprising", "prompt": "When the utterance is complete, report all instances of the word \"surprising\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.24, "surprising"]], "response_mode": "offline", "answer": "[{\"word\":\"surprising\",\"start\":0.71,\"end\":1.24}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[26.78, 27.36, "vandals"]], "response_mode": "offline", "answer": "[{\"word\":\"vandals\",\"start\":26.78,\"end\":27.36}]"}
{"utterance_id": "1462-170138-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.21, 0.55, "hugh's"]], "response_mode": "offline", "answer": "[{\"word\":\"hugh's\",\"start\":0.21,\"end\":0.55}]"}
{"utterance_id": "2086-149220-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0032.flac", "task_type": "lexical_detection", "query": "smile", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"smile\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.71, "smile"]], "response_mode": "offline", "answer": "[{\"word\":\"smile\",\"start\":2.3,\"end\":2.71}]"}
{"utterance_id": "5694-64025-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0012.flac", "task_type": "lexical_detection", "query": "officers", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"officers\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.15, 0.7, "officers"]], "response_mode": "offline", "answer": "[{\"word\":\"officers\",\"start\":0.15,\"end\":0.7}]"}
{"utterance_id": "422-122949-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0004.flac", "task_type": "lexical_detection", "query": "equals", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"equals\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 4.98, "equals"]], "response_mode": "offline", "answer": "[{\"word\":\"equals\",\"start\":4.44,\"end\":4.98}]"}
{"utterance_id": "652-130726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "You will hear a complete utterance. When it ends, list every instance of a date you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.42, 6.26, "nineteen o six"]], "response_mode": "offline", "answer": "[{\"word\":\"nineteen o six\",\"start\":5.42,\"end\":6.26}]"}
{"utterance_id": "2086-149220-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0010.flac", "task_type": "lexical_detection", "query": "unquestionably", "prompt": "Process the full audio offline. Find every instance of the word \"unquestionably\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.75, "unquestionably"]], "response_mode": "offline", "answer": "[{\"word\":\"unquestionably\",\"start\":1.84,\"end\":2.75}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.28, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.93,\"end\":1.28}]"}
{"utterance_id": "6241-61943-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.45, 2.79, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":2.45,\"end\":2.79}]"}
{"utterance_id": "7976-105575-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0002.flac", "task_type": "lexical_detection", "query": "brigade", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"brigade\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.21, "brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"brigade\",\"start\":0.75,\"end\":1.21}]"}
{"utterance_id": "7850-111771-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "When the utterance is complete, report all instances of a date including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.02, 7.21, "eighteen thirty nine"]], "response_mode": "offline", "answer": "[{\"word\":\"eighteen thirty nine\",\"start\":6.02,\"end\":7.21}]"}
{"utterance_id": "7850-281318-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0014.flac", "task_type": "lexical_detection", "query": "careful", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"careful\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 3.15, "careful"]], "response_mode": "offline", "answer": "[{\"word\":\"careful\",\"start\":2.52,\"end\":3.15}]"}
{"utterance_id": "3081-166546-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0005.flac", "task_type": "lexical_detection", "query": "george", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"george\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.51, 0.9, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":0.51,\"end\":0.9}]"}
{"utterance_id": "652-130726-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0008.flac", "task_type": "lexical_detection", "query": "cost", "prompt": "After listening to the whole recording, output every instance of the word \"cost\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.55, "cost"]], "response_mode": "offline", "answer": "[{\"word\":\"cost\",\"start\":3.11,\"end\":3.55}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.74, "fairies"], [13.97, 14.36, "fairy"]], "response_mode": "offline", "answer": "[{\"word\":\"fairies\",\"start\":1.18,\"end\":1.74},{\"word\":\"fairy\",\"start\":13.97,\"end\":14.36}]"}
{"utterance_id": "2035-152373-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0010.flac", "task_type": "lexical_detection", "query": "kingdom", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"kingdom\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 1.03, "kingdom"]], "response_mode": "offline", "answer": "[{\"word\":\"kingdom\",\"start\":0.64,\"end\":1.03}]"}
{"utterance_id": "7850-111771-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0007.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.99, 6.99, "confederate army"]], "response_mode": "offline", "answer": "[{\"word\":\"confederate army\",\"start\":5.99,\"end\":6.99}]"}
{"utterance_id": "8297-275156-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.22, 9.72, "editor"]], "response_mode": "offline", "answer": "[{\"word\":\"editor\",\"start\":9.22,\"end\":9.72}]"}
{"utterance_id": "652-130737-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0010.flac", "task_type": "lexical_detection", "query": "bordeaux", "prompt": "Once the audio has finished, identify every instance of the word \"bordeaux\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.18, 2.76, "bordeaux"]], "response_mode": "offline", "answer": "[{\"word\":\"bordeaux\",\"start\":2.18,\"end\":2.76}]"}
{"utterance_id": "6345-93302-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0018.flac", "task_type": "lexical_counting", "query": "don't", "prompt": "Process the full audio offline. For every occurrence of the word \"don't\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.64, "one"], [2.65, 2.87, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.43,\"end\":0.64},{\"word\":\"two\",\"start\":2.65,\"end\":2.87}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "semantic_counting", "query": "city", "prompt": "Listen to the full audio. Count every occurrence of a city name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.29, "one"], [4.32, 4.73, "two"], [7.81, 8.59, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.86,\"end\":1.29},{\"word\":\"two\",\"start\":4.32,\"end\":4.73},{\"word\":\"three\",\"start\":7.81,\"end\":8.59}]"}
{"utterance_id": "2428-83705-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0011.flac", "task_type": "lexical_detection", "query": "possessing", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"possessing\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.15, 3.75, "possessing"]], "response_mode": "offline", "answer": "[{\"word\":\"possessing\",\"start\":3.15,\"end\":3.75}]"}
{"utterance_id": "2035-147960-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0008.flac", "task_type": "lexical_detection", "query": "liquid", "prompt": "Process the full audio offline. Find every instance of the word \"liquid\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.22, "liquid"]], "response_mode": "offline", "answer": "[{\"word\":\"liquid\",\"start\":3.78,\"end\":4.22}]"}
{"utterance_id": "7850-111771-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0003.flac", "task_type": "lexical_detection", "query": "speaking", "prompt": "When the utterance is complete, report all instances of the word \"speaking\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.1, "speaking"]], "response_mode": "offline", "answer": "[{\"word\":\"speaking\",\"start\":1.67,\"end\":2.1}]"}
{"utterance_id": "1272-128104-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 1.24, "mister quilter"]], "response_mode": "offline", "answer": "[{\"word\":\"mister quilter\",\"start\":0.48,\"end\":1.24}]"}
{"utterance_id": "6345-64257-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0007.flac", "task_type": "lexical_detection", "query": "nobody", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"nobody\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.87, 1.51, "nobody"]], "response_mode": "offline", "answer": "[{\"word\":\"nobody\",\"start\":0.87,\"end\":1.51}]"}
{"utterance_id": "6295-244435-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0039.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.19, 4.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":4.19,\"end\":4.66}]"}
{"utterance_id": "2277-149897-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0026.flac", "task_type": "lexical_detection", "query": "loved", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"loved\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.58, 0.78, "loved"]], "response_mode": "offline", "answer": "[{\"word\":\"loved\",\"start\":0.58,\"end\":0.78}]"}
{"utterance_id": "5895-34615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0005.flac", "task_type": "lexical_detection", "query": "mountebank", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"mountebank\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.46, 2.23, "mountebank"]], "response_mode": "offline", "answer": "[{\"word\":\"mountebank\",\"start\":1.46,\"end\":2.23}]"}
{"utterance_id": "6241-61943-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0005.flac", "task_type": "lexical_detection", "query": "heroic", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"heroic\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.3, 3.73, "heroic"]], "response_mode": "offline", "answer": "[{\"word\":\"heroic\",\"start\":3.3,\"end\":3.73}]"}
{"utterance_id": "3752-4943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0024.flac", "task_type": "lexical_detection", "query": "stones", "prompt": "Once the audio has finished, identify every instance of the word \"stones\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.9, 5.54, "stones"]], "response_mode": "offline", "answer": "[{\"word\":\"stones\",\"start\":4.9,\"end\":5.54}]"}
{"utterance_id": "2803-154320-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0013.flac", "task_type": "lexical_detection", "query": "score", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"score\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.09, "score"]], "response_mode": "offline", "answer": "[{\"word\":\"score\",\"start\":1.62,\"end\":2.09}]"}
{"utterance_id": "777-126732-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0007.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a disease name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.91, 5.91, "gouty swellings"]], "response_mode": "offline", "answer": "[{\"word\":\"gouty swellings\",\"start\":4.91,\"end\":5.91}]"}
{"utterance_id": "422-122949-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0000.flac", "task_type": "lexical_detection", "query": "scale", "prompt": "After listening to the whole recording, output every instance of the word \"scale\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.31, 11.89, "scale"]], "response_mode": "offline", "answer": "[{\"word\":\"scale\",\"start\":11.31,\"end\":11.89}]"}
{"utterance_id": "3853-163249-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0055.flac", "task_type": "lexical_detection", "query": "david", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"david\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.5, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.96,\"end\":1.5}]"}
{"utterance_id": "3853-163249-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0013.flac", "task_type": "lexical_detection", "query": "cottage", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"cottage\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.75, 5.25, "cottage"]], "response_mode": "offline", "answer": "[{\"word\":\"cottage\",\"start\":4.75,\"end\":5.25}]"}
{"utterance_id": "5895-34615-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.4, 4.84, "features"], [6.88, 7.22, "flesh"], [11.51, 11.77, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"features\",\"start\":4.4,\"end\":4.84},{\"word\":\"flesh\",\"start\":6.88,\"end\":7.22},{\"word\":\"head\",\"start\":11.51,\"end\":11.77}]"}
{"utterance_id": "1993-147965-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0001.flac", "task_type": "lexical_detection", "query": "prayers", "prompt": "Process the full audio offline. Find every instance of the word \"prayers\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.24, "prayers"]], "response_mode": "offline", "answer": "[{\"word\":\"prayers\",\"start\":0.83,\"end\":1.24}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "lexical_detection", "query": "chef", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"chef\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.54, "chef"], [4.28, 4.59, "chef"]], "response_mode": "offline", "answer": "[{\"word\":\"chef\",\"start\":1.25,\"end\":1.54},{\"word\":\"chef\",\"start\":4.28,\"end\":4.59}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the full audio. Count every occurrence of a food name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.83, "one"], [6.32, 6.78, "two"], [14.72, 15.23, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.45,\"end\":4.83},{\"word\":\"two\",\"start\":6.32,\"end\":6.78},{\"word\":\"three\",\"start\":14.72,\"end\":15.23}]"}
{"utterance_id": "6319-64726-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0019.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.93, 6.4, "supper"]], "response_mode": "offline", "answer": "[{\"word\":\"supper\",\"start\":5.93,\"end\":6.4}]"}
{"utterance_id": "8842-304647-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0004.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.19, 0.48, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":0.19,\"end\":0.48}]"}
{"utterance_id": "3000-15664-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0001.flac", "task_type": "lexical_counting", "query": "around", "prompt": "After listening to the whole recording, output timed count words for the word \"around\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[13.03, 13.51, "one"], [18.9, 19.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":13.03,\"end\":13.51},{\"word\":\"two\",\"start\":18.9,\"end\":19.36}]"}
{"utterance_id": "251-136532-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0008.flac", "task_type": "lexical_detection", "query": "gloria", "prompt": "Process the full audio offline. Find every instance of the word \"gloria\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.89, "gloria"]], "response_mode": "offline", "answer": "[{\"word\":\"gloria\",\"start\":0.41,\"end\":0.89}]"}
{"utterance_id": "2803-154320-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 1.41, "will halley"]], "response_mode": "offline", "answer": "[{\"word\":\"will halley\",\"start\":0.39,\"end\":1.41}]"}
{"utterance_id": "6345-93302-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.39, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.26,\"end\":3.39}]"}
{"utterance_id": "2902-9006-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.14, 4.64, "church"], [19.29, 19.76, "church's"]], "response_mode": "offline", "answer": "[{\"word\":\"church\",\"start\":4.14,\"end\":4.64},{\"word\":\"church's\",\"start\":19.29,\"end\":19.76}]"}
{"utterance_id": "6241-61943-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.06, 3.59, "horace"]], "response_mode": "offline", "answer": "[{\"word\":\"horace\",\"start\":3.06,\"end\":3.59}]"}
{"utterance_id": "1993-147966-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0005.flac", "task_type": "lexical_detection", "query": "calling", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"calling\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.14, "calling"]], "response_mode": "offline", "answer": "[{\"word\":\"calling\",\"start\":2.72,\"end\":3.14}]"}
{"utterance_id": "422-122949-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[9.21, 9.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":9.21,\"end\":9.46}]"}
{"utterance_id": "251-136532-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.52, 5.53, "anthony lattimer ph d"]], "response_mode": "offline", "answer": "[{\"word\":\"anthony lattimer ph d\",\"start\":3.52,\"end\":5.53}]"}
{"utterance_id": "2277-149897-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0017.flac", "task_type": "lexical_detection", "query": "rapidly", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"rapidly\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.49, "rapidly"]], "response_mode": "offline", "answer": "[{\"word\":\"rapidly\",\"start\":4.04,\"end\":4.49}]"}
{"utterance_id": "3853-163249-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0042.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.77, 3.12, "wife"], [10.15, 10.58, "widow"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":2.77,\"end\":3.12},{\"word\":\"widow\",\"start\":10.15,\"end\":10.58}]"}
{"utterance_id": "8297-275155-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.98, "catherine"]], "response_mode": "offline", "answer": "[{\"word\":\"catherine\",\"start\":1.47,\"end\":1.98}]"}
{"utterance_id": "7850-286674-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0012.flac", "task_type": "lexical_detection", "query": "scared", "prompt": "Once the audio has finished, identify every instance of the word \"scared\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.2, 1.03, "scared"]], "response_mode": "offline", "answer": "[{\"word\":\"scared\",\"start\":0.2,\"end\":1.03}]"}
{"utterance_id": "5694-64029-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0012.flac", "task_type": "lexical_counting", "query": "regiment", "prompt": "Process the full audio offline. For every occurrence of the word \"regiment\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.98, 4.35, "one"], [6.56, 7.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.98,\"end\":4.35},{\"word\":\"two\",\"start\":6.56,\"end\":7.01}]"}
{"utterance_id": "6313-66129-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0005.flac", "task_type": "lexical_detection", "query": "jam", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"jam\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.91, "jam"]], "response_mode": "offline", "answer": "[{\"word\":\"jam\",\"start\":0.43,\"end\":0.91}]"}
{"utterance_id": "6313-66129-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0033.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.55, 2.86, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":2.55,\"end\":2.86}]"}
{"utterance_id": "1272-135031-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0006.flac", "task_type": "lexical_detection", "query": "offered", "prompt": "Once the audio has finished, identify every instance of the word \"offered\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.89, 1.23, "offered"]], "response_mode": "offline", "answer": "[{\"word\":\"offered\",\"start\":0.89,\"end\":1.23}]"}
{"utterance_id": "5338-24615-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0009.flac", "task_type": "lexical_detection", "query": "security", "prompt": "After listening to the whole recording, output every instance of the word \"security\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.58, 3.19, "security"]], "response_mode": "offline", "answer": "[{\"word\":\"security\",\"start\":2.58,\"end\":3.19}]"}
{"utterance_id": "8297-275155-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.0, 4.59, "sydenham"]], "response_mode": "offline", "answer": "[{\"word\":\"sydenham\",\"start\":4.0,\"end\":4.59}]"}
{"utterance_id": "3853-163249-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0033.flac", "task_type": "lexical_detection", "query": "ain't", "prompt": "Process the full audio offline. Find every instance of the word \"ain't\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.99, "ain't"]], "response_mode": "offline", "answer": "[{\"word\":\"ain't\",\"start\":0.57,\"end\":0.99}]"}
{"utterance_id": "3853-163249-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0016.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.05, 4.45, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":4.05,\"end\":4.45}]"}
{"utterance_id": "8842-302196-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0001.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Once the audio has finished, list each occurrence of a profession as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[5.74, 6.43, "one"], [7.87, 8.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.74,\"end\":6.43},{\"word\":\"two\",\"start\":7.87,\"end\":8.3}]"}
{"utterance_id": "1673-143397-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0020.flac", "task_type": "lexical_detection", "query": "dismissed", "prompt": "After listening to the whole recording, output every instance of the word \"dismissed\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.05, 6.5, "dismissed"]], "response_mode": "offline", "answer": "[{\"word\":\"dismissed\",\"start\":6.05,\"end\":6.5}]"}
{"utterance_id": "5536-43363-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.18, "fourteen"]], "response_mode": "offline", "answer": "[{\"word\":\"fourteen\",\"start\":1.76,\"end\":2.18}]"}
{"utterance_id": "7850-286674-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0003.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the full audio. Count every occurrence of a food name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 1.99, "one"], [5.35, 5.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.55,\"end\":1.99},{\"word\":\"two\",\"start\":5.35,\"end\":5.91}]"}
{"utterance_id": "2412-153948-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0014.flac", "task_type": "lexical_detection", "query": "smallest", "prompt": "Process the full audio offline. Find every instance of the word \"smallest\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.95, 2.44, "smallest"]], "response_mode": "offline", "answer": "[{\"word\":\"smallest\",\"start\":1.95,\"end\":2.44}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "semantic_counting", "query": "disease", "prompt": "Listen to the full audio. Count every occurrence of a disease name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.27, "one"], [4.37, 4.81, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.86,\"end\":1.27},{\"word\":\"two\",\"start\":4.37,\"end\":4.81}]"}
{"utterance_id": "251-118436-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0000.flac", "task_type": "lexical_detection", "query": "poison", "prompt": "After the audio ends, extract all occurrences of the word \"poison\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.12, 4.74, "poison"]], "response_mode": "offline", "answer": "[{\"word\":\"poison\",\"start\":4.12,\"end\":4.74}]"}
{"utterance_id": "2428-83705-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0042.flac", "task_type": "lexical_detection", "query": "coupons", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"coupons\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.32, 7.0, "coupons"]], "response_mode": "offline", "answer": "[{\"word\":\"coupons\",\"start\":6.32,\"end\":7.0}]"}
{"utterance_id": "84-121550-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0014.flac", "task_type": "lexical_detection", "query": "ten", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"ten\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.21, 5.68, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":5.21,\"end\":5.68}]"}
{"utterance_id": "1272-141231-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.35, 5.71, "two"], [7.47, 7.85, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":5.35,\"end\":5.71},{\"word\":\"one\",\"start\":7.47,\"end\":7.85}]"}
{"utterance_id": "3000-15664-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0029.flac", "task_type": "lexical_detection", "query": "everything", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"everything\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.14, "everything"]], "response_mode": "offline", "answer": "[{\"word\":\"everything\",\"start\":2.74,\"end\":3.14}]"}
{"utterance_id": "2078-142845-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0008.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After listening to the whole recording, output timed count words for a food name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[9.8, 10.4, "one"], [15.25, 15.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.8,\"end\":10.4},{\"word\":\"two\",\"start\":15.25,\"end\":15.89}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.73, 6.06, "clubs"]], "response_mode": "offline", "answer": "[{\"word\":\"clubs\",\"start\":5.73,\"end\":6.06}]"}
{"utterance_id": "174-168635-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0022.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.68, "child's"]], "response_mode": "offline", "answer": "[{\"word\":\"child's\",\"start\":1.18,\"end\":1.68}]"}
{"utterance_id": "1272-128104-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.08, 1.71, "mister quilter"]], "response_mode": "offline", "answer": "[{\"word\":\"mister quilter\",\"start\":1.08,\"end\":1.71}]"}
{"utterance_id": "5895-34629-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.82, 1.36, "comedian"]], "response_mode": "offline", "answer": "[{\"word\":\"comedian\",\"start\":0.82,\"end\":1.36}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "lexical_detection", "query": "imitate", "prompt": "After the audio ends, extract all occurrences of the word \"imitate\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.83, 10.25, "imitate"]], "response_mode": "offline", "answer": "[{\"word\":\"imitate\",\"start\":9.83,\"end\":10.25}]"}
{"utterance_id": "2086-149220-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0018.flac", "task_type": "lexical_detection", "query": "pyncheon", "prompt": "Process the full audio offline. Find every instance of the word \"pyncheon\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.39, "pyncheon"]], "response_mode": "offline", "answer": "[{\"word\":\"pyncheon\",\"start\":5.9,\"end\":6.39}]"}
{"utterance_id": "1272-135031-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.42, 7.03, "kaliko"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":6.42,\"end\":7.03}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.73, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.57,\"end\":0.73}]"}
{"utterance_id": "1673-143397-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 1.73, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.45,\"end\":1.73}]"}
{"utterance_id": "5536-43358-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0011.flac", "task_type": "lexical_detection", "query": "evils", "prompt": "When the utterance is complete, report all instances of the word \"evils\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.75, 5.26, "evils"]], "response_mode": "offline", "answer": "[{\"word\":\"evils\",\"start\":4.75,\"end\":5.26}]"}
{"utterance_id": "2902-9008-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0009.flac", "task_type": "lexical_detection", "query": "sphere", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sphere\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.72, "sphere"]], "response_mode": "offline", "answer": "[{\"word\":\"sphere\",\"start\":1.36,\"end\":1.72}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "lexical_detection", "query": "termed", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"termed\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.47, 3.84, "termed"]], "response_mode": "offline", "answer": "[{\"word\":\"termed\",\"start\":3.47,\"end\":3.84}]"}
{"utterance_id": "7976-110124-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0018.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 1.05, "merchant's"]], "response_mode": "offline", "answer": "[{\"word\":\"merchant's\",\"start\":0.58,\"end\":1.05}]"}
{"utterance_id": "3536-23268-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.96, 10.66, "miss woodley"]], "response_mode": "offline", "answer": "[{\"word\":\"miss woodley\",\"start\":9.96,\"end\":10.66}]"}
{"utterance_id": "2035-147961-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0016.flac", "task_type": "lexical_detection", "query": "lit", "prompt": "After the audio ends, extract all occurrences of the word \"lit\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.28, "lit"]], "response_mode": "offline", "answer": "[{\"word\":\"lit\",\"start\":2.02,\"end\":2.28}]"}
{"utterance_id": "5536-43358-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0004.flac", "task_type": "lexical_detection", "query": "erect", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"erect\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.29, 3.73, "erect"]], "response_mode": "offline", "answer": "[{\"word\":\"erect\",\"start\":3.29,\"end\":3.73}]"}
{"utterance_id": "2035-147961-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.81, "peter"], [3.63, 4.04, "cutter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":0.47,\"end\":0.81},{\"word\":\"cutter\",\"start\":3.63,\"end\":4.04}]"}
{"utterance_id": "3170-137482-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0034.flac", "task_type": "lexical_detection", "query": "declared", "prompt": "When the utterance is complete, report all instances of the word \"declared\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 1.01, "declared"]], "response_mode": "offline", "answer": "[{\"word\":\"declared\",\"start\":0.61,\"end\":1.01}]"}
{"utterance_id": "6345-93306-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0002.flac", "task_type": "lexical_detection", "query": "moving", "prompt": "Process the full audio offline. Find every instance of the word \"moving\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.45, 3.82, "moving"]], "response_mode": "offline", "answer": "[{\"word\":\"moving\",\"start\":3.45,\"end\":3.82}]"}
{"utterance_id": "1993-147149-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0015.flac", "task_type": "lexical_detection", "query": "wished", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"wished\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.21, 6.52, "wished"]], "response_mode": "offline", "answer": "[{\"word\":\"wished\",\"start\":6.21,\"end\":6.52}]"}
{"utterance_id": "5694-64025-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0007.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "You will hear a complete utterance. When it ends, list every instance of a state or province name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 2.09, "tennessee"]], "response_mode": "offline", "answer": "[{\"word\":\"tennessee\",\"start\":1.47,\"end\":2.09}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "After listening to the whole recording, output timed count words for a family relation (one, two, three, ...) as JSON. If none, reply [].", "labels": [[3.69, 4.12, "one"], [6.97, 7.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.69,\"end\":4.12},{\"word\":\"two\",\"start\":6.97,\"end\":7.35}]"}
{"utterance_id": "3752-4944-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0034.flac", "task_type": "lexical_detection", "query": "find", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"find\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 0.97, "find"]], "response_mode": "offline", "answer": "[{\"word\":\"find\",\"start\":0.66,\"end\":0.97}]"}
{"utterance_id": "2078-142845-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0040.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.05, 2.75, "buns"], [6.07, 6.79, "buns"]], "response_mode": "offline", "answer": "[{\"word\":\"buns\",\"start\":2.05,\"end\":2.75},{\"word\":\"buns\",\"start\":6.07,\"end\":6.79}]"}
{"utterance_id": "5694-64029-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0001.flac", "task_type": "lexical_detection", "query": "federal", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"federal\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.65, 1.03, "federal"]], "response_mode": "offline", "answer": "[{\"word\":\"federal\",\"start\":0.65,\"end\":1.03}]"}
{"utterance_id": "3576-138058-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0021.flac", "task_type": "lexical_detection", "query": "ease", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"ease\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.37, 5.84, "ease"]], "response_mode": "offline", "answer": "[{\"word\":\"ease\",\"start\":5.37,\"end\":5.84}]"}
{"utterance_id": "3000-15664-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0028.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "When the utterance is complete, report all instances of an ordinal number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.11, 5.45, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.11,\"end\":5.45}]"}
{"utterance_id": "3536-23268-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0014.flac", "task_type": "lexical_detection", "query": "madam", "prompt": "Once the audio has finished, identify every instance of the word \"madam\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.08, 1.45, "madam"]], "response_mode": "offline", "answer": "[{\"word\":\"madam\",\"start\":1.08,\"end\":1.45}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.58, "hair"], [3.73, 4.06, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":1.21,\"end\":1.58},{\"word\":\"hands\",\"start\":3.73,\"end\":4.06}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "lexical_counting", "query": "new", "prompt": "After listening to the whole recording, output timed count words for the word \"new\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.65, 2.87, "one"], [8.16, 8.33, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.65,\"end\":2.87},{\"word\":\"two\",\"start\":8.16,\"end\":8.33}]"}
{"utterance_id": "1272-135031-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.75, 4.96, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.75,\"end\":4.96}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an ordinal number with word/start/end. If none, reply [].", "labels": [[0.59, 0.83, "one"], [2.68, 3.07, "two"], [6.19, 6.57, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.59,\"end\":0.83},{\"word\":\"two\",\"start\":2.68,\"end\":3.07},{\"word\":\"three\",\"start\":6.19,\"end\":6.57}]"}
{"utterance_id": "1272-141231-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 2.24, "twenties"]], "response_mode": "offline", "answer": "[{\"word\":\"twenties\",\"start\":1.59,\"end\":2.24}]"}
{"utterance_id": "2428-83699-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0011.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "You will hear a complete utterance. When it ends, list every instance of a time of day you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.88, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":0.52,\"end\":0.88}]"}
{"utterance_id": "1988-147956-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0011.flac", "task_type": "semantic_counting", "query": "color", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a color name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.88, 1.19, "one"], [4.07, 4.44, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.88,\"end\":1.19},{\"word\":\"two\",\"start\":4.07,\"end\":4.44}]"}
{"utterance_id": "5338-24615-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0011.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "After listening to the whole recording, output timed count words for a number (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.21, 0.4, "one"], [1.59, 1.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.21,\"end\":0.4},{\"word\":\"two\",\"start\":1.59,\"end\":1.77}]"}
{"utterance_id": "3081-166546-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0000.flac", "task_type": "lexical_detection", "query": "connected", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"connected\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.33, 8.0, "connected"]], "response_mode": "offline", "answer": "[{\"word\":\"connected\",\"start\":7.33,\"end\":8.0}]"}
{"utterance_id": "6319-275224-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0011.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Process the full audio offline. For every occurrence of a plant name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.71, 5.48, "one"], [9.04, 9.5, "two"], [10.91, 12.13, "three"], [14.27, 14.89, "four"], [16.34, 17.09, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.71,\"end\":5.48},{\"word\":\"two\",\"start\":9.04,\"end\":9.5},{\"word\":\"three\",\"start\":10.91,\"end\":12.13},{\"word\":\"four\",\"start\":14.27,\"end\":14.89},{\"word\":\"five\",\"start\":16.34,\"end\":17.09}]"}
{"utterance_id": "6295-64301-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0023.flac", "task_type": "lexical_counting", "query": "thou", "prompt": "Listen to the entire clip, then report the count sequence for the word \"thou\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 0.84, "one"], [7.08, 7.25, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.68,\"end\":0.84},{\"word\":\"two\",\"start\":7.08,\"end\":7.25}]"}
{"utterance_id": "777-126732-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0030.flac", "task_type": "lexical_detection", "query": "disregarded", "prompt": "When the utterance is complete, report all instances of the word \"disregarded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.73, "disregarded"]], "response_mode": "offline", "answer": "[{\"word\":\"disregarded\",\"start\":2.05,\"end\":2.73}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "lexical_detection", "query": "hunt", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hunt\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.57, "hunt"]], "response_mode": "offline", "answer": "[{\"word\":\"hunt\",\"start\":2.14,\"end\":2.57}]"}
{"utterance_id": "5694-64038-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0022.flac", "task_type": "lexical_counting", "query": "past", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"past\" with word/start/end. If none, reply [].", "labels": [[6.04, 6.7, "one"], [14.77, 15.41, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.04,\"end\":6.7},{\"word\":\"two\",\"start\":14.77,\"end\":15.41}]"}
{"utterance_id": "5895-34615-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 0.84, "hair"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":0.6,\"end\":0.84}]"}
{"utterance_id": "6241-61946-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0005.flac", "task_type": "lexical_detection", "query": "rarely", "prompt": "Process the full audio offline. Find every instance of the word \"rarely\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.06, "rarely"]], "response_mode": "offline", "answer": "[{\"word\":\"rarely\",\"start\":0.7,\"end\":1.06}]"}
{"utterance_id": "7850-281318-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0009.flac", "task_type": "lexical_detection", "query": "twittering", "prompt": "Once the audio has finished, identify every instance of the word \"twittering\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.44, 4.08, "twittering"]], "response_mode": "offline", "answer": "[{\"word\":\"twittering\",\"start\":3.44,\"end\":4.08}]"}
{"utterance_id": "2035-147961-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.22, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":1.96,\"end\":2.22}]"}
{"utterance_id": "1462-170138-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0012.flac", "task_type": "lexical_detection", "query": "introduce", "prompt": "After listening to the whole recording, output every instance of the word \"introduce\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.24, "introduce"]], "response_mode": "offline", "answer": "[{\"word\":\"introduce\",\"start\":1.64,\"end\":2.24}]"}
{"utterance_id": "777-126732-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0035.flac", "task_type": "lexical_detection", "query": "vision", "prompt": "Once the audio has finished, identify every instance of the word \"vision\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.47, 0.75, "vision"]], "response_mode": "offline", "answer": "[{\"word\":\"vision\",\"start\":0.47,\"end\":0.75}]"}
{"utterance_id": "3081-166546-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.11, 1.67, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":1.11,\"end\":1.67}]"}
{"utterance_id": "6319-57405-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.51, 3.21, "deucalion"]], "response_mode": "offline", "answer": "[{\"word\":\"deucalion\",\"start\":2.51,\"end\":3.21}]"}
{"utterance_id": "6319-57405-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0005.flac", "task_type": "lexical_detection", "query": "flood", "prompt": "After listening to the whole recording, output every instance of the word \"flood\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.62, 6.08, "flood"]], "response_mode": "offline", "answer": "[{\"word\":\"flood\",\"start\":5.62,\"end\":6.08}]"}
{"utterance_id": "5338-24615-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0011.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.21, 0.4, "two"], [1.59, 1.77, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.21,\"end\":0.4},{\"word\":\"one\",\"start\":1.59,\"end\":1.77}]"}
{"utterance_id": "8297-275154-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0027.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 1.72, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.61,\"end\":1.72}]"}
{"utterance_id": "2277-149874-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.0, 0.38, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.0,\"end\":0.38}]"}
{"utterance_id": "2035-147961-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0019.flac", "task_type": "lexical_detection", "query": "echoed", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"echoed\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.86, 2.36, "echoed"]], "response_mode": "offline", "answer": "[{\"word\":\"echoed\",\"start\":1.86,\"end\":2.36}]"}
{"utterance_id": "1919-142785-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0045.flac", "task_type": "lexical_detection", "query": "account", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"account\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.26, 4.64, "account"]], "response_mode": "offline", "answer": "[{\"word\":\"account\",\"start\":4.26,\"end\":4.64}]"}
{"utterance_id": "2412-153948-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0004.flac", "task_type": "lexical_counting", "query": "cattle", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"cattle\" (word, start, end). If none, reply [].", "labels": [[0.67, 0.99, "one"], [17.11, 17.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.67,\"end\":0.99},{\"word\":\"two\",\"start\":17.11,\"end\":17.54}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "lexical_detection", "query": "prime", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"prime\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.01, 5.31, "prime"], [6.71, 6.96, "prime"]], "response_mode": "offline", "answer": "[{\"word\":\"prime\",\"start\":5.01,\"end\":5.31},{\"word\":\"prime\",\"start\":6.71,\"end\":6.96}]"}
{"utterance_id": "8297-275154-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.88, "randal"], [7.61, 8.13, "catherine"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":0.38,\"end\":0.88},{\"word\":\"catherine\",\"start\":7.61,\"end\":8.13}]"}
{"utterance_id": "8842-302201-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0007.flac", "task_type": "lexical_detection", "query": "weakness", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"weakness\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.24, 3.76, "weakness"]], "response_mode": "offline", "answer": "[{\"word\":\"weakness\",\"start\":3.24,\"end\":3.76}]"}
{"utterance_id": "174-50561-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.85, "singer"], [8.43, 8.8, "singer"]], "response_mode": "offline", "answer": "[{\"word\":\"singer\",\"start\":1.44,\"end\":1.85},{\"word\":\"singer\",\"start\":8.43,\"end\":8.8}]"}
{"utterance_id": "2078-142845-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0048.flac", "task_type": "lexical_detection", "query": "four", "prompt": "Process the full audio offline. Find every instance of the word \"four\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.98, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.47,\"end\":1.98}]"}
{"utterance_id": "174-50561-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0011.flac", "task_type": "lexical_detection", "query": "lute", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"lute\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.69, 9.0, "lute"]], "response_mode": "offline", "answer": "[{\"word\":\"lute\",\"start\":8.69,\"end\":9.0}]"}
{"utterance_id": "1988-24833-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0023.flac", "task_type": "lexical_detection", "query": "looks", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"looks\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.79, "looks"]], "response_mode": "offline", "answer": "[{\"word\":\"looks\",\"start\":1.62,\"end\":1.79}]"}
{"utterance_id": "2277-149897-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0033.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.45, 0.8, "wednesday"]], "response_mode": "offline", "answer": "[{\"word\":\"wednesday\",\"start\":0.45,\"end\":0.8}]"}
{"utterance_id": "3081-166546-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0030.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.77, 4.2, "one"], [7.36, 7.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.77,\"end\":4.2},{\"word\":\"two\",\"start\":7.36,\"end\":7.69}]"}
{"utterance_id": "8842-302203-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0004.flac", "task_type": "lexical_detection", "query": "shaken", "prompt": "When the utterance is complete, report all instances of the word \"shaken\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.57, 11.97, "shaken"]], "response_mode": "offline", "answer": "[{\"word\":\"shaken\",\"start\":11.57,\"end\":11.97}]"}
{"utterance_id": "2428-83705-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0007.flac", "task_type": "lexical_detection", "query": "notice", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"notice\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.56, 2.93, "notice"]], "response_mode": "offline", "answer": "[{\"word\":\"notice\",\"start\":2.56,\"end\":2.93}]"}
{"utterance_id": "7976-110523-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.21, 2.6, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":2.21,\"end\":2.6}]"}
{"utterance_id": "5536-43363-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.6, 3.99, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":3.6,\"end\":3.99}]"}
{"utterance_id": "3752-4944-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0064.flac", "task_type": "lexical_detection", "query": "crisis", "prompt": "After listening to the whole recording, output every instance of the word \"crisis\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.54, "crisis"]], "response_mode": "offline", "answer": "[{\"word\":\"crisis\",\"start\":2.01,\"end\":2.54}]"}
{"utterance_id": "3170-137482-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.02, 1.29, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.02,\"end\":1.29}]"}
{"utterance_id": "5895-34629-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0022.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Process the full audio offline. Find every instance of a city name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.47, "southwark"]], "response_mode": "offline", "answer": "[{\"word\":\"southwark\",\"start\":2.84,\"end\":3.47}]"}
{"utterance_id": "6313-66129-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.44, 0.79, "water"]], "response_mode": "offline", "answer": "[{\"word\":\"water\",\"start\":0.44,\"end\":0.79}]"}
{"utterance_id": "6241-61946-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0012.flac", "task_type": "lexical_detection", "query": "occasionally", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"occasionally\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 2.8, "occasionally"]], "response_mode": "offline", "answer": "[{\"word\":\"occasionally\",\"start\":2.13,\"end\":2.8}]"}
{"utterance_id": "2277-149874-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0002.flac", "task_type": "lexical_detection", "query": "wife's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wife's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.22, 2.6, "wife's"]], "response_mode": "offline", "answer": "[{\"word\":\"wife's\",\"start\":2.22,\"end\":2.6}]"}
{"utterance_id": "251-118436-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.49, 5.04, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":4.49,\"end\":5.04}]"}
{"utterance_id": "2803-154328-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0010.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.81, "major"]], "response_mode": "offline", "answer": "[{\"word\":\"major\",\"start\":1.42,\"end\":1.81}]"}
{"utterance_id": "6319-57405-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.54, 3.17, "mercury"]], "response_mode": "offline", "answer": "[{\"word\":\"mercury\",\"start\":2.54,\"end\":3.17}]"}
{"utterance_id": "2035-147961-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0033.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Process the full audio offline. Find every instance of a city name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.82, "village"]], "response_mode": "offline", "answer": "[{\"word\":\"village\",\"start\":1.26,\"end\":1.82}]"}
{"utterance_id": "1919-142785-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0058.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.56, 1.31, "fried bread"]], "response_mode": "offline", "answer": "[{\"word\":\"fried bread\",\"start\":0.56,\"end\":1.31}]"}
{"utterance_id": "5536-43358-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0019.flac", "task_type": "lexical_detection", "query": "oath", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"oath\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.7, 5.03, "oath"]], "response_mode": "offline", "answer": "[{\"word\":\"oath\",\"start\":4.7,\"end\":5.03}]"}
{"utterance_id": "3170-137482-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0014.flac", "task_type": "lexical_detection", "query": "fear", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fear\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.24, "fear"]], "response_mode": "offline", "answer": "[{\"word\":\"fear\",\"start\":0.76,\"end\":1.24}]"}
{"utterance_id": "3170-137482-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0006.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.87, "physicians"]], "response_mode": "offline", "answer": "[{\"word\":\"physicians\",\"start\":1.08,\"end\":1.87}]"}
{"utterance_id": "2803-154328-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0005.flac", "task_type": "lexical_detection", "query": "view", "prompt": "When the utterance is complete, report all instances of the word \"view\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.21, 5.61, "view"]], "response_mode": "offline", "answer": "[{\"word\":\"view\",\"start\":5.21,\"end\":5.61}]"}
{"utterance_id": "3081-166546-0063", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0063.flac", "task_type": "lexical_detection", "query": "doubt", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"doubt\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.53, "doubt"]], "response_mode": "offline", "answer": "[{\"word\":\"doubt\",\"start\":1.04,\"end\":1.53}]"}
{"utterance_id": "8297-275154-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0024.flac", "task_type": "lexical_detection", "query": "herbert", "prompt": "Once the audio has finished, identify every instance of the word \"herbert\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.43, 1.95, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":1.43,\"end\":1.95}]"}
{"utterance_id": "6241-66616-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0015.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.33, "forests"]], "response_mode": "offline", "answer": "[{\"word\":\"forests\",\"start\":3.7,\"end\":4.33}]"}
{"utterance_id": "2412-153954-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0009.flac", "task_type": "lexical_detection", "query": "splendid", "prompt": "When the utterance is complete, report all instances of the word \"splendid\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.53, 5.94, "splendid"]], "response_mode": "offline", "answer": "[{\"word\":\"splendid\",\"start\":5.53,\"end\":5.94}]"}
{"utterance_id": "8297-275155-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0011.flac", "task_type": "lexical_detection", "query": "newspaper", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"newspaper\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.47, "newspaper"]], "response_mode": "offline", "answer": "[{\"word\":\"newspaper\",\"start\":0.88,\"end\":1.47}]"}
{"utterance_id": "5338-24615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0006.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.74, 8.14, "boughs"]], "response_mode": "offline", "answer": "[{\"word\":\"boughs\",\"start\":7.74,\"end\":8.14}]"}
{"utterance_id": "3536-23268-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0025.flac", "task_type": "lexical_detection", "query": "troublesome", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"troublesome\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.49, 3.13, "troublesome"]], "response_mode": "offline", "answer": "[{\"word\":\"troublesome\",\"start\":2.49,\"end\":3.13}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.26, 6.63, "driver"]], "response_mode": "offline", "answer": "[{\"word\":\"driver\",\"start\":6.26,\"end\":6.63}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "lexical_detection", "query": "leaven", "prompt": "Process the full audio offline. Find every instance of the word \"leaven\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.37, "leaven"]], "response_mode": "offline", "answer": "[{\"word\":\"leaven\",\"start\":0.94,\"end\":1.37}]"}
{"utterance_id": "8842-302203-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.09, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.77,\"end\":1.09}]"}
{"utterance_id": "3576-138058-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0008.flac", "task_type": "lexical_detection", "query": "rise", "prompt": "Once the audio has finished, identify every instance of the word \"rise\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.6, 1.05, "rise"]], "response_mode": "offline", "answer": "[{\"word\":\"rise\",\"start\":0.6,\"end\":1.05}]"}
{"utterance_id": "1988-148538-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0001.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen carefully to the entire clip, then report every instance of a country name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.17, 8.54, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"europe\",\"start\":8.17,\"end\":8.54}]"}
{"utterance_id": "6319-275224-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0013.flac", "task_type": "lexical_detection", "query": "climbing", "prompt": "After listening to the whole recording, output every instance of the word \"climbing\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 2.69, "climbing"]], "response_mode": "offline", "answer": "[{\"word\":\"climbing\",\"start\":2.13,\"end\":2.69}]"}
{"utterance_id": "2086-149220-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0024.flac", "task_type": "lexical_detection", "query": "wonderful", "prompt": "After listening to the whole recording, output every instance of the word \"wonderful\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.24, "wonderful"]], "response_mode": "offline", "answer": "[{\"word\":\"wonderful\",\"start\":0.81,\"end\":1.24}]"}
{"utterance_id": "2412-153947-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.29, 7.1, "mister darwin's"]], "response_mode": "offline", "answer": "[{\"word\":\"mister darwin's\",\"start\":6.29,\"end\":7.1}]"}
{"utterance_id": "2803-161169-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0002.flac", "task_type": "lexical_detection", "query": "odd", "prompt": "After the audio ends, extract all occurrences of the word \"odd\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.74, "odd"]], "response_mode": "offline", "answer": "[{\"word\":\"odd\",\"start\":1.42,\"end\":1.74}]"}
{"utterance_id": "3000-15664-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0040.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.88, 1.23, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"blood\",\"start\":0.88,\"end\":1.23}]"}
{"utterance_id": "6241-61943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0007.flac", "task_type": "lexical_detection", "query": "day", "prompt": "When the utterance is complete, report all instances of the word \"day\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.16, "day"]], "response_mode": "offline", "answer": "[{\"word\":\"day\",\"start\":1.85,\"end\":2.16}]"}
{"utterance_id": "174-84280-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0006.flac", "task_type": "lexical_detection", "query": "waste", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"waste\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.4, "waste"]], "response_mode": "offline", "answer": "[{\"word\":\"waste\",\"start\":1.79,\"end\":2.4}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "lexical_detection", "query": "discarded", "prompt": "After listening to the whole recording, output every instance of the word \"discarded\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.41, 8.18, "discarded"]], "response_mode": "offline", "answer": "[{\"word\":\"discarded\",\"start\":7.41,\"end\":8.18}]"}
{"utterance_id": "7976-110523-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0017.flac", "task_type": "lexical_detection", "query": "oven", "prompt": "Process the full audio offline. Find every instance of the word \"oven\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.79, 9.16, "oven"]], "response_mode": "offline", "answer": "[{\"word\":\"oven\",\"start\":8.79,\"end\":9.16}]"}
{"utterance_id": "6313-76958-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0013.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[2.9, 3.15, "night"], [8.45, 8.88, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":2.9,\"end\":3.15},{\"word\":\"evening\",\"start\":8.45,\"end\":8.88}]"}
{"utterance_id": "2803-161169-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0009.flac", "task_type": "lexical_counting", "query": "coal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"coal\" with word/start/end. If none, reply [].", "labels": [[13.66, 14.15, "one"], [18.44, 18.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":13.66,\"end\":14.15},{\"word\":\"two\",\"start\":18.44,\"end\":18.8}]"}
{"utterance_id": "1993-147149-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.31, 3.73, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":3.31,\"end\":3.73}]"}
{"utterance_id": "3853-163249-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.53, 3.46, "ann lizy"]], "response_mode": "offline", "answer": "[{\"word\":\"ann lizy\",\"start\":2.53,\"end\":3.46}]"}
{"utterance_id": "3853-163249-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0003.flac", "task_type": "lexical_detection", "query": "surprise", "prompt": "After listening to the whole recording, output every instance of the word \"surprise\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.28, "surprise"]], "response_mode": "offline", "answer": "[{\"word\":\"surprise\",\"start\":0.78,\"end\":1.28}]"}
{"utterance_id": "6313-66125-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0024.flac", "task_type": "lexical_detection", "query": "glared", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"glared\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.81, "glared"]], "response_mode": "offline", "answer": "[{\"word\":\"glared\",\"start\":3.4,\"end\":3.81}]"}
{"utterance_id": "3752-4944-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0058.flac", "task_type": "lexical_detection", "query": "report", "prompt": "After listening to the whole recording, output every instance of the word \"report\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.82, "report"]], "response_mode": "offline", "answer": "[{\"word\":\"report\",\"start\":0.48,\"end\":0.82}]"}
{"utterance_id": "6345-93302-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0001.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.57, "bluebells"]], "response_mode": "offline", "answer": "[{\"word\":\"bluebells\",\"start\":1.84,\"end\":2.57}]"}
{"utterance_id": "251-118436-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0016.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.82, 3.4, "hands"], [4.91, 5.43, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.82,\"end\":3.4},{\"word\":\"eyes\",\"start\":4.91,\"end\":5.43}]"}
{"utterance_id": "7976-110523-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.23, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":1.64,\"end\":2.23}]"}
{"utterance_id": "84-121123-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0028.flac", "task_type": "lexical_detection", "query": "mine", "prompt": "After listening to the whole recording, output every instance of the word \"mine\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.21, 4.63, "mine"]], "response_mode": "offline", "answer": "[{\"word\":\"mine\",\"start\":4.21,\"end\":4.63}]"}
{"utterance_id": "3752-4944-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0013.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.02, 5.5, "bishop"]], "response_mode": "offline", "answer": "[{\"word\":\"bishop\",\"start\":5.02,\"end\":5.5}]"}
{"utterance_id": "84-121550-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0018.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "When the utterance is complete, report all instances of a weather term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.94, 7.53, "snow"]], "response_mode": "offline", "answer": "[{\"word\":\"snow\",\"start\":6.94,\"end\":7.53}]"}
{"utterance_id": "7976-110124-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0023.flac", "task_type": "lexical_detection", "query": "air", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"air\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.27, "air"]], "response_mode": "offline", "answer": "[{\"word\":\"air\",\"start\":2.9,\"end\":3.27}]"}
{"utterance_id": "5338-24615-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0013.flac", "task_type": "lexical_detection", "query": "paved", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"paved\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.49, "paved"]], "response_mode": "offline", "answer": "[{\"word\":\"paved\",\"start\":1.96,\"end\":2.49}]"}
{"utterance_id": "2412-153954-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0023.flac", "task_type": "lexical_detection", "query": "effect", "prompt": "After listening to the whole recording, output every instance of the word \"effect\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.38, "effect"]], "response_mode": "offline", "answer": "[{\"word\":\"effect\",\"start\":1.04,\"end\":1.38}]"}
{"utterance_id": "2428-83705-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.08, 1.42, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.08,\"end\":1.42}]"}
{"utterance_id": "2803-161169-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0009.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.54, "animals"]], "response_mode": "offline", "answer": "[{\"word\":\"animals\",\"start\":3.78,\"end\":4.54}]"}
{"utterance_id": "3081-166546-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0030.flac", "task_type": "lexical_detection", "query": "relaxed", "prompt": "When the utterance is complete, report all instances of the word \"relaxed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.2, 4.78, "relaxed"]], "response_mode": "offline", "answer": "[{\"word\":\"relaxed\",\"start\":4.2,\"end\":4.78}]"}
{"utterance_id": "6345-93306-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0018.flac", "task_type": "lexical_detection", "query": "catch", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"catch\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.75, "catch"]], "response_mode": "offline", "answer": "[{\"word\":\"catch\",\"start\":0.49,\"end\":0.75}]"}
{"utterance_id": "2035-147961-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.18, "peter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":0.89,\"end\":1.18}]"}
{"utterance_id": "1988-147956-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.29, 8.79, "saturday's"]], "response_mode": "offline", "answer": "[{\"word\":\"saturday's\",\"start\":8.29,\"end\":8.79}]"}
{"utterance_id": "1993-147964-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0007.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[1.86, 2.26, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":1.86,\"end\":2.26}]"}
{"utterance_id": "1673-143397-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0001.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After the audio ends, extract all occurrences of a time of day. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.92, 5.17, "dawn"]], "response_mode": "offline", "answer": "[{\"word\":\"dawn\",\"start\":4.92,\"end\":5.17}]"}
{"utterance_id": "1272-135031-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0000.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.88, 4.33, "rose"]], "response_mode": "offline", "answer": "[{\"word\":\"rose\",\"start\":3.88,\"end\":4.33}]"}
{"utterance_id": "3853-163249-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0049.flac", "task_type": "lexical_detection", "query": "brides", "prompt": "After the audio ends, extract all occurrences of the word \"brides\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.43, "brides"]], "response_mode": "offline", "answer": "[{\"word\":\"brides\",\"start\":1.06,\"end\":1.43}]"}
{"utterance_id": "2277-149896-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.51, 2.06, "drouet"]], "response_mode": "offline", "answer": "[{\"word\":\"drouet\",\"start\":1.51,\"end\":2.06}]"}
{"utterance_id": "2428-83705-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.52, 0.84, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":0.52,\"end\":0.84}]"}
{"utterance_id": "2428-83705-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0034.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.57, 2.73, "missus macpherson"]], "response_mode": "offline", "answer": "[{\"word\":\"missus macpherson\",\"start\":1.57,\"end\":2.73}]"}
{"utterance_id": "6241-61943-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0022.flac", "task_type": "lexical_detection", "query": "absent", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"absent\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.61, 2.0, "absent"]], "response_mode": "offline", "answer": "[{\"word\":\"absent\",\"start\":1.61,\"end\":2.0}]"}
{"utterance_id": "3081-166546-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0025.flac", "task_type": "lexical_detection", "query": "wish", "prompt": "Process the full audio offline. Find every instance of the word \"wish\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.81, "wish"]], "response_mode": "offline", "answer": "[{\"word\":\"wish\",\"start\":0.49,\"end\":0.81}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.58, 12.01, "shasta"]], "response_mode": "offline", "answer": "[{\"word\":\"shasta\",\"start\":11.58,\"end\":12.01}]"}
{"utterance_id": "2277-149897-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0003.flac", "task_type": "lexical_detection", "query": "recent", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"recent\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 1.94, "recent"]], "response_mode": "offline", "answer": "[{\"word\":\"recent\",\"start\":1.65,\"end\":1.94}]"}
{"utterance_id": "1993-147965-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0006.flac", "task_type": "lexical_detection", "query": "resting", "prompt": "Process the full audio offline. Find every instance of the word \"resting\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.64, 4.04, "resting"]], "response_mode": "offline", "answer": "[{\"word\":\"resting\",\"start\":3.64,\"end\":4.04}]"}
{"utterance_id": "5536-43363-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0019.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.37, "sioux"]], "response_mode": "offline", "answer": "[{\"word\":\"sioux\",\"start\":1.14,\"end\":1.37}]"}
{"utterance_id": "84-121550-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0029.flac", "task_type": "lexical_detection", "query": "weeping", "prompt": "After listening to the whole recording, output every instance of the word \"weeping\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.72, 5.1, "weeping"]], "response_mode": "offline", "answer": "[{\"word\":\"weeping\",\"start\":4.72,\"end\":5.1}]"}
{"utterance_id": "2277-149896-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.58, 2.01, "rector's"]], "response_mode": "offline", "answer": "[{\"word\":\"rector's\",\"start\":1.58,\"end\":2.01}]"}
{"utterance_id": "2086-149220-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0007.flac", "task_type": "lexical_detection", "query": "chanticleer", "prompt": "Process the full audio offline. Find every instance of the word \"chanticleer\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.34, "chanticleer"]], "response_mode": "offline", "answer": "[{\"word\":\"chanticleer\",\"start\":1.6,\"end\":2.34}]"}
{"utterance_id": "3081-166546-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0028.flac", "task_type": "lexical_detection", "query": "triumph", "prompt": "When the utterance is complete, report all instances of the word \"triumph\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 2.91, "triumph"]], "response_mode": "offline", "answer": "[{\"word\":\"triumph\",\"start\":2.32,\"end\":2.91}]"}
{"utterance_id": "7850-73752-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.96, 10.31, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":9.96,\"end\":10.31}]"}
{"utterance_id": "7850-73752-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0008.flac", "task_type": "lexical_detection", "query": "forlorn", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"forlorn\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.1, 5.82, "forlorn"]], "response_mode": "offline", "answer": "[{\"word\":\"forlorn\",\"start\":5.1,\"end\":5.82}]"}
{"utterance_id": "8297-275154-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0027.flac", "task_type": "lexical_detection", "query": "servants", "prompt": "Process the full audio offline. Find every instance of the word \"servants\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.86, "servants"]], "response_mode": "offline", "answer": "[{\"word\":\"servants\",\"start\":2.14,\"end\":2.86}]"}
{"utterance_id": "3853-163249-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0031.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.6, 2.97, "bait"], [6.45, 6.85, "cakes"]], "response_mode": "offline", "answer": "[{\"word\":\"bait\",\"start\":2.6,\"end\":2.97},{\"word\":\"cakes\",\"start\":6.45,\"end\":6.85}]"}
{"utterance_id": "1919-142785-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0016.flac", "task_type": "lexical_detection", "query": "lemon", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"lemon\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.28, "lemon"]], "response_mode": "offline", "answer": "[{\"word\":\"lemon\",\"start\":1.93,\"end\":2.28}]"}
{"utterance_id": "7850-111771-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 1.13, "general halleck"]], "response_mode": "offline", "answer": "[{\"word\":\"general halleck\",\"start\":0.28,\"end\":1.13}]"}
{"utterance_id": "3853-163249-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 5.61, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"christie\",\"start\":5.14,\"end\":5.61}]"}
{"utterance_id": "1272-135031-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0001.flac", "task_type": "lexical_detection", "query": "managed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"managed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.32, 4.75, "managed"]], "response_mode": "offline", "answer": "[{\"word\":\"managed\",\"start\":4.32,\"end\":4.75}]"}
{"utterance_id": "3752-4944-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0028.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After listening to the whole recording, output every instance of a date with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.66, "morrow"]], "response_mode": "offline", "answer": "[{\"word\":\"morrow\",\"start\":1.23,\"end\":1.66}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a number: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.73, 2.03, "one"], [5.27, 5.64, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.73,\"end\":2.03},{\"word\":\"two\",\"start\":5.27,\"end\":5.64}]"}
{"utterance_id": "2035-147961-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0018.flac", "task_type": "lexical_detection", "query": "parents", "prompt": "After the audio ends, extract all occurrences of the word \"parents\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.08, 4.51, "parents"]], "response_mode": "offline", "answer": "[{\"word\":\"parents\",\"start\":4.08,\"end\":4.51}]"}
{"utterance_id": "3536-23268-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0017.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[8.24, 8.49, "one"], [16.26, 16.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.24,\"end\":8.49},{\"word\":\"two\",\"start\":16.26,\"end\":16.69}]"}
{"utterance_id": "6241-66616-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.64, 1.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.64,\"end\":1.88}]"}
{"utterance_id": "2803-161169-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0009.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.31, 17.84, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":17.31,\"end\":17.84}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 1.16, "waiters"]], "response_mode": "offline", "answer": "[{\"word\":\"waiters\",\"start\":0.68,\"end\":1.16}]"}
{"utterance_id": "2412-153954-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.57, 4.01, "village"]], "response_mode": "offline", "answer": "[{\"word\":\"village\",\"start\":3.57,\"end\":4.01}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.88, 13.14, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":12.88,\"end\":13.14}]"}
{"utterance_id": "2035-147960-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.66, 3.19, "twenty four"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty four\",\"start\":2.66,\"end\":3.19}]"}
{"utterance_id": "6319-64726-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0020.flac", "task_type": "lexical_detection", "query": "behold", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"behold\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.59, "behold"]], "response_mode": "offline", "answer": "[{\"word\":\"behold\",\"start\":2.93,\"end\":3.59}]"}
{"utterance_id": "6319-275224-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.41, "silver"]], "response_mode": "offline", "answer": "[{\"word\":\"silver\",\"start\":1.88,\"end\":2.41}]"}
{"utterance_id": "1993-147149-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.95, "mary"], [10.92, 11.43, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":0.51,\"end\":0.95},{\"word\":\"mary\",\"start\":10.92,\"end\":11.43}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.79, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":0.39,\"end\":0.79}]"}
{"utterance_id": "7850-286674-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0014.flac", "task_type": "lexical_detection", "query": "sunlight", "prompt": "After the audio ends, extract all occurrences of the word \"sunlight\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.91, 7.51, "sunlight"]], "response_mode": "offline", "answer": "[{\"word\":\"sunlight\",\"start\":6.91,\"end\":7.51}]"}
{"utterance_id": "2086-149220-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0021.flac", "task_type": "lexical_detection", "query": "come", "prompt": "Once the audio has finished, identify every instance of the word \"come\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.25, 2.45, "come"]], "response_mode": "offline", "answer": "[{\"word\":\"come\",\"start\":2.25,\"end\":2.45}]"}
{"utterance_id": "3536-8226-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.42, 3.84, "husband's"], [8.94, 9.38, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband's\",\"start\":3.42,\"end\":3.84},{\"word\":\"husband\",\"start\":8.94,\"end\":9.38}]"}
{"utterance_id": "2412-153954-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0019.flac", "task_type": "lexical_detection", "query": "pocket", "prompt": "Process the full audio offline. Find every instance of the word \"pocket\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.91, 4.29, "pocket"]], "response_mode": "offline", "answer": "[{\"word\":\"pocket\",\"start\":3.91,\"end\":4.29}]"}
{"utterance_id": "6241-61943-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0027.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.46, 2.78, "fair"]], "response_mode": "offline", "answer": "[{\"word\":\"fair\",\"start\":2.46,\"end\":2.78}]"}
{"utterance_id": "2035-147961-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.5, "peter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":1.19,\"end\":1.5}]"}
{"utterance_id": "1462-170142-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.38, 6.94, "savoy"]], "response_mode": "offline", "answer": "[{\"word\":\"savoy\",\"start\":6.38,\"end\":6.94}]"}
{"utterance_id": "2086-149220-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0011.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.35, 8.81, "one"], [9.84, 10.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.35,\"end\":8.81},{\"word\":\"two\",\"start\":9.84,\"end\":10.38}]"}
{"utterance_id": "6241-66616-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.64, 2.95, "newsome"], [4.13, 4.83, "minnetaki"]], "response_mode": "offline", "answer": "[{\"word\":\"newsome\",\"start\":2.64,\"end\":2.95},{\"word\":\"minnetaki\",\"start\":4.13,\"end\":4.83}]"}
{"utterance_id": "2902-9008-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.42, 1.94, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":1.42,\"end\":1.94}]"}
{"utterance_id": "6319-57405-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0011.flac", "task_type": "lexical_detection", "query": "deucalion", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"deucalion\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.51, 3.21, "deucalion"]], "response_mode": "offline", "answer": "[{\"word\":\"deucalion\",\"start\":2.51,\"end\":3.21}]"}
{"utterance_id": "6313-66125-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.55, 4.93, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":4.55,\"end\":4.93}]"}
{"utterance_id": "5895-34622-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0006.flac", "task_type": "lexical_detection", "query": "smoke", "prompt": "Process the full audio offline. Find every instance of the word \"smoke\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.01, 3.42, "smoke"]], "response_mode": "offline", "answer": "[{\"word\":\"smoke\",\"start\":3.01,\"end\":3.42}]"}
{"utterance_id": "1993-147149-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.7, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":1.33,\"end\":1.7}]"}
{"utterance_id": "3576-138058-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0039.flac", "task_type": "lexical_detection", "query": "generous", "prompt": "After listening to the whole recording, output every instance of the word \"generous\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.82, 10.3, "generous"]], "response_mode": "offline", "answer": "[{\"word\":\"generous\",\"start\":9.82,\"end\":10.3}]"}
{"utterance_id": "6295-64301-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.92, 1.6, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":0.92,\"end\":1.6}]"}
{"utterance_id": "6313-66125-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0025.flac", "task_type": "lexical_detection", "query": "making", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"making\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.99, 2.29, "making"]], "response_mode": "offline", "answer": "[{\"word\":\"making\",\"start\":1.99,\"end\":2.29}]"}
{"utterance_id": "6345-93302-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0003.flac", "task_type": "lexical_detection", "query": "limit", "prompt": "Process the full audio offline. Find every instance of the word \"limit\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.76, "limit"]], "response_mode": "offline", "answer": "[{\"word\":\"limit\",\"start\":1.42,\"end\":1.76}]"}
{"utterance_id": "5338-24640-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0004.flac", "task_type": "lexical_detection", "query": "expedition", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"expedition\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.85, 4.71, "expedition"]], "response_mode": "offline", "answer": "[{\"word\":\"expedition\",\"start\":3.85,\"end\":4.71}]"}
{"utterance_id": "174-50561-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0003.flac", "task_type": "lexical_detection", "query": "approaches", "prompt": "When the utterance is complete, report all instances of the word \"approaches\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.96, "approaches"]], "response_mode": "offline", "answer": "[{\"word\":\"approaches\",\"start\":1.44,\"end\":1.96}]"}
{"utterance_id": "1673-143397-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0007.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Process the full audio offline. Find every instance of a country name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.95, 4.65, "persians"]], "response_mode": "offline", "answer": "[{\"word\":\"persians\",\"start\":3.95,\"end\":4.65}]"}
{"utterance_id": "5536-43363-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0019.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.14, "prophet"]], "response_mode": "offline", "answer": "[{\"word\":\"prophet\",\"start\":1.63,\"end\":2.14}]"}
{"utterance_id": "7976-105575-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0008.flac", "task_type": "lexical_detection", "query": "tried", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"tried\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.49, 3.76, "tried"]], "response_mode": "offline", "answer": "[{\"word\":\"tried\",\"start\":3.49,\"end\":3.76}]"}
{"utterance_id": "1988-147956-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.77, 7.31, "finger"]], "response_mode": "offline", "answer": "[{\"word\":\"finger\",\"start\":6.77,\"end\":7.31}]"}
{"utterance_id": "777-126732-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0064.flac", "task_type": "lexical_detection", "query": "clock", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"clock\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.17, 5.57, "clock"]], "response_mode": "offline", "answer": "[{\"word\":\"clock\",\"start\":5.17,\"end\":5.57}]"}
{"utterance_id": "2902-9008-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0010.flac", "task_type": "lexical_detection", "query": "immortal", "prompt": "After listening to the whole recording, output every instance of the word \"immortal\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.66, 7.07, "immortal"]], "response_mode": "offline", "answer": "[{\"word\":\"immortal\",\"start\":6.66,\"end\":7.07}]"}
{"utterance_id": "422-122949-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0013.flac", "task_type": "lexical_counting", "query": "growth", "prompt": "After the audio ends, tally every instance of the word \"growth\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[10.49, 10.86, "one"], [14.77, 15.24, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.49,\"end\":10.86},{\"word\":\"two\",\"start\":14.77,\"end\":15.24}]"}
{"utterance_id": "2035-152373-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0005.flac", "task_type": "lexical_detection", "query": "trembled", "prompt": "Process the full audio offline. Find every instance of the word \"trembled\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.94, 9.43, "trembled"]], "response_mode": "offline", "answer": "[{\"word\":\"trembled\",\"start\":8.94,\"end\":9.43}]"}
{"utterance_id": "1462-170145-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0022.flac", "task_type": "lexical_counting", "query": "don't", "prompt": "Listen to the full audio. Count every occurrence of the word \"don't\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.29, 0.47, "one"], [2.05, 2.26, "two"], [3.84, 4.29, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.29,\"end\":0.47},{\"word\":\"two\",\"start\":2.05,\"end\":2.26},{\"word\":\"three\",\"start\":3.84,\"end\":4.29}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.62, 3.97, "heart"], [9.11, 9.57, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":3.62,\"end\":3.97},{\"word\":\"face\",\"start\":9.11,\"end\":9.57}]"}
{"utterance_id": "7850-111771-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.17, 5.83, "west point"]], "response_mode": "offline", "answer": "[{\"word\":\"west point\",\"start\":5.17,\"end\":5.83}]"}
{"utterance_id": "7850-281318-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0015.flac", "task_type": "lexical_detection", "query": "begin", "prompt": "After the audio ends, extract all occurrences of the word \"begin\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.59, "begin"]], "response_mode": "offline", "answer": "[{\"word\":\"begin\",\"start\":1.87,\"end\":2.59}]"}
{"utterance_id": "5338-284437-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0033.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 1.9, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.38,\"end\":1.9}]"}
{"utterance_id": "7850-286674-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0002.flac", "task_type": "lexical_counting", "query": "bringing", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"bringing\" with word/start/end. If none, reply [].", "labels": [[2.3, 2.67, "one"], [4.12, 4.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.3,\"end\":2.67},{\"word\":\"two\",\"start\":4.12,\"end\":4.53}]"}
{"utterance_id": "6313-66129-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0025.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.19, 3.61, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.19,\"end\":3.61}]"}
{"utterance_id": "1993-147149-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0030.flac", "task_type": "lexical_counting", "query": "father", "prompt": "Once the audio has finished, list each occurrence of the word \"father\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[8.74, 9.2, "one"], [10.43, 10.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.74,\"end\":9.2},{\"word\":\"two\",\"start\":10.43,\"end\":10.9}]"}
{"utterance_id": "1462-170138-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.98, 3.62, "alexander's"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander's\",\"start\":2.98,\"end\":3.62}]"}
{"utterance_id": "7976-105575-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0003.flac", "task_type": "lexical_detection", "query": "hamilton's", "prompt": "When the utterance is complete, report all instances of the word \"hamilton's\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.57, 4.27, "hamilton's"]], "response_mode": "offline", "answer": "[{\"word\":\"hamilton's\",\"start\":3.57,\"end\":4.27}]"}
{"utterance_id": "3000-15664-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0017.flac", "task_type": "lexical_detection", "query": "elevation", "prompt": "When the utterance is complete, report all instances of the word \"elevation\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.71, 5.52, "elevation"]], "response_mode": "offline", "answer": "[{\"word\":\"elevation\",\"start\":4.71,\"end\":5.52}]"}
{"utterance_id": "6345-93306-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0017.flac", "task_type": "lexical_counting", "query": "house", "prompt": "After the audio ends, tally every instance of the word \"house\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.83, 2.31, "one"], [6.8, 7.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.83,\"end\":2.31},{\"word\":\"two\",\"start\":6.8,\"end\":7.31}]"}
{"utterance_id": "6295-64301-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.06, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"father's\",\"start\":5.59,\"end\":6.06}]"}
{"utterance_id": "5536-43359-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0013.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.44, "meat"]], "response_mode": "offline", "answer": "[{\"word\":\"meat\",\"start\":2.09,\"end\":2.44}]"}
{"utterance_id": "6345-93306-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0000.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the full audio. Count every occurrence of a number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.34, 12.69, "one"], [19.21, 19.42, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":12.34,\"end\":12.69},{\"word\":\"two\",\"start\":19.21,\"end\":19.42}]"}
{"utterance_id": "84-121123-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0022.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 1.0, "doctors"]], "response_mode": "offline", "answer": "[{\"word\":\"doctors\",\"start\":0.55,\"end\":1.0}]"}
{"utterance_id": "422-122949-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0005.flac", "task_type": "lexical_detection", "query": "regard", "prompt": "Once the audio has finished, identify every instance of the word \"regard\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.36, 2.76, "regard"]], "response_mode": "offline", "answer": "[{\"word\":\"regard\",\"start\":2.36,\"end\":2.76}]"}
{"utterance_id": "2428-83699-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0040.flac", "task_type": "lexical_detection", "query": "lighted", "prompt": "Once the audio has finished, identify every instance of the word \"lighted\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.19, 6.52, "lighted"]], "response_mode": "offline", "answer": "[{\"word\":\"lighted\",\"start\":6.19,\"end\":6.52}]"}
{"utterance_id": "3536-23268-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0015.flac", "task_type": "lexical_detection", "query": "command", "prompt": "Once the audio has finished, identify every instance of the word \"command\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.97, 1.4, "command"]], "response_mode": "offline", "answer": "[{\"word\":\"command\",\"start\":0.97,\"end\":1.4}]"}
{"utterance_id": "8842-302201-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.5, "eyes"], [5.93, 6.23, "eyes"], [8.1, 8.42, "heart's"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.21,\"end\":1.5},{\"word\":\"eyes\",\"start\":5.93,\"end\":6.23},{\"word\":\"heart's\",\"start\":8.1,\"end\":8.42}]"}
{"utterance_id": "84-121123-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0024.flac", "task_type": "lexical_detection", "query": "district", "prompt": "Once the audio has finished, identify every instance of the word \"district\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.29, 0.74, "district"]], "response_mode": "offline", "answer": "[{\"word\":\"district\",\"start\":0.29,\"end\":0.74}]"}
{"utterance_id": "1993-147149-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0020.flac", "task_type": "lexical_detection", "query": "brothers", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"brothers\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.26, 5.65, "brothers"]], "response_mode": "offline", "answer": "[{\"word\":\"brothers\",\"start\":5.26,\"end\":5.65}]"}
{"utterance_id": "6319-57405-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0005.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.29, "one"], [4.3, 4.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.46,\"end\":3.29},{\"word\":\"two\",\"start\":4.3,\"end\":4.89}]"}
{"utterance_id": "2803-154320-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.39, 1.04, "minds"]], "response_mode": "offline", "answer": "[{\"word\":\"minds\",\"start\":0.39,\"end\":1.04}]"}
{"utterance_id": "6319-64726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.56, 2.04, "fairy"]], "response_mode": "offline", "answer": "[{\"word\":\"fairy\",\"start\":1.56,\"end\":2.04}]"}
{"utterance_id": "2086-149220-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0037.flac", "task_type": "lexical_detection", "query": "hepzibah's", "prompt": "After listening to the whole recording, output every instance of the word \"hepzibah's\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.58, 2.19, "hepzibah's"]], "response_mode": "offline", "answer": "[{\"word\":\"hepzibah's\",\"start\":1.58,\"end\":2.19}]"}
{"utterance_id": "5694-64025-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[5.11, 5.45, "ear"], [6.45, 7.01, "nose"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":5.11,\"end\":5.45},{\"word\":\"nose\",\"start\":6.45,\"end\":7.01}]"}
{"utterance_id": "1993-147149-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0029.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.49, 3.25, "miss simmonds"]], "response_mode": "offline", "answer": "[{\"word\":\"miss simmonds\",\"start\":2.49,\"end\":3.25}]"}
{"utterance_id": "3752-4944-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0016.flac", "task_type": "lexical_detection", "query": "knowledge", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"knowledge\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.74, 7.14, "knowledge"]], "response_mode": "offline", "answer": "[{\"word\":\"knowledge\",\"start\":6.74,\"end\":7.14}]"}
{"utterance_id": "2902-9006-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0016.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[7.29, 8.18, "north africa"]], "response_mode": "offline", "answer": "[{\"word\":\"north africa\",\"start\":7.29,\"end\":8.18}]"}
{"utterance_id": "6345-93306-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0012.flac", "task_type": "lexical_detection", "query": "impatiently", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"impatiently\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.66, 2.42, "impatiently"]], "response_mode": "offline", "answer": "[{\"word\":\"impatiently\",\"start\":1.66,\"end\":2.42}]"}
{"utterance_id": "3170-137482-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0046.flac", "task_type": "lexical_detection", "query": "gondola", "prompt": "Process the full audio offline. Find every instance of the word \"gondola\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.5, 5.09, "gondola"]], "response_mode": "offline", "answer": "[{\"word\":\"gondola\",\"start\":4.5,\"end\":5.09}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "When the utterance is complete, report how a profession accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.57, "one"], [4.01, 4.55, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.09,\"end\":2.57},{\"word\":\"two\",\"start\":4.01,\"end\":4.55}]"}
{"utterance_id": "7976-110124-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0002.flac", "task_type": "lexical_detection", "query": "evils", "prompt": "After the audio ends, extract all occurrences of the word \"evils\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.83, 6.32, "evils"]], "response_mode": "offline", "answer": "[{\"word\":\"evils\",\"start\":5.83,\"end\":6.32}]"}
{"utterance_id": "6295-244435-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.54, 0.98, "seven"]], "response_mode": "offline", "answer": "[{\"word\":\"seven\",\"start\":0.54,\"end\":0.98}]"}
{"utterance_id": "1462-170142-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0032.flac", "task_type": "lexical_detection", "query": "want", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"want\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 0.82, "want"]], "response_mode": "offline", "answer": "[{\"word\":\"want\",\"start\":0.65,\"end\":0.82}]"}
{"utterance_id": "251-137823-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0008.flac", "task_type": "lexical_detection", "query": "raised", "prompt": "After listening to the whole recording, output every instance of the word \"raised\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.96, 4.36, "raised"]], "response_mode": "offline", "answer": "[{\"word\":\"raised\",\"start\":3.96,\"end\":4.36}]"}
{"utterance_id": "652-130726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.98, "thompson"]], "response_mode": "offline", "answer": "[{\"word\":\"thompson\",\"start\":0.4,\"end\":0.98}]"}
{"utterance_id": "652-129742-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0017.flac", "task_type": "lexical_detection", "query": "attached", "prompt": "Process the full audio offline. Find every instance of the word \"attached\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.24, "attached"]], "response_mode": "offline", "answer": "[{\"word\":\"attached\",\"start\":1.8,\"end\":2.24}]"}
{"utterance_id": "5536-43363-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.91, 4.28, "feast"]], "response_mode": "offline", "answer": "[{\"word\":\"feast\",\"start\":3.91,\"end\":4.28}]"}
{"utterance_id": "8842-302196-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0005.flac", "task_type": "lexical_counting", "query": "cadence", "prompt": "When the utterance is complete, report how the word \"cadence\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.36, 7.84, "one"], [12.1, 12.71, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.36,\"end\":7.84},{\"word\":\"two\",\"start\":12.1,\"end\":12.71}]"}
{"utterance_id": "1993-147149-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0016.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.46, "frame"]], "response_mode": "offline", "answer": "[{\"word\":\"frame\",\"start\":3.11,\"end\":3.46}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Listen to the full audio. Count every occurrence of a color name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.69, 2.88, "one"], [6.84, 7.15, "two"], [10.71, 10.97, "three"], [12.06, 12.35, "four"], [14.96, 15.27, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.69,\"end\":2.88},{\"word\":\"two\",\"start\":6.84,\"end\":7.15},{\"word\":\"three\",\"start\":10.71,\"end\":10.97},{\"word\":\"four\",\"start\":12.06,\"end\":12.35},{\"word\":\"five\",\"start\":14.96,\"end\":15.27}]"}
{"utterance_id": "2035-152373-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0008.flac", "task_type": "lexical_detection", "query": "leading", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"leading\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.0, "leading"]], "response_mode": "offline", "answer": "[{\"word\":\"leading\",\"start\":0.66,\"end\":1.0}]"}
{"utterance_id": "174-50561-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0018.flac", "task_type": "lexical_detection", "query": "bed", "prompt": "After listening to the whole recording, output every instance of the word \"bed\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.74, "bed"]], "response_mode": "offline", "answer": "[{\"word\":\"bed\",\"start\":0.5,\"end\":0.74}]"}
{"utterance_id": "7976-105575-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0029.flac", "task_type": "lexical_detection", "query": "discovery", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"discovery\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.88, 6.57, "discovery"]], "response_mode": "offline", "answer": "[{\"word\":\"discovery\",\"start\":5.88,\"end\":6.57}]"}
{"utterance_id": "1988-24833-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0026.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.74, 2.09, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":1.74,\"end\":2.09}]"}
{"utterance_id": "3081-166546-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0041.flac", "task_type": "lexical_detection", "query": "gryce", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"gryce\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.6, 2.23, "gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"gryce\",\"start\":1.6,\"end\":2.23}]"}
{"utterance_id": "2078-142845-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0041.flac", "task_type": "lexical_detection", "query": "buns", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"buns\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.53, 2.28, "buns"]], "response_mode": "offline", "answer": "[{\"word\":\"buns\",\"start\":1.53,\"end\":2.28}]"}
{"utterance_id": "6345-93306-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0025.flac", "task_type": "lexical_detection", "query": "taste", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"taste\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.31, 8.74, "taste"]], "response_mode": "offline", "answer": "[{\"word\":\"taste\",\"start\":8.31,\"end\":8.74}]"}
{"utterance_id": "3752-4944-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0056.flac", "task_type": "lexical_detection", "query": "roared", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"roared\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.42, 2.77, "roared"]], "response_mode": "offline", "answer": "[{\"word\":\"roared\",\"start\":2.42,\"end\":2.77}]"}
{"utterance_id": "5536-43363-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0001.flac", "task_type": "lexical_detection", "query": "disgraceful", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"disgraceful\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.4, 5.08, "disgraceful"]], "response_mode": "offline", "answer": "[{\"word\":\"disgraceful\",\"start\":4.4,\"end\":5.08}]"}
{"utterance_id": "3170-137482-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.76, 15.46, "shoulders"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulders\",\"start\":14.76,\"end\":15.46}]"}
{"utterance_id": "7976-110124-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0019.flac", "task_type": "lexical_detection", "query": "harm", "prompt": "After the audio ends, extract all occurrences of the word \"harm\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.25, "harm"]], "response_mode": "offline", "answer": "[{\"word\":\"harm\",\"start\":1.76,\"end\":2.25}]"}
{"utterance_id": "1462-170138-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.45, 9.09, "shakespeare's"], [12.02, 13.06, "ernest dowson"]], "response_mode": "offline", "answer": "[{\"word\":\"shakespeare's\",\"start\":8.45,\"end\":9.09},{\"word\":\"ernest dowson\",\"start\":12.02,\"end\":13.06}]"}
{"utterance_id": "5694-64038-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0003.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.05, 6.53, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":6.05,\"end\":6.53}]"}
{"utterance_id": "777-126732-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0032.flac", "task_type": "lexical_detection", "query": "hurt", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"hurt\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.86, 3.15, "hurt"]], "response_mode": "offline", "answer": "[{\"word\":\"hurt\",\"start\":2.86,\"end\":3.15}]"}
{"utterance_id": "1462-170142-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 5.12, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander\",\"start\":4.44,\"end\":5.12}]"}
{"utterance_id": "5694-64025-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 1.0, "gladden"]], "response_mode": "offline", "answer": "[{\"word\":\"gladden\",\"start\":0.62,\"end\":1.0}]"}
{"utterance_id": "2277-149874-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.22, 3.12, "wife's sister"]], "response_mode": "offline", "answer": "[{\"word\":\"wife's sister\",\"start\":2.22,\"end\":3.12}]"}
{"utterance_id": "7976-105575-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0014.flac", "task_type": "lexical_detection", "query": "division", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"division\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.41, 4.01, "division"]], "response_mode": "offline", "answer": "[{\"word\":\"division\",\"start\":3.41,\"end\":4.01}]"}
{"utterance_id": "652-129742-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0005.flac", "task_type": "lexical_detection", "query": "seasoned", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"seasoned\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.01, 5.57, "seasoned"]], "response_mode": "offline", "answer": "[{\"word\":\"seasoned\",\"start\":5.01,\"end\":5.57}]"}
{"utterance_id": "1919-142785-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0043.flac", "task_type": "lexical_detection", "query": "forcemeat", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"forcemeat\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.0, 5.62, "forcemeat"]], "response_mode": "offline", "answer": "[{\"word\":\"forcemeat\",\"start\":5.0,\"end\":5.62}]"}
{"utterance_id": "5694-64025-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0006.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.11, 2.78, "brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"brigade\",\"start\":2.11,\"end\":2.78}]"}
{"utterance_id": "1462-170138-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.34, 8.01, "sir harry's"]], "response_mode": "offline", "answer": "[{\"word\":\"sir harry's\",\"start\":7.34,\"end\":8.01}]"}
{"utterance_id": "1673-143396-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0007.flac", "task_type": "lexical_detection", "query": "prevailing", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"prevailing\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.62, 1.19, "prevailing"]], "response_mode": "offline", "answer": "[{\"word\":\"prevailing\",\"start\":0.62,\"end\":1.19}]"}
{"utterance_id": "5694-64025-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 1.26, "shiloh"]], "response_mode": "offline", "answer": "[{\"word\":\"shiloh\",\"start\":0.54,\"end\":1.26}]"}
{"utterance_id": "2428-83699-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0034.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.1, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":0.75,\"end\":1.1}]"}
{"utterance_id": "1919-142785-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0034.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.3, 1.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.3,\"end\":1.67}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an animal name with word/start/end. If none, reply [].", "labels": [[2.9, 3.32, "one"], [6.92, 7.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.9,\"end\":3.32},{\"word\":\"two\",\"start\":6.92,\"end\":7.48}]"}
{"utterance_id": "1988-147956-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0024.flac", "task_type": "lexical_detection", "query": "hands", "prompt": "Process the full audio offline. Find every instance of the word \"hands\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.75, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.23,\"end\":2.75}]"}
{"utterance_id": "1988-147956-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0011.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[0.44, 0.71, "one"], [2.44, 2.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.44,\"end\":0.71},{\"word\":\"two\",\"start\":2.44,\"end\":2.89}]"}
{"utterance_id": "6313-76958-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 0.89, "walter"]], "response_mode": "offline", "answer": "[{\"word\":\"walter\",\"start\":0.52,\"end\":0.89}]"}
{"utterance_id": "5338-24640-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0005.flac", "task_type": "lexical_detection", "query": "talents", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"talents\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 2.07, "talents"]], "response_mode": "offline", "answer": "[{\"word\":\"talents\",\"start\":1.56,\"end\":2.07}]"}
{"utterance_id": "2078-142845-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0049.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.62, 2.37, "rusks"]], "response_mode": "offline", "answer": "[{\"word\":\"rusks\",\"start\":1.62,\"end\":2.37}]"}
{"utterance_id": "2035-147960-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0011.flac", "task_type": "lexical_detection", "query": "ride", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"ride\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.3, "ride"]], "response_mode": "offline", "answer": "[{\"word\":\"ride\",\"start\":2.02,\"end\":2.3}]"}
{"utterance_id": "251-137823-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.5, 1.0, "inventor"]], "response_mode": "offline", "answer": "[{\"word\":\"inventor\",\"start\":0.5,\"end\":1.0}]"}
{"utterance_id": "1919-142785-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0005.flac", "task_type": "lexical_detection", "query": "sprinkle", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sprinkle\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.66, 5.17, "sprinkle"]], "response_mode": "offline", "answer": "[{\"word\":\"sprinkle\",\"start\":4.66,\"end\":5.17}]"}
{"utterance_id": "2035-147961-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0011.flac", "task_type": "lexical_detection", "query": "covered", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"covered\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.48, "covered"]], "response_mode": "offline", "answer": "[{\"word\":\"covered\",\"start\":1.1,\"end\":1.48}]"}
{"utterance_id": "7976-105575-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0006.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "Listen to the entire clip, then report the count sequence for an organization name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.63, 5.5, "one"], [8.24, 8.93, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.63,\"end\":5.5},{\"word\":\"two\",\"start\":8.24,\"end\":8.93}]"}
{"utterance_id": "2086-149220-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0041.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.83, 12.23, "cheek"]], "response_mode": "offline", "answer": "[{\"word\":\"cheek\",\"start\":11.83,\"end\":12.23}]"}
{"utterance_id": "2803-154328-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.14, "jailer"]], "response_mode": "offline", "answer": "[{\"word\":\"jailer\",\"start\":0.66,\"end\":1.14}]"}
{"utterance_id": "7976-110124-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0012.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the entire clip, then report the count sequence for a food name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 3.01, "one"], [4.55, 5.11, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.5,\"end\":3.01},{\"word\":\"two\",\"start\":4.55,\"end\":5.11}]"}
{"utterance_id": "6313-76958-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0014.flac", "task_type": "lexical_detection", "query": "waking", "prompt": "After listening to the whole recording, output every instance of the word \"waking\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.98, 6.38, "waking"]], "response_mode": "offline", "answer": "[{\"word\":\"waking\",\"start\":5.98,\"end\":6.38}]"}
{"utterance_id": "8297-275155-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.67, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":2.16,\"end\":2.67}]"}
{"utterance_id": "1988-148538-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0011.flac", "task_type": "lexical_detection", "query": "privileges", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"privileges\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.45, 4.29, "privileges"]], "response_mode": "offline", "answer": "[{\"word\":\"privileges\",\"start\":3.45,\"end\":4.29}]"}
{"utterance_id": "2412-153954-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0017.flac", "task_type": "lexical_detection", "query": "contained", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"contained\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.2, 3.86, "contained"]], "response_mode": "offline", "answer": "[{\"word\":\"contained\",\"start\":3.2,\"end\":3.86}]"}
{"utterance_id": "3576-138058-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0018.flac", "task_type": "lexical_detection", "query": "carrying", "prompt": "After the audio ends, extract all occurrences of the word \"carrying\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 5.84, "carrying"]], "response_mode": "offline", "answer": "[{\"word\":\"carrying\",\"start\":5.52,\"end\":5.84}]"}
{"utterance_id": "6241-66616-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0021.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Do not answer until the audio ends. Then list each instance of a money or currency term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.78, 1.02, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":0.78,\"end\":1.02}]"}
{"utterance_id": "2035-147961-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0032.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.1, 11.67, "village"]], "response_mode": "offline", "answer": "[{\"word\":\"village\",\"start\":11.1,\"end\":11.67}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.42, 15.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":15.42,\"end\":15.65}]"}
{"utterance_id": "3576-138058-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0015.flac", "task_type": "lexical_detection", "query": "pistols", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pistols\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.16, 5.75, "pistols"]], "response_mode": "offline", "answer": "[{\"word\":\"pistols\",\"start\":5.16,\"end\":5.75}]"}
{"utterance_id": "84-121123-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0018.flac", "task_type": "lexical_detection", "query": "suffered", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"suffered\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.97, "suffered"]], "response_mode": "offline", "answer": "[{\"word\":\"suffered\",\"start\":0.55,\"end\":0.97}]"}
{"utterance_id": "5338-24640-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.82, 1.61, "thirty three"]], "response_mode": "offline", "answer": "[{\"word\":\"thirty three\",\"start\":0.82,\"end\":1.61}]"}
{"utterance_id": "6313-66125-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0001.flac", "task_type": "lexical_detection", "query": "clump", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"clump\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.86, 1.16, "clump"]], "response_mode": "offline", "answer": "[{\"word\":\"clump\",\"start\":0.86,\"end\":1.16}]"}
{"utterance_id": "652-130726-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0028.flac", "task_type": "lexical_detection", "query": "meat", "prompt": "When the utterance is complete, report all instances of the word \"meat\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 1.98, "meat"]], "response_mode": "offline", "answer": "[{\"word\":\"meat\",\"start\":1.76,\"end\":1.98}]"}
{"utterance_id": "5694-64029-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0017.flac", "task_type": "lexical_detection", "query": "yankees", "prompt": "After the audio ends, extract all occurrences of the word \"yankees\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 0.93, "yankees"]], "response_mode": "offline", "answer": "[{\"word\":\"yankees\",\"start\":0.62,\"end\":0.93}]"}
{"utterance_id": "5895-34615-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0009.flac", "task_type": "lexical_detection", "query": "escape", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"escape\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.94, 1.38, "escape"]], "response_mode": "offline", "answer": "[{\"word\":\"escape\",\"start\":0.94,\"end\":1.38}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.1, 6.43, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":6.1,\"end\":6.43}]"}
{"utterance_id": "6241-66616-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0025.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.84, 7.15, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":6.84,\"end\":7.15}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "lexical_counting", "query": "red", "prompt": "After the audio ends, tally every instance of the word \"red\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.69, 2.88, "one"], [14.96, 15.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.69,\"end\":2.88},{\"word\":\"two\",\"start\":14.96,\"end\":15.27}]"}
{"utterance_id": "3000-15664-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0040.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.05, 6.51, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.05,\"end\":6.51}]"}
{"utterance_id": "1919-142785-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0029.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.99, "lemon juice"]], "response_mode": "offline", "answer": "[{\"word\":\"lemon juice\",\"start\":0.38,\"end\":0.99}]"}
{"utterance_id": "6319-275224-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0004.flac", "task_type": "lexical_detection", "query": "silver", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"silver\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.41, "silver"]], "response_mode": "offline", "answer": "[{\"word\":\"silver\",\"start\":1.88,\"end\":2.41}]"}
{"utterance_id": "1462-170142-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0027.flac", "task_type": "lexical_detection", "query": "sweetness", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sweetness\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.68, 2.18, "sweetness"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetness\",\"start\":1.68,\"end\":2.18}]"}
{"utterance_id": "251-136532-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0010.flac", "task_type": "lexical_detection", "query": "survey", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"survey\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.93, 4.43, "survey"]], "response_mode": "offline", "answer": "[{\"word\":\"survey\",\"start\":3.93,\"end\":4.43}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.05, 11.53, "rhodes"]], "response_mode": "offline", "answer": "[{\"word\":\"rhodes\",\"start\":11.05,\"end\":11.53}]"}
{"utterance_id": "1462-170142-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.69, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":1.21,\"end\":1.69}]"}
{"utterance_id": "3853-163249-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0031.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.24, "fish"]], "response_mode": "offline", "answer": "[{\"word\":\"fish\",\"start\":0.94,\"end\":1.24}]"}
{"utterance_id": "8842-302196-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0003.flac", "task_type": "lexical_detection", "query": "poetry", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"poetry\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.18, "poetry"]], "response_mode": "offline", "answer": "[{\"word\":\"poetry\",\"start\":1.63,\"end\":2.18}]"}
{"utterance_id": "3752-4944-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0045.flac", "task_type": "lexical_detection", "query": "sir", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sir\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.46, "sir"]], "response_mode": "offline", "answer": "[{\"word\":\"sir\",\"start\":2.03,\"end\":2.46}]"}
{"utterance_id": "84-121123-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0008.flac", "task_type": "lexical_counting", "query": "rave", "prompt": "After the audio ends, tally every instance of the word \"rave\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.66, 1.04, "one"], [6.42, 6.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.66,\"end\":1.04},{\"word\":\"two\",\"start\":6.42,\"end\":6.91}]"}
{"utterance_id": "5694-64025-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0020.flac", "task_type": "lexical_detection", "query": "persuade", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"persuade\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.07, 3.5, "persuade"]], "response_mode": "offline", "answer": "[{\"word\":\"persuade\",\"start\":3.07,\"end\":3.5}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "lexical_detection", "query": "stands", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"stands\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.62, 5.28, "stands"]], "response_mode": "offline", "answer": "[{\"word\":\"stands\",\"start\":4.62,\"end\":5.28}]"}
{"utterance_id": "6313-76958-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0029.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.61, 6.41, "cowpunchers"]], "response_mode": "offline", "answer": "[{\"word\":\"cowpunchers\",\"start\":5.61,\"end\":6.41}]"}
{"utterance_id": "7850-73752-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0009.flac", "task_type": "lexical_detection", "query": "entertained", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"entertained\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.38, 4.91, "entertained"]], "response_mode": "offline", "answer": "[{\"word\":\"entertained\",\"start\":4.38,\"end\":4.91}]"}
{"utterance_id": "3752-4944-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0040.flac", "task_type": "lexical_detection", "query": "doctor", "prompt": "When the utterance is complete, report all instances of the word \"doctor\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.22, 0.64, "doctor"]], "response_mode": "offline", "answer": "[{\"word\":\"doctor\",\"start\":0.22,\"end\":0.64}]"}
{"utterance_id": "2035-152373-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0014.flac", "task_type": "lexical_detection", "query": "process", "prompt": "After listening to the whole recording, output every instance of the word \"process\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.46, "process"]], "response_mode": "offline", "answer": "[{\"word\":\"process\",\"start\":1.86,\"end\":2.46}]"}
{"utterance_id": "6319-57405-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0002.flac", "task_type": "lexical_detection", "query": "company", "prompt": "Process the full audio offline. Find every instance of the word \"company\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.0, 3.56, "company"]], "response_mode": "offline", "answer": "[{\"word\":\"company\",\"start\":3.0,\"end\":3.56}]"}
{"utterance_id": "3576-138058-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 2.92, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.54,\"end\":2.92}]"}
{"utterance_id": "8297-275154-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.74, 2.31, "captain's"]], "response_mode": "offline", "answer": "[{\"word\":\"captain's\",\"start\":1.74,\"end\":2.31}]"}
{"utterance_id": "3000-15664-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0019.flac", "task_type": "lexical_detection", "query": "reappear", "prompt": "Once the audio has finished, identify every instance of the word \"reappear\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.09, 5.5, "reappear"]], "response_mode": "offline", "answer": "[{\"word\":\"reappear\",\"start\":5.09,\"end\":5.5}]"}
{"utterance_id": "6313-76958-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0029.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.03, 4.46, "meals"]], "response_mode": "offline", "answer": "[{\"word\":\"meals\",\"start\":4.03,\"end\":4.46}]"}
{"utterance_id": "1673-143396-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.56, 5.16, "socrates"], [12.13, 12.64, "jesus"]], "response_mode": "offline", "answer": "[{\"word\":\"socrates\",\"start\":4.56,\"end\":5.16},{\"word\":\"jesus\",\"start\":12.13,\"end\":12.64}]"}
{"utterance_id": "1272-141231-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0022.flac", "task_type": "lexical_detection", "query": "immediate", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"immediate\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.7, 2.22, "immediate"]], "response_mode": "offline", "answer": "[{\"word\":\"immediate\",\"start\":1.7,\"end\":2.22}]"}
{"utterance_id": "1988-24833-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.79, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cat\",\"start\":1.52,\"end\":1.79}]"}
{"utterance_id": "2086-149220-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0035.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[2.66, 3.27, "one"], [5.02, 5.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.66,\"end\":3.27},{\"word\":\"two\",\"start\":5.02,\"end\":5.45}]"}
{"utterance_id": "5338-284437-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0013.flac", "task_type": "lexical_detection", "query": "button", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"button\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.25, 4.6, "button"]], "response_mode": "offline", "answer": "[{\"word\":\"button\",\"start\":4.25,\"end\":4.6}]"}
{"utterance_id": "2428-83705-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.62, "snellings"]], "response_mode": "offline", "answer": "[{\"word\":\"snellings\",\"start\":1.9,\"end\":2.62}]"}
{"utterance_id": "777-126732-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0000.flac", "task_type": "lexical_detection", "query": "poorer", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"poorer\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.56, "poorer"]], "response_mode": "offline", "answer": "[{\"word\":\"poorer\",\"start\":2.05,\"end\":2.56}]"}
{"utterance_id": "6313-66129-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0023.flac", "task_type": "lexical_detection", "query": "sombrero", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sombrero\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.6, "sombrero"]], "response_mode": "offline", "answer": "[{\"word\":\"sombrero\",\"start\":2.08,\"end\":2.6}]"}
{"utterance_id": "84-121550-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0018.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.49, 5.12, "emerald"]], "response_mode": "offline", "answer": "[{\"word\":\"emerald\",\"start\":4.49,\"end\":5.12}]"}
{"utterance_id": "174-50561-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0016.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Listen to the entire clip, then report the count sequence for a profession as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.85, "one"], [8.43, 8.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.44,\"end\":1.85},{\"word\":\"two\",\"start\":8.43,\"end\":8.8}]"}
{"utterance_id": "777-126732-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0057.flac", "task_type": "lexical_detection", "query": "bare", "prompt": "After the audio ends, extract all occurrences of the word \"bare\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.68, "bare"]], "response_mode": "offline", "answer": "[{\"word\":\"bare\",\"start\":0.43,\"end\":0.68}]"}
{"utterance_id": "1988-148538-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0006.flac", "task_type": "lexical_detection", "query": "embrace", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"embrace\" with fields word, start, end (seconds). If none, reply [].", "labels": [[11.54, 12.03, "embrace"]], "response_mode": "offline", "answer": "[{\"word\":\"embrace\",\"start\":11.54,\"end\":12.03}]"}
{"utterance_id": "251-137823-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.14, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.59,\"end\":1.14}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a person's name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[3.29, 4.32, "one"], [5.42, 6.3, "two"], [12.96, 13.79, "three"], [21.05, 21.85, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.29,\"end\":4.32},{\"word\":\"two\",\"start\":5.42,\"end\":6.3},{\"word\":\"three\",\"start\":12.96,\"end\":13.79},{\"word\":\"four\",\"start\":21.05,\"end\":21.85}]"}
{"utterance_id": "1673-143397-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0007.flac", "task_type": "lexical_counting", "query": "exterminate", "prompt": "After listening to the whole recording, output timed count words for the word \"exterminate\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.31, 1.01, "one"], [3.19, 3.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.31,\"end\":1.01},{\"word\":\"two\",\"start\":3.19,\"end\":3.88}]"}
{"utterance_id": "3170-137482-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0047.flac", "task_type": "lexical_counting", "query": "everything", "prompt": "Listen to the full audio. Count every occurrence of the word \"everything\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.18, 7.85, "one"], [9.95, 10.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.18,\"end\":7.85},{\"word\":\"two\",\"start\":9.95,\"end\":10.53}]"}
{"utterance_id": "1919-142785-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0022.flac", "task_type": "semantic_detection", "query": "date", "prompt": "You will hear a complete utterance. When it ends, list every instance of a date you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.43, "easter"]], "response_mode": "offline", "answer": "[{\"word\":\"easter\",\"start\":2.93,\"end\":3.43}]"}
{"utterance_id": "8297-275155-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0012.flac", "task_type": "lexical_detection", "query": "catherine", "prompt": "After the audio ends, extract all occurrences of the word \"catherine\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.98, "catherine"]], "response_mode": "offline", "answer": "[{\"word\":\"catherine\",\"start\":1.47,\"end\":1.98}]"}
{"utterance_id": "1272-128104-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0008.flac", "task_type": "lexical_detection", "query": "kinds", "prompt": "Process the full audio offline. Find every instance of the word \"kinds\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.91, "kinds"]], "response_mode": "offline", "answer": "[{\"word\":\"kinds\",\"start\":2.35,\"end\":2.91}]"}
{"utterance_id": "5694-64029-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0023.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.15, "hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"hundred\",\"start\":1.74,\"end\":2.15}]"}
{"utterance_id": "6345-93306-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0006.flac", "task_type": "lexical_detection", "query": "towards", "prompt": "When the utterance is complete, report all instances of the word \"towards\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.29, "towards"]], "response_mode": "offline", "answer": "[{\"word\":\"towards\",\"start\":0.87,\"end\":1.29}]"}
{"utterance_id": "84-121550-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0017.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.45, 5.84, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":5.45,\"end\":5.84}]"}
{"utterance_id": "3853-163249-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0031.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Process the full audio offline. For every occurrence of a food name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.6, 2.97, "one"], [6.45, 6.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.6,\"end\":2.97},{\"word\":\"two\",\"start\":6.45,\"end\":6.85}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "semantic_detection", "query": "color", "prompt": "You will hear a complete utterance. When it ends, list every instance of a color name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.13, 5.45, "blue"]], "response_mode": "offline", "answer": "[{\"word\":\"blue\",\"start\":5.13,\"end\":5.45}]"}
{"utterance_id": "422-122949-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0014.flac", "task_type": "lexical_detection", "query": "frightfully", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"frightfully\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[12.26, 12.83, "frightfully"]], "response_mode": "offline", "answer": "[{\"word\":\"frightfully\",\"start\":12.26,\"end\":12.83}]"}
{"utterance_id": "3081-166546-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 4.53, "detective gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"detective gryce\",\"start\":3.26,\"end\":4.53}]"}
{"utterance_id": "3170-137482-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0004.flac", "task_type": "lexical_detection", "query": "exposed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"exposed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.52, "exposed"]], "response_mode": "offline", "answer": "[{\"word\":\"exposed\",\"start\":1.97,\"end\":2.52}]"}
{"utterance_id": "1988-24833-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0001.flac", "task_type": "lexical_counting", "query": "he's", "prompt": "Process the full audio offline. For every occurrence of the word \"he's\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.59, 2.88, "one"], [6.31, 6.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.59,\"end\":2.88},{\"word\":\"two\",\"start\":6.31,\"end\":6.46}]"}
{"utterance_id": "7850-281318-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.92, 1.3, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":0.92,\"end\":1.3}]"}
{"utterance_id": "7976-110523-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.26, 5.63, "crust"]], "response_mode": "offline", "answer": "[{\"word\":\"crust\",\"start\":5.26,\"end\":5.63}]"}
{"utterance_id": "6295-244435-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0033.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Process the full audio offline. Find every instance of a city name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.42, 2.96, "winton"]], "response_mode": "offline", "answer": "[{\"word\":\"winton\",\"start\":2.42,\"end\":2.96}]"}
{"utterance_id": "2803-154328-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.13, "john"], [3.7, 4.26, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"john\",\"start\":1.74,\"end\":2.13},{\"word\":\"god\",\"start\":3.7,\"end\":4.26}]"}
{"utterance_id": "5694-64029-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 1.92, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.7,\"end\":1.92}]"}
{"utterance_id": "1988-24833-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0005.flac", "task_type": "lexical_detection", "query": "english", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"english\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.5, 3.86, "english"]], "response_mode": "offline", "answer": "[{\"word\":\"english\",\"start\":3.5,\"end\":3.86}]"}
{"utterance_id": "3853-163249-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.84, 7.4, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":6.84,\"end\":7.4}]"}
{"utterance_id": "1673-143397-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.64, "celestine"]], "response_mode": "offline", "answer": "[{\"word\":\"celestine\",\"start\":1.01,\"end\":1.64}]"}
{"utterance_id": "777-126732-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0003.flac", "task_type": "lexical_detection", "query": "ticket", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"ticket\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.4, 1.74, "ticket"]], "response_mode": "offline", "answer": "[{\"word\":\"ticket\",\"start\":1.4,\"end\":1.74}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "When the utterance is complete, report how a number accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 1.93, "one"], [3.12, 3.39, "two"], [8.81, 9.39, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.75,\"end\":1.93},{\"word\":\"two\",\"start\":3.12,\"end\":3.39},{\"word\":\"three\",\"start\":8.81,\"end\":9.39}]"}
{"utterance_id": "5895-34622-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0013.flac", "task_type": "lexical_detection", "query": "chariot", "prompt": "When the utterance is complete, report all instances of the word \"chariot\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 6.06, "chariot"]], "response_mode": "offline", "answer": "[{\"word\":\"chariot\",\"start\":5.52,\"end\":6.06}]"}
{"utterance_id": "5536-43359-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0018.flac", "task_type": "lexical_detection", "query": "bond", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"bond\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.42, 0.82, "bond"]], "response_mode": "offline", "answer": "[{\"word\":\"bond\",\"start\":0.42,\"end\":0.82}]"}
{"utterance_id": "2035-147961-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0031.flac", "task_type": "lexical_detection", "query": "crouching", "prompt": "Process the full audio offline. Find every instance of the word \"crouching\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.47, "crouching"]], "response_mode": "offline", "answer": "[{\"word\":\"crouching\",\"start\":0.84,\"end\":1.47}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.9, 1.43, "wines"], [4.15, 4.59, "wines"]], "response_mode": "offline", "answer": "[{\"word\":\"wines\",\"start\":0.9,\"end\":1.43},{\"word\":\"wines\",\"start\":4.15,\"end\":4.59}]"}
{"utterance_id": "3536-8226-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0002.flac", "task_type": "lexical_detection", "query": "favour", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"favour\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.58, 9.07, "favour"]], "response_mode": "offline", "answer": "[{\"word\":\"favour\",\"start\":8.58,\"end\":9.07}]"}
{"utterance_id": "7976-105575-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0008.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the full audio. After it finishes, find every instance of a time of day. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.34, 6.79, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":6.34,\"end\":6.79}]"}
{"utterance_id": "2086-149220-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0010.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "After the audio ends, tally every instance of an animal name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.29, 5.84, "one"], [12.89, 13.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.29,\"end\":5.84},{\"word\":\"two\",\"start\":12.89,\"end\":13.47}]"}
{"utterance_id": "2412-153948-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0001.flac", "task_type": "lexical_detection", "query": "succeed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"succeed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.2, "succeed"]], "response_mode": "offline", "answer": "[{\"word\":\"succeed\",\"start\":1.8,\"end\":2.2}]"}
{"utterance_id": "2803-154328-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0007.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 1.02, "one"], [2.41, 2.95, "two"], [3.98, 4.84, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.56,\"end\":1.02},{\"word\":\"two\",\"start\":2.41,\"end\":2.95},{\"word\":\"three\",\"start\":3.98,\"end\":4.84}]"}
{"utterance_id": "2428-83705-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0015.flac", "task_type": "lexical_detection", "query": "engagement", "prompt": "When the utterance is complete, report all instances of the word \"engagement\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.88, "engagement"]], "response_mode": "offline", "answer": "[{\"word\":\"engagement\",\"start\":1.34,\"end\":1.88}]"}
{"utterance_id": "3170-137482-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0048.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.29, 6.99, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":6.29,\"end\":6.99}]"}
{"utterance_id": "1272-141231-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0009.flac", "task_type": "lexical_detection", "query": "nights", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"nights\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.5, 5.92, "nights"]], "response_mode": "offline", "answer": "[{\"word\":\"nights\",\"start\":5.5,\"end\":5.92}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.27, 7.95, "liveryman"]], "response_mode": "offline", "answer": "[{\"word\":\"liveryman\",\"start\":7.27,\"end\":7.95}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After listening to the whole recording, output every instance of a plant name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.75, 7.2, "juniper"], [8.49, 9.23, "cedars"]], "response_mode": "offline", "answer": "[{\"word\":\"juniper\",\"start\":6.75,\"end\":7.2},{\"word\":\"cedars\",\"start\":8.49,\"end\":9.23}]"}
{"utterance_id": "1988-24833-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0003.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Listen to the full audio. Count every occurrence of an animal name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.89, "one"], [4.03, 4.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.29,\"end\":1.89},{\"word\":\"two\",\"start\":4.03,\"end\":4.31}]"}
{"utterance_id": "5895-34615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 1.09, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.52,\"end\":1.09}]"}
{"utterance_id": "6313-66125-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.3, 1.6, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":1.3,\"end\":1.6}]"}
{"utterance_id": "3536-8226-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[9.87, 10.5, "naples"]], "response_mode": "offline", "answer": "[{\"word\":\"naples\",\"start\":9.87,\"end\":10.5}]"}
{"utterance_id": "1993-147149-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0025.flac", "task_type": "lexical_detection", "query": "else", "prompt": "When the utterance is complete, report all instances of the word \"else\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.53, 5.73, "else"]], "response_mode": "offline", "answer": "[{\"word\":\"else\",\"start\":5.53,\"end\":5.73}]"}
{"utterance_id": "7850-281318-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 3.19, "jackdaw"]], "response_mode": "offline", "answer": "[{\"word\":\"jackdaw\",\"start\":2.52,\"end\":3.19}]"}
{"utterance_id": "2277-149874-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.86, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":0.47,\"end\":0.86}]"}
{"utterance_id": "2277-149874-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0016.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Listen to the full audio. After it finishes, find every instance of a money or currency term. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.52, 3.88, "dollars"]], "response_mode": "offline", "answer": "[{\"word\":\"dollars\",\"start\":3.52,\"end\":3.88}]"}
{"utterance_id": "2277-149897-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0008.flac", "task_type": "lexical_detection", "query": "guess", "prompt": "Once the audio has finished, identify every instance of the word \"guess\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.88, 2.12, "guess"]], "response_mode": "offline", "answer": "[{\"word\":\"guess\",\"start\":1.88,\"end\":2.12}]"}
{"utterance_id": "8297-275155-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0018.flac", "task_type": "lexical_detection", "query": "concluded", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"concluded\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.64, 2.57, "concluded"]], "response_mode": "offline", "answer": "[{\"word\":\"concluded\",\"start\":1.64,\"end\":2.57}]"}
{"utterance_id": "3000-15664-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0018.flac", "task_type": "semantic_detection", "query": "color", "prompt": "You will hear a complete utterance. When it ends, list every instance of a color name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 0.97, "gray"]], "response_mode": "offline", "answer": "[{\"word\":\"gray\",\"start\":0.73,\"end\":0.97}]"}
{"utterance_id": "6295-64301-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0025.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.91, 2.22, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":1.91,\"end\":2.22}]"}
{"utterance_id": "3752-4944-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0014.flac", "task_type": "lexical_detection", "query": "put", "prompt": "After the audio ends, extract all occurrences of the word \"put\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.62, "put"]], "response_mode": "offline", "answer": "[{\"word\":\"put\",\"start\":1.43,\"end\":1.62}]"}
{"utterance_id": "1272-135031-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.18, 1.55, "ann"]], "response_mode": "offline", "answer": "[{\"word\":\"ann\",\"start\":1.18,\"end\":1.55}]"}
{"utterance_id": "6313-76958-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0021.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.43, 5.85, "cowmen"]], "response_mode": "offline", "answer": "[{\"word\":\"cowmen\",\"start\":5.43,\"end\":5.85}]"}
{"utterance_id": "174-50561-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.41, "gillian"]], "response_mode": "offline", "answer": "[{\"word\":\"gillian\",\"start\":1.89,\"end\":2.41}]"}
{"utterance_id": "8842-302201-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.66, 3.92, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":3.66,\"end\":3.92}]"}
{"utterance_id": "2035-147961-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0027.flac", "task_type": "lexical_detection", "query": "yes", "prompt": "Once the audio has finished, identify every instance of the word \"yes\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.46, 0.94, "yes"]], "response_mode": "offline", "answer": "[{\"word\":\"yes\",\"start\":0.46,\"end\":0.94}]"}
{"utterance_id": "6345-93306-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.16, 4.52, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.16,\"end\":4.52}]"}
{"utterance_id": "2078-142845-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0023.flac", "task_type": "lexical_detection", "query": "hot", "prompt": "Process the full audio offline. Find every instance of the word \"hot\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.83, "hot"]], "response_mode": "offline", "answer": "[{\"word\":\"hot\",\"start\":0.51,\"end\":0.83}]"}
{"utterance_id": "3576-138058-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0035.flac", "task_type": "lexical_detection", "query": "salvation", "prompt": "After the audio ends, extract all occurrences of the word \"salvation\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.03, 5.7, "salvation"]], "response_mode": "offline", "answer": "[{\"word\":\"salvation\",\"start\":5.03,\"end\":5.7}]"}
{"utterance_id": "6319-64726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0000.flac", "task_type": "lexical_counting", "query": "gave", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"gave\" (word, start, end). If none, reply [].", "labels": [[1.79, 2.08, "one"], [5.29, 5.5, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.79,\"end\":2.08},{\"word\":\"two\",\"start\":5.29,\"end\":5.5}]"}
{"utterance_id": "6319-64726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0018.flac", "task_type": "lexical_detection", "query": "splendour", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"splendour\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.25, 6.0, "splendour"]], "response_mode": "offline", "answer": "[{\"word\":\"splendour\",\"start\":5.25,\"end\":6.0}]"}
{"utterance_id": "5536-43363-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0003.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.44, 7.7, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":7.44,\"end\":7.7}]"}
{"utterance_id": "6295-244435-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 1.69, "colonel leonidas talbot"]], "response_mode": "offline", "answer": "[{\"word\":\"colonel leonidas talbot\",\"start\":0.26,\"end\":1.69}]"}
{"utterance_id": "6241-61943-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0026.flac", "task_type": "lexical_detection", "query": "inhabitants", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"inhabitants\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.24, 1.83, "inhabitants"]], "response_mode": "offline", "answer": "[{\"word\":\"inhabitants\",\"start\":1.24,\"end\":1.83}]"}
{"utterance_id": "2902-9006-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0009.flac", "task_type": "lexical_detection", "query": "creed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"creed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.2, 3.57, "creed"]], "response_mode": "offline", "answer": "[{\"word\":\"creed\",\"start\":3.2,\"end\":3.57}]"}
{"utterance_id": "1462-170142-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.66, 6.16, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":5.66,\"end\":6.16}]"}
{"utterance_id": "652-130737-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0001.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.81, 2.12, "white"], [4.95, 5.21, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.81,\"end\":2.12},{\"word\":\"white\",\"start\":4.95,\"end\":5.21}]"}
{"utterance_id": "8842-302203-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0000.flac", "task_type": "lexical_counting", "query": "thou", "prompt": "After the audio ends, tally every instance of the word \"thou\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.48, 5.63, "one"], [10.3, 10.51, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.48,\"end\":5.63},{\"word\":\"two\",\"start\":10.3,\"end\":10.51}]"}
{"utterance_id": "174-168635-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.07, 6.73, "jean valjean"]], "response_mode": "offline", "answer": "[{\"word\":\"jean valjean\",\"start\":6.07,\"end\":6.73}]"}
{"utterance_id": "2277-149897-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0002.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "You will hear a complete utterance. When it ends, list every instance of a time of day you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.31, "evening"], [5.81, 6.27, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":3.86,\"end\":4.31},{\"word\":\"morning\",\"start\":5.81,\"end\":6.27}]"}
{"utterance_id": "6313-66129-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.18, 3.66, "ponies"]], "response_mode": "offline", "answer": "[{\"word\":\"ponies\",\"start\":3.18,\"end\":3.66}]"}
{"utterance_id": "5694-64029-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0010.flac", "task_type": "lexical_detection", "query": "motive", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"motive\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.33, "motive"]], "response_mode": "offline", "answer": "[{\"word\":\"motive\",\"start\":0.91,\"end\":1.33}]"}
{"utterance_id": "1993-147965-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0002.flac", "task_type": "lexical_detection", "query": "prayed", "prompt": "When the utterance is complete, report all instances of the word \"prayed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 3.74, "prayed"]], "response_mode": "offline", "answer": "[{\"word\":\"prayed\",\"start\":3.48,\"end\":3.74}]"}
{"utterance_id": "1462-170142-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0009.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.24, 1.83, "one"], [3.09, 3.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.24,\"end\":1.83},{\"word\":\"two\",\"start\":3.09,\"end\":3.45}]"}
{"utterance_id": "5895-34629-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.99, 4.67, "mountebank"]], "response_mode": "offline", "answer": "[{\"word\":\"mountebank\",\"start\":3.99,\"end\":4.67}]"}
{"utterance_id": "3853-163249-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.29, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.02,\"end\":2.29}]"}
{"utterance_id": "5536-43358-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0019.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 1.15, "historians"]], "response_mode": "offline", "answer": "[{\"word\":\"historians\",\"start\":0.55,\"end\":1.15}]"}
{"utterance_id": "5338-284437-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.21, 4.47, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":4.21,\"end\":4.47}]"}
{"utterance_id": "5895-34629-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0031.flac", "task_type": "lexical_detection", "query": "pity", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pity\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.51, "pity"]], "response_mode": "offline", "answer": "[{\"word\":\"pity\",\"start\":0.95,\"end\":1.51}]"}
{"utterance_id": "8842-302201-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0003.flac", "task_type": "lexical_detection", "query": "pleased", "prompt": "Once the audio has finished, identify every instance of the word \"pleased\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.08, 1.5, "pleased"]], "response_mode": "offline", "answer": "[{\"word\":\"pleased\",\"start\":1.08,\"end\":1.5}]"}
{"utterance_id": "1988-24833-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0028.flac", "task_type": "lexical_detection", "query": "happy", "prompt": "After listening to the whole recording, output every instance of the word \"happy\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.12, "happy"]], "response_mode": "offline", "answer": "[{\"word\":\"happy\",\"start\":1.81,\"end\":2.12}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.47, 10.8, "one"], [13.77, 14.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.47,\"end\":10.8},{\"word\":\"two\",\"start\":13.77,\"end\":14.69}]"}
{"utterance_id": "6295-244435-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0012.flac", "task_type": "lexical_detection", "query": "europe", "prompt": "After the audio ends, extract all occurrences of the word \"europe\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.24, 0.75, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"europe\",\"start\":0.24,\"end\":0.75}]"}
{"utterance_id": "174-168635-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.67, 1.08, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":0.67,\"end\":1.08}]"}
{"utterance_id": "1272-135031-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 1.86, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.63,\"end\":1.86}]"}
{"utterance_id": "8842-302201-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0010.flac", "task_type": "lexical_detection", "query": "divided", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"divided\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.19, 1.57, "divided"]], "response_mode": "offline", "answer": "[{\"word\":\"divided\",\"start\":1.19,\"end\":1.57}]"}
{"utterance_id": "652-129742-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0015.flac", "task_type": "lexical_counting", "query": "two", "prompt": "When the utterance is complete, report how the word \"two\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.34, "one"], [4.07, 4.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.34},{\"word\":\"two\",\"start\":4.07,\"end\":4.27}]"}
{"utterance_id": "8297-275155-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0032.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.03, 7.67, "sydney"], [8.86, 9.27, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"sydney\",\"start\":7.03,\"end\":7.67},{\"word\":\"london\",\"start\":8.86,\"end\":9.27}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "semantic_detection", "query": "money", "prompt": "You will hear a complete utterance. When it ends, list every instance of a money or currency term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.39, 3.9, "crowns"]], "response_mode": "offline", "answer": "[{\"word\":\"crowns\",\"start\":3.39,\"end\":3.9}]"}
{"utterance_id": "2078-142845-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0040.flac", "task_type": "semantic_counting", "query": "food", "prompt": "When the utterance is complete, report how a food name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.75, "one"], [6.07, 6.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.05,\"end\":2.75},{\"word\":\"two\",\"start\":6.07,\"end\":6.79}]"}
{"utterance_id": "251-137823-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.57, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":1.27,\"end\":1.57}]"}
{"utterance_id": "5694-64038-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0017.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Once the audio has finished, list each occurrence of a body part as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.06, 1.47, "one"], [3.7, 4.1, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.06,\"end\":1.47},{\"word\":\"two\",\"start\":3.7,\"end\":4.1}]"}
{"utterance_id": "1272-141231-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0005.flac", "task_type": "lexical_detection", "query": "measure", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"measure\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.31, "measure"]], "response_mode": "offline", "answer": "[{\"word\":\"measure\",\"start\":2.0,\"end\":2.31}]"}
{"utterance_id": "6313-66125-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.04, 1.28, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":1.04,\"end\":1.28}]"}
{"utterance_id": "3752-4943-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0028.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.82, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.46,\"end\":0.82}]"}
{"utterance_id": "7850-73752-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.37, 3.59, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":3.37,\"end\":3.59}]"}
{"utterance_id": "1272-135031-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0018.flac", "task_type": "lexical_detection", "query": "didn't", "prompt": "Once the audio has finished, identify every instance of the word \"didn't\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.65, 2.05, "didn't"]], "response_mode": "offline", "answer": "[{\"word\":\"didn't\",\"start\":1.65,\"end\":2.05}]"}
{"utterance_id": "652-130726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0007.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.96, "poodle dog"]], "response_mode": "offline", "answer": "[{\"word\":\"poodle dog\",\"start\":0.36,\"end\":0.96}]"}
{"utterance_id": "2803-161169-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.99, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":1.44,\"end\":1.99}]"}
{"utterance_id": "2412-153954-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0011.flac", "task_type": "lexical_detection", "query": "expressed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"expressed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 1.08, "expressed"]], "response_mode": "offline", "answer": "[{\"word\":\"expressed\",\"start\":0.49,\"end\":1.08}]"}
{"utterance_id": "6295-244435-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.27, 0.6, "lincoln"], [3.8, 4.08, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"lincoln\",\"start\":0.27,\"end\":0.6},{\"word\":\"harry\",\"start\":3.8,\"end\":4.08}]"}
{"utterance_id": "777-126732-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0048.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.55, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.26,\"end\":2.55}]"}
{"utterance_id": "251-118436-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0016.flac", "task_type": "semantic_detection", "query": "color", "prompt": "You will hear a complete utterance. When it ends, list every instance of a color name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.56, 6.18, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":5.56,\"end\":6.18}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "lexical_counting", "query": "empire", "prompt": "When the utterance is complete, report how the word \"empire\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.67, 9.21, "one"], [14.03, 14.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.67,\"end\":9.21},{\"word\":\"two\",\"start\":14.03,\"end\":14.48}]"}
{"utterance_id": "7850-286674-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0001.flac", "task_type": "lexical_detection", "query": "disgraceful", "prompt": "After listening to the whole recording, output every instance of the word \"disgraceful\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 2.51, "disgraceful"]], "response_mode": "offline", "answer": "[{\"word\":\"disgraceful\",\"start\":1.02,\"end\":2.51}]"}
{"utterance_id": "6313-76958-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0010.flac", "task_type": "lexical_detection", "query": "doubled", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"doubled\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.29, 5.59, "doubled"]], "response_mode": "offline", "answer": "[{\"word\":\"doubled\",\"start\":5.29,\"end\":5.59}]"}
{"utterance_id": "2277-149896-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0030.flac", "task_type": "lexical_detection", "query": "money", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"money\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.35, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":3.97,\"end\":4.35}]"}
{"utterance_id": "251-118436-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0005.flac", "task_type": "lexical_detection", "query": "broad", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"broad\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.49, 1.85, "broad"]], "response_mode": "offline", "answer": "[{\"word\":\"broad\",\"start\":1.49,\"end\":1.85}]"}
{"utterance_id": "3853-163249-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0035.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.78, 3.23, "breast"]], "response_mode": "offline", "answer": "[{\"word\":\"breast\",\"start\":2.78,\"end\":3.23}]"}
{"utterance_id": "2412-153948-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0004.flac", "task_type": "lexical_detection", "query": "bred", "prompt": "After listening to the whole recording, output every instance of the word \"bred\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.22, "bred"]], "response_mode": "offline", "answer": "[{\"word\":\"bred\",\"start\":2.02,\"end\":2.22}]"}
{"utterance_id": "2277-149897-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0022.flac", "task_type": "lexical_detection", "query": "ability", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"ability\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.57, 8.0, "ability"]], "response_mode": "offline", "answer": "[{\"word\":\"ability\",\"start\":7.57,\"end\":8.0}]"}
{"utterance_id": "3536-23268-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.14, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss milner\",\"start\":2.54,\"end\":3.14}]"}
{"utterance_id": "2078-142845-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0012.flac", "task_type": "semantic_detection", "query": "country", "prompt": "You will hear a complete utterance. When it ends, list every instance of a country name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.19, "italy"]], "response_mode": "offline", "answer": "[{\"word\":\"italy\",\"start\":1.72,\"end\":2.19}]"}
{"utterance_id": "2086-149220-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0010.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.29, 5.84, "chicken"], [12.89, 13.47, "fowls"]], "response_mode": "offline", "answer": "[{\"word\":\"chicken\",\"start\":5.29,\"end\":5.84},{\"word\":\"fowls\",\"start\":12.89,\"end\":13.47}]"}
{"utterance_id": "6295-244435-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0018.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.41, 2.79, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.41,\"end\":2.79}]"}
{"utterance_id": "1272-128104-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 3.24, "sir frederick leighton's"]], "response_mode": "offline", "answer": "[{\"word\":\"sir frederick leighton's\",\"start\":2.18,\"end\":3.24}]"}
{"utterance_id": "84-121550-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.06, 3.36, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"blood\",\"start\":3.06,\"end\":3.36}]"}
{"utterance_id": "5895-34629-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0021.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the full audio. After it finishes, find every instance of a disease name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.02, 6.58, "sickness"]], "response_mode": "offline", "answer": "[{\"word\":\"sickness\",\"start\":6.02,\"end\":6.58}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.23, 6.82, "two thirds"], [13.3, 13.52, "few"]], "response_mode": "offline", "answer": "[{\"word\":\"two thirds\",\"start\":6.23,\"end\":6.82},{\"word\":\"few\",\"start\":13.3,\"end\":13.52}]"}
{"utterance_id": "5895-34622-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[14.6, 15.26, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":14.6,\"end\":15.26}]"}
{"utterance_id": "5694-64029-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0019.flac", "task_type": "lexical_detection", "query": "yankee", "prompt": "Process the full audio offline. Find every instance of the word \"yankee\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.51, "yankee"]], "response_mode": "offline", "answer": "[{\"word\":\"yankee\",\"start\":1.15,\"end\":1.51}]"}
{"utterance_id": "84-121550-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.21, 5.68, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":5.21,\"end\":5.68}]"}
{"utterance_id": "84-121550-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0031.flac", "task_type": "lexical_detection", "query": "age", "prompt": "After the audio ends, extract all occurrences of the word \"age\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.49, "age"]], "response_mode": "offline", "answer": "[{\"word\":\"age\",\"start\":2.2,\"end\":2.49}]"}
{"utterance_id": "652-129742-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0000.flac", "task_type": "lexical_counting", "query": "asparagus", "prompt": "After the audio ends, tally every instance of the word \"asparagus\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.51, 1.13, "one"], [2.89, 3.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.13},{\"word\":\"two\",\"start\":2.89,\"end\":3.59}]"}
{"utterance_id": "1272-135031-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0021.flac", "task_type": "lexical_detection", "query": "know", "prompt": "When the utterance is complete, report all instances of the word \"know\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.38, "know"]], "response_mode": "offline", "answer": "[{\"word\":\"know\",\"start\":0.98,\"end\":1.38}]"}
{"utterance_id": "8297-275155-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0028.flac", "task_type": "lexical_detection", "query": "hotel", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hotel\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.97, "hotel"]], "response_mode": "offline", "answer": "[{\"word\":\"hotel\",\"start\":1.12,\"end\":1.97}]"}
{"utterance_id": "3576-138058-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0024.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Once the audio has finished, identify every instance of a weather term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[13.99, 14.46, "winds"]], "response_mode": "offline", "answer": "[{\"word\":\"winds\",\"start\":13.99,\"end\":14.46}]"}
{"utterance_id": "7976-105575-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0014.flac", "task_type": "semantic_counting", "query": "clock_time", "prompt": "When the utterance is complete, report how a time of day accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.24, "one"], [6.41, 7.07, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.84,\"end\":1.24},{\"word\":\"two\",\"start\":6.41,\"end\":7.07}]"}
{"utterance_id": "6345-64257-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0009.flac", "task_type": "lexical_detection", "query": "vessels", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"vessels\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.66, 5.04, "vessels"]], "response_mode": "offline", "answer": "[{\"word\":\"vessels\",\"start\":4.66,\"end\":5.04}]"}
{"utterance_id": "2277-149897-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0006.flac", "task_type": "lexical_detection", "query": "harry", "prompt": "After the audio ends, extract all occurrences of the word \"harry\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.44, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":1.03,\"end\":1.44}]"}
{"utterance_id": "7976-110523-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.53, 7.93, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":7.53,\"end\":7.93}]"}
{"utterance_id": "2428-83699-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.31, 5.83, "madge"]], "response_mode": "offline", "answer": "[{\"word\":\"madge\",\"start\":5.31,\"end\":5.83}]"}
{"utterance_id": "8842-302203-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0007.flac", "task_type": "lexical_detection", "query": "phantasies", "prompt": "Process the full audio offline. Find every instance of the word \"phantasies\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.74, "phantasies"]], "response_mode": "offline", "answer": "[{\"word\":\"phantasies\",\"start\":1.1,\"end\":1.74}]"}
{"utterance_id": "251-118436-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0022.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.68, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":3.24,\"end\":3.68}]"}
{"utterance_id": "2035-152373-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0005.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "When the utterance is complete, report all instances of a disease name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.59, 13.06, "plague"]], "response_mode": "offline", "answer": "[{\"word\":\"plague\",\"start\":12.59,\"end\":13.06}]"}
{"utterance_id": "652-130726-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0022.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[11.65, 12.17, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":11.65,\"end\":12.17}]"}
{"utterance_id": "2086-149220-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.44, 0.73, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":0.44,\"end\":0.73}]"}
{"utterance_id": "5536-43358-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.94, 8.27, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":7.94,\"end\":8.27}]"}
{"utterance_id": "1462-170142-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0005.flac", "task_type": "lexical_detection", "query": "threshold", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"threshold\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.31, "threshold"]], "response_mode": "offline", "answer": "[{\"word\":\"threshold\",\"start\":1.64,\"end\":2.31}]"}
{"utterance_id": "1272-135031-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0000.flac", "task_type": "lexical_detection", "query": "fiddle", "prompt": "After the audio ends, extract all occurrences of the word \"fiddle\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.39, 5.92, "fiddle"]], "response_mode": "offline", "answer": "[{\"word\":\"fiddle\",\"start\":5.39,\"end\":5.92}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.05, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":2.54,\"end\":3.05}]"}
{"utterance_id": "7976-110523-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0015.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[0.4, 0.75, "one"], [3.82, 4.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.4,\"end\":0.75},{\"word\":\"two\",\"start\":3.82,\"end\":4.32}]"}
{"utterance_id": "5895-34615-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0003.flac", "task_type": "lexical_detection", "query": "industrious", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"industrious\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.53, 3.3, "industrious"]], "response_mode": "offline", "answer": "[{\"word\":\"industrious\",\"start\":2.53,\"end\":3.3}]"}
{"utterance_id": "7976-110523-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0001.flac", "task_type": "lexical_detection", "query": "feed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"feed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.07, "feed"]], "response_mode": "offline", "answer": "[{\"word\":\"feed\",\"start\":0.78,\"end\":1.07}]"}
{"utterance_id": "1462-170145-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 1.96, "arm"]], "response_mode": "offline", "answer": "[{\"word\":\"arm\",\"start\":1.66,\"end\":1.96}]"}
{"utterance_id": "1993-147149-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.35, 3.71, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":3.35,\"end\":3.71}]"}
{"utterance_id": "174-50561-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0011.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.27, 6.6, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":6.27,\"end\":6.6}]"}
{"utterance_id": "1462-170138-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0012.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a person's name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.5, 1.12, "one"], [2.24, 3.63, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.5,\"end\":1.12},{\"word\":\"two\",\"start\":2.24,\"end\":3.63}]"}
{"utterance_id": "1272-141231-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0029.flac", "task_type": "lexical_detection", "query": "energy", "prompt": "Process the full audio offline. Find every instance of the word \"energy\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.83, "energy"]], "response_mode": "offline", "answer": "[{\"word\":\"energy\",\"start\":2.23,\"end\":2.83}]"}
{"utterance_id": "7850-281318-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0008.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.06, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":1.61,\"end\":2.06}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.56, 5.99, "lily"]], "response_mode": "offline", "answer": "[{\"word\":\"lily\",\"start\":5.56,\"end\":5.99}]"}
{"utterance_id": "251-118436-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0015.flac", "task_type": "lexical_detection", "query": "cord", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"cord\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.29, 2.78, "cord"]], "response_mode": "offline", "answer": "[{\"word\":\"cord\",\"start\":2.29,\"end\":2.78}]"}
{"utterance_id": "2078-142845-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 2.0, "seventeen eighteen"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen eighteen\",\"start\":0.42,\"end\":2.0}]"}
{"utterance_id": "6345-64257-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0012.flac", "task_type": "lexical_detection", "query": "living", "prompt": "After listening to the whole recording, output every instance of the word \"living\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.77, "living"]], "response_mode": "offline", "answer": "[{\"word\":\"living\",\"start\":0.43,\"end\":0.77}]"}
{"utterance_id": "5694-64029-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0032.flac", "task_type": "lexical_detection", "query": "reported", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"reported\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.0, "reported"]], "response_mode": "offline", "answer": "[{\"word\":\"reported\",\"start\":1.55,\"end\":2.0}]"}
{"utterance_id": "3853-163249-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0042.flac", "task_type": "lexical_detection", "query": "support", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"support\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[12.01, 12.49, "support"]], "response_mode": "offline", "answer": "[{\"word\":\"support\",\"start\":12.01,\"end\":12.49}]"}
{"utterance_id": "777-126732-0081", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0081.flac", "task_type": "lexical_detection", "query": "comfortable", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"comfortable\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 1.03, "comfortable"]], "response_mode": "offline", "answer": "[{\"word\":\"comfortable\",\"start\":0.53,\"end\":1.03}]"}
{"utterance_id": "5694-64038-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0004.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Process the full audio offline. Find every instance of a country name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.53, "united states"]], "response_mode": "offline", "answer": "[{\"word\":\"united states\",\"start\":0.84,\"end\":1.53}]"}
{"utterance_id": "6319-275224-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0011.flac", "task_type": "lexical_detection", "query": "pole", "prompt": "Process the full audio offline. Find every instance of the word \"pole\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.48, 5.9, "pole"]], "response_mode": "offline", "answer": "[{\"word\":\"pole\",\"start\":5.48,\"end\":5.9}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.01, 4.46, "bay"]], "response_mode": "offline", "answer": "[{\"word\":\"bay\",\"start\":4.01,\"end\":4.46}]"}
{"utterance_id": "5694-64029-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0012.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an organization name with word/start/end. If none, reply [].", "labels": [[3.98, 4.35, "one"], [6.56, 7.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.98,\"end\":4.35},{\"word\":\"two\",\"start\":6.56,\"end\":7.01}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.87, 13.26, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":12.87,\"end\":13.26}]"}
{"utterance_id": "7976-110124-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.36, 2.65, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.36,\"end\":2.65}]"}
{"utterance_id": "7850-73752-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0001.flac", "task_type": "lexical_detection", "query": "bursts", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"bursts\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.07, 3.42, "bursts"]], "response_mode": "offline", "answer": "[{\"word\":\"bursts\",\"start\":3.07,\"end\":3.42}]"}
{"utterance_id": "2412-153947-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0016.flac", "task_type": "lexical_detection", "query": "probably", "prompt": "After listening to the whole recording, output every instance of the word \"probably\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 2.91, "probably"]], "response_mode": "offline", "answer": "[{\"word\":\"probably\",\"start\":2.52,\"end\":2.91}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.13, 3.5, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.13,\"end\":3.5}]"}
{"utterance_id": "1993-147964-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.79, 7.15, "jake"], [13.12, 13.48, "otto"]], "response_mode": "offline", "answer": "[{\"word\":\"jake\",\"start\":6.79,\"end\":7.15},{\"word\":\"otto\",\"start\":13.12,\"end\":13.48}]"}
{"utterance_id": "84-121123-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0001.flac", "task_type": "lexical_detection", "query": "less", "prompt": "When the utterance is complete, report all instances of the word \"less\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.73, "less"]], "response_mode": "offline", "answer": "[{\"word\":\"less\",\"start\":0.53,\"end\":0.73}]"}
{"utterance_id": "3752-4943-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0026.flac", "task_type": "lexical_detection", "query": "stop", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"stop\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.16, 3.55, "stop"]], "response_mode": "offline", "answer": "[{\"word\":\"stop\",\"start\":3.16,\"end\":3.55}]"}
{"utterance_id": "2803-154328-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0011.flac", "task_type": "lexical_detection", "query": "forget", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"forget\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.47, 2.11, "forget"]], "response_mode": "offline", "answer": "[{\"word\":\"forget\",\"start\":1.47,\"end\":2.11}]"}
{"utterance_id": "6313-66129-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.18, 2.6, "stacy"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":2.18,\"end\":2.6}]"}
{"utterance_id": "8297-275155-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.82, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":0.5,\"end\":0.82}]"}
{"utterance_id": "7976-110124-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0015.flac", "task_type": "lexical_detection", "query": "doors", "prompt": "Process the full audio offline. Find every instance of the word \"doors\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.34, 3.73, "doors"]], "response_mode": "offline", "answer": "[{\"word\":\"doors\",\"start\":3.34,\"end\":3.73}]"}
{"utterance_id": "1993-147149-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the full audio. After it finishes, find every instance of a disease name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.87, "fever"]], "response_mode": "offline", "answer": "[{\"word\":\"fever\",\"start\":1.31,\"end\":1.87}]"}
{"utterance_id": "2277-149897-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.42, 4.84, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":4.42,\"end\":4.84}]"}
{"utterance_id": "2428-83705-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0037.flac", "task_type": "lexical_detection", "query": "pretend", "prompt": "Once the audio has finished, identify every instance of the word \"pretend\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.02, 1.4, "pretend"]], "response_mode": "offline", "answer": "[{\"word\":\"pretend\",\"start\":1.02,\"end\":1.4}]"}
{"utterance_id": "1462-170142-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0010.flac", "task_type": "lexical_detection", "query": "tremulously", "prompt": "After the audio ends, extract all occurrences of the word \"tremulously\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.55, 4.36, "tremulously"]], "response_mode": "offline", "answer": "[{\"word\":\"tremulously\",\"start\":3.55,\"end\":4.36}]"}
{"utterance_id": "6345-93306-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0019.flac", "task_type": "lexical_detection", "query": "stopped", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"stopped\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.04, 7.61, "stopped"]], "response_mode": "offline", "answer": "[{\"word\":\"stopped\",\"start\":7.04,\"end\":7.61}]"}
{"utterance_id": "5694-64025-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.16, 3.6, "sheep"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":3.16,\"end\":3.6}]"}
{"utterance_id": "422-122949-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0011.flac", "task_type": "lexical_detection", "query": "victorious", "prompt": "Process the full audio offline. Find every instance of the word \"victorious\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.11, 10.97, "victorious"]], "response_mode": "offline", "answer": "[{\"word\":\"victorious\",\"start\":10.11,\"end\":10.97}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.54, 8.46, "stoke claypole"]], "response_mode": "offline", "answer": "[{\"word\":\"stoke claypole\",\"start\":7.54,\"end\":8.46}]"}
{"utterance_id": "6295-64301-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.82, 2.46, "joseph"]], "response_mode": "offline", "answer": "[{\"word\":\"joseph\",\"start\":1.82,\"end\":2.46}]"}
{"utterance_id": "174-168635-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0009.flac", "task_type": "lexical_detection", "query": "meet", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"meet\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.04, "meet"]], "response_mode": "offline", "answer": "[{\"word\":\"meet\",\"start\":0.69,\"end\":1.04}]"}
{"utterance_id": "1272-141231-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0026.flac", "task_type": "lexical_detection", "query": "breathing", "prompt": "Once the audio has finished, identify every instance of the word \"breathing\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.51, 0.92, "breathing"]], "response_mode": "offline", "answer": "[{\"word\":\"breathing\",\"start\":0.51,\"end\":0.92}]"}
{"utterance_id": "3576-138058-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.9, "sancho"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho\",\"start\":0.51,\"end\":0.9}]"}
{"utterance_id": "777-126732-0070", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0070.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[0.96, 1.36, "mind"], [3.36, 3.88, "flesh"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":0.96,\"end\":1.36},{\"word\":\"flesh\",\"start\":3.36,\"end\":3.88}]"}
{"utterance_id": "8842-302201-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0013.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.97, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.45,\"end\":0.97}]"}
{"utterance_id": "3081-166546-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0062.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.88, 2.94, "assistant manager"]], "response_mode": "offline", "answer": "[{\"word\":\"assistant manager\",\"start\":1.88,\"end\":2.94}]"}
{"utterance_id": "2277-149896-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0015.flac", "task_type": "lexical_detection", "query": "examined", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"examined\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.5, "examined"]], "response_mode": "offline", "answer": "[{\"word\":\"examined\",\"start\":1.02,\"end\":1.5}]"}
{"utterance_id": "6319-64726-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.27, 6.87, "shoulders"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulders\",\"start\":6.27,\"end\":6.87}]"}
{"utterance_id": "7976-105575-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0012.flac", "task_type": "lexical_detection", "query": "mississippi", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"mississippi\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.92, 4.68, "mississippi"]], "response_mode": "offline", "answer": "[{\"word\":\"mississippi\",\"start\":3.92,\"end\":4.68}]"}
{"utterance_id": "6313-76958-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0015.flac", "task_type": "lexical_detection", "query": "stacy", "prompt": "After the audio ends, extract all occurrences of the word \"stacy\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.67, "stacy"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":0.28,\"end\":0.67}]"}
{"utterance_id": "652-130726-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0014.flac", "task_type": "lexical_detection", "query": "changed", "prompt": "Process the full audio offline. Find every instance of the word \"changed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.91, "changed"]], "response_mode": "offline", "answer": "[{\"word\":\"changed\",\"start\":0.46,\"end\":0.91}]"}
{"utterance_id": "652-130737-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0007.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.31, 1.08, "one"], [3.14, 3.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.31,\"end\":1.08},{\"word\":\"two\",\"start\":3.14,\"end\":3.54}]"}
{"utterance_id": "2412-153947-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0014.flac", "task_type": "lexical_counting", "query": "made", "prompt": "Process the full audio offline. For every occurrence of the word \"made\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 0.81, "one"], [11.97, 12.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.61,\"end\":0.81},{\"word\":\"two\",\"start\":11.97,\"end\":12.32}]"}
{"utterance_id": "7976-110124-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.27, "daughters"]], "response_mode": "offline", "answer": "[{\"word\":\"daughters\",\"start\":2.65,\"end\":3.27}]"}
{"utterance_id": "3752-4944-0069", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0069.flac", "task_type": "lexical_detection", "query": "ludicrous", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"ludicrous\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.9, 8.66, "ludicrous"]], "response_mode": "offline", "answer": "[{\"word\":\"ludicrous\",\"start\":7.9,\"end\":8.66}]"}
{"utterance_id": "1993-147149-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.75, "hand"], [4.44, 4.92, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.42,\"end\":1.75},{\"word\":\"heart\",\"start\":4.44,\"end\":4.92}]"}
{"utterance_id": "5694-64029-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0013.flac", "task_type": "lexical_detection", "query": "formed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"formed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 1.99, "formed"]], "response_mode": "offline", "answer": "[{\"word\":\"formed\",\"start\":1.57,\"end\":1.99}]"}
{"utterance_id": "3000-15664-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0006.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.33, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":0.99,\"end\":1.33}]"}
{"utterance_id": "1919-142785-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0027.flac", "task_type": "lexical_detection", "query": "balls", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"balls\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.74, 9.25, "balls"]], "response_mode": "offline", "answer": "[{\"word\":\"balls\",\"start\":8.74,\"end\":9.25}]"}
{"utterance_id": "3576-138058-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0035.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.12, 10.15, "knight errant"]], "response_mode": "offline", "answer": "[{\"word\":\"knight errant\",\"start\":9.12,\"end\":10.15}]"}
{"utterance_id": "2277-149896-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0026.flac", "task_type": "lexical_detection", "query": "bottom", "prompt": "After the audio ends, extract all occurrences of the word \"bottom\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.12, 4.61, "bottom"]], "response_mode": "offline", "answer": "[{\"word\":\"bottom\",\"start\":4.12,\"end\":4.61}]"}
{"utterance_id": "1919-142785-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0019.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.06, 1.38, "first"], [3.82, 4.43, "seventeenth"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.06,\"end\":1.38},{\"word\":\"seventeenth\",\"start\":3.82,\"end\":4.43}]"}
{"utterance_id": "251-137823-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.3, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":2.84,\"end\":3.3}]"}
{"utterance_id": "5895-34629-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0026.flac", "task_type": "lexical_detection", "query": "clouds", "prompt": "Process the full audio offline. Find every instance of the word \"clouds\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.03, 11.74, "clouds"]], "response_mode": "offline", "answer": "[{\"word\":\"clouds\",\"start\":11.03,\"end\":11.74}]"}
{"utterance_id": "6345-64257-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0001.flac", "task_type": "lexical_counting", "query": "time", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"time\" (word, start, end). If none, reply [].", "labels": [[0.79, 1.23, "one"], [6.28, 6.61, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.79,\"end\":1.23},{\"word\":\"two\",\"start\":6.28,\"end\":6.61}]"}
{"utterance_id": "6313-66125-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.46, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":2.17,\"end\":2.46}]"}
{"utterance_id": "3000-15664-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0024.flac", "task_type": "lexical_detection", "query": "lava", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"lava\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.48, 7.81, "lava"]], "response_mode": "offline", "answer": "[{\"word\":\"lava\",\"start\":7.48,\"end\":7.81}]"}
{"utterance_id": "2078-142845-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.28, 7.89, "fly"]], "response_mode": "offline", "answer": "[{\"word\":\"fly\",\"start\":7.28,\"end\":7.89}]"}
{"utterance_id": "6319-64726-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.29, "faces"]], "response_mode": "offline", "answer": "[{\"word\":\"faces\",\"start\":0.78,\"end\":1.29}]"}
{"utterance_id": "174-50561-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0013.flac", "task_type": "lexical_counting", "query": "apple", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"apple\" (word, start, end). If none, reply [].", "labels": [[1.69, 2.05, "one"], [8.97, 9.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.69,\"end\":2.05},{\"word\":\"two\",\"start\":8.97,\"end\":9.32}]"}
{"utterance_id": "5895-34615-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 1.02, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.4,\"end\":1.02}]"}
{"utterance_id": "6295-64301-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0031.flac", "task_type": "lexical_detection", "query": "treasure", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"treasure\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.1, "treasure"]], "response_mode": "offline", "answer": "[{\"word\":\"treasure\",\"start\":0.69,\"end\":1.1}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.17, "aunt"], [7.27, 7.68, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"aunt\",\"start\":2.93,\"end\":3.17},{\"word\":\"father's\",\"start\":7.27,\"end\":7.68}]"}
{"utterance_id": "7976-110523-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.37, 4.76, "neck"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":4.37,\"end\":4.76}]"}
{"utterance_id": "1272-128104-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0006.flac", "task_type": "lexical_detection", "query": "writes", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"writes\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.19, 3.58, "writes"]], "response_mode": "offline", "answer": "[{\"word\":\"writes\",\"start\":3.19,\"end\":3.58}]"}
{"utterance_id": "5338-24640-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0009.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[9.1, 10.17, "seventeen o seven"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen o seven\",\"start\":9.1,\"end\":10.17}]"}
{"utterance_id": "6319-64726-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0014.flac", "task_type": "lexical_detection", "query": "waited", "prompt": "After the audio ends, extract all occurrences of the word \"waited\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.31, "waited"]], "response_mode": "offline", "answer": "[{\"word\":\"waited\",\"start\":0.86,\"end\":1.31}]"}
{"utterance_id": "1993-147964-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0007.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.58, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":2.26,\"end\":2.58}]"}
{"utterance_id": "2803-154328-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.14, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.85,\"end\":1.14}]"}
{"utterance_id": "2902-9008-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0000.flac", "task_type": "lexical_detection", "query": "song", "prompt": "Process the full audio offline. Find every instance of the word \"song\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.74, 4.26, "song"]], "response_mode": "offline", "answer": "[{\"word\":\"song\",\"start\":3.74,\"end\":4.26}]"}
{"utterance_id": "3576-138058-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.23, 8.79, "waist"]], "response_mode": "offline", "answer": "[{\"word\":\"waist\",\"start\":8.23,\"end\":8.79}]"}
{"utterance_id": "7976-110523-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.0, 1.35, "parents"]], "response_mode": "offline", "answer": "[{\"word\":\"parents\",\"start\":1.0,\"end\":1.35}]"}
{"utterance_id": "3853-163249-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0035.flac", "task_type": "lexical_detection", "query": "inconsistent", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"inconsistent\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.46, 1.23, "inconsistent"]], "response_mode": "offline", "answer": "[{\"word\":\"inconsistent\",\"start\":0.46,\"end\":1.23}]"}
{"utterance_id": "8842-302203-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0005.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[3.29, 3.5, "one"], [4.86, 5.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.29,\"end\":3.5},{\"word\":\"two\",\"start\":4.86,\"end\":5.18}]"}
{"utterance_id": "1919-142785-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0039.flac", "task_type": "lexical_detection", "query": "illustration", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"illustration\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.35, "illustration"]], "response_mode": "offline", "answer": "[{\"word\":\"illustration\",\"start\":0.52,\"end\":1.35}]"}
{"utterance_id": "2078-142845-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0027.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.91, "bread"], [3.29, 3.97, "toast"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":1.43,\"end\":1.91},{\"word\":\"toast\",\"start\":3.29,\"end\":3.97}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.77, 11.27, "engineer"]], "response_mode": "offline", "answer": "[{\"word\":\"engineer\",\"start\":10.77,\"end\":11.27}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "lexical_counting", "query": "second", "prompt": "Process the full audio offline. For every occurrence of the word \"second\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 0.83, "one"], [6.19, 6.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.59,\"end\":0.83},{\"word\":\"two\",\"start\":6.19,\"end\":6.57}]"}
{"utterance_id": "5895-34622-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.73, 1.27, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.73,\"end\":1.27}]"}
{"utterance_id": "6241-66616-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0007.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After listening to the whole recording, output timed count words for a person's name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.64, 2.95, "one"], [4.13, 4.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.64,\"end\":2.95},{\"word\":\"two\",\"start\":4.13,\"end\":4.83}]"}
{"utterance_id": "6295-244435-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0025.flac", "task_type": "lexical_detection", "query": "charleston", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"charleston\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.09, 1.83, "charleston"]], "response_mode": "offline", "answer": "[{\"word\":\"charleston\",\"start\":1.09,\"end\":1.83}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.12, 8.49, "one"], [20.54, 20.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.12,\"end\":8.49},{\"word\":\"two\",\"start\":20.54,\"end\":20.92}]"}
{"utterance_id": "7976-105575-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0017.flac", "task_type": "lexical_detection", "query": "help", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"help\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.18, "help"]], "response_mode": "offline", "answer": "[{\"word\":\"help\",\"start\":0.96,\"end\":1.18}]"}
{"utterance_id": "7976-105575-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0011.flac", "task_type": "lexical_detection", "query": "king", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"king\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.79, 6.23, "king"]], "response_mode": "offline", "answer": "[{\"word\":\"king\",\"start\":5.79,\"end\":6.23}]"}
{"utterance_id": "3752-4944-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0040.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.22, 0.64, "doctor"]], "response_mode": "offline", "answer": "[{\"word\":\"doctor\",\"start\":0.22,\"end\":0.64}]"}
{"utterance_id": "2412-153954-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0018.flac", "task_type": "lexical_detection", "query": "pulse", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"pulse\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.96, 1.43, "pulse"]], "response_mode": "offline", "answer": "[{\"word\":\"pulse\",\"start\":0.96,\"end\":1.43}]"}
{"utterance_id": "7850-286674-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 1.67, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":1.38,\"end\":1.67}]"}
{"utterance_id": "3081-166546-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0056.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.19, "next"]], "response_mode": "offline", "answer": "[{\"word\":\"next\",\"start\":0.67,\"end\":1.19}]"}
{"utterance_id": "5895-34615-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[5.71, 6.21, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":5.71,\"end\":6.21}]"}
{"utterance_id": "5694-64029-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0029.flac", "task_type": "lexical_detection", "query": "wearing", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"wearing\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.34, "wearing"]], "response_mode": "offline", "answer": "[{\"word\":\"wearing\",\"start\":1.87,\"end\":2.34}]"}
{"utterance_id": "1272-128104-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0009.flac", "task_type": "lexical_counting", "query": "art", "prompt": "Once the audio has finished, list each occurrence of the word \"art\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[4.83, 5.17, "one"], [14.13, 14.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.83,\"end\":5.17},{\"word\":\"two\",\"start\":14.13,\"end\":14.59}]"}
{"utterance_id": "3576-138058-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0040.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.27, 2.63, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.27,\"end\":2.63}]"}
{"utterance_id": "1988-148538-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0003.flac", "task_type": "lexical_detection", "query": "democratic", "prompt": "After listening to the whole recording, output every instance of the word \"democratic\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.84, 4.39, "democratic"]], "response_mode": "offline", "answer": "[{\"word\":\"democratic\",\"start\":3.84,\"end\":4.39}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.35, 6.64, "six"]], "response_mode": "offline", "answer": "[{\"word\":\"six\",\"start\":6.35,\"end\":6.64}]"}
{"utterance_id": "1993-147964-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.68, 0.91, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":0.68,\"end\":0.91}]"}
{"utterance_id": "5694-64025-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0001.flac", "task_type": "lexical_detection", "query": "engaged", "prompt": "After the audio ends, extract all occurrences of the word \"engaged\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.9, "engaged"]], "response_mode": "offline", "answer": "[{\"word\":\"engaged\",\"start\":3.15,\"end\":3.9}]"}
{"utterance_id": "2035-147961-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0026.flac", "task_type": "lexical_detection", "query": "village", "prompt": "Process the full audio offline. Find every instance of the word \"village\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.32, "village"]], "response_mode": "offline", "answer": "[{\"word\":\"village\",\"start\":1.93,\"end\":2.32}]"}
{"utterance_id": "3752-4944-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0009.flac", "task_type": "lexical_detection", "query": "pleasure", "prompt": "Process the full audio offline. Find every instance of the word \"pleasure\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 1.04, "pleasure"]], "response_mode": "offline", "answer": "[{\"word\":\"pleasure\",\"start\":0.68,\"end\":1.04}]"}
{"utterance_id": "2412-153947-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.87, 6.27, "printer"]], "response_mode": "offline", "answer": "[{\"word\":\"printer\",\"start\":5.87,\"end\":6.27}]"}
{"utterance_id": "2803-154320-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0007.flac", "task_type": "lexical_detection", "query": "meeting", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"meeting\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.76, "meeting"]], "response_mode": "offline", "answer": "[{\"word\":\"meeting\",\"start\":2.09,\"end\":2.76}]"}
{"utterance_id": "5694-64029-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0002.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.87, 1.15, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":0.87,\"end\":1.15}]"}
{"utterance_id": "1988-24833-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0005.flac", "task_type": "lexical_counting", "query": "music", "prompt": "Listen to the entire clip, then report the count sequence for the word \"music\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.94, "one"], [4.79, 5.4, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.47,\"end\":1.94},{\"word\":\"two\",\"start\":4.79,\"end\":5.4}]"}
{"utterance_id": "3752-4944-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0000.flac", "task_type": "lexical_detection", "query": "captain", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"captain\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 0.77, "captain"]], "response_mode": "offline", "answer": "[{\"word\":\"captain\",\"start\":0.37,\"end\":0.77}]"}
{"utterance_id": "5694-64029-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0006.flac", "task_type": "lexical_detection", "query": "trust", "prompt": "When the utterance is complete, report all instances of the word \"trust\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.94, 2.27, "trust"]], "response_mode": "offline", "answer": "[{\"word\":\"trust\",\"start\":1.94,\"end\":2.27}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[11.15, 11.6, "egypt"]], "response_mode": "offline", "answer": "[{\"word\":\"egypt\",\"start\":11.15,\"end\":11.6}]"}
{"utterance_id": "3081-166546-0081", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0081.flac", "task_type": "lexical_detection", "query": "wound", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"wound\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.55, "wound"]], "response_mode": "offline", "answer": "[{\"word\":\"wound\",\"start\":1.9,\"end\":2.55}]"}
{"utterance_id": "3853-163249-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0049.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.22, 6.7, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"christie\",\"start\":6.22,\"end\":6.7}]"}
{"utterance_id": "6319-57405-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0001.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After the audio ends, extract all occurrences of a disease name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.86, 5.54, "diseases"]], "response_mode": "offline", "answer": "[{\"word\":\"diseases\",\"start\":4.86,\"end\":5.54}]"}
{"utterance_id": "1673-143396-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0019.flac", "task_type": "lexical_detection", "query": "occasional", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"occasional\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 2.22, "occasional"]], "response_mode": "offline", "answer": "[{\"word\":\"occasional\",\"start\":1.65,\"end\":2.22}]"}
{"utterance_id": "251-136532-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0016.flac", "task_type": "lexical_detection", "query": "couldn't", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"couldn't\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.74, 4.98, "couldn't"]], "response_mode": "offline", "answer": "[{\"word\":\"couldn't\",\"start\":4.74,\"end\":4.98}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.77, 11.24, "ass"]], "response_mode": "offline", "answer": "[{\"word\":\"ass\",\"start\":10.77,\"end\":11.24}]"}
{"utterance_id": "777-126732-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0061.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.18, 1.65, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"black\",\"start\":1.18,\"end\":1.65}]"}
{"utterance_id": "1462-170138-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0025.flac", "task_type": "lexical_detection", "query": "here's", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"here's\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.02, 3.28, "here's"]], "response_mode": "offline", "answer": "[{\"word\":\"here's\",\"start\":3.02,\"end\":3.28}]"}
{"utterance_id": "251-136532-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0005.flac", "task_type": "lexical_detection", "query": "terra", "prompt": "After listening to the whole recording, output every instance of the word \"terra\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.46, "terra"]], "response_mode": "offline", "answer": "[{\"word\":\"terra\",\"start\":2.01,\"end\":2.46}]"}
{"utterance_id": "652-130726-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0016.flac", "task_type": "lexical_detection", "query": "sutter", "prompt": "Once the audio has finished, identify every instance of the word \"sutter\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.3, 3.64, "sutter"], [5.97, 6.31, "sutter"]], "response_mode": "offline", "answer": "[{\"word\":\"sutter\",\"start\":3.3,\"end\":3.64},{\"word\":\"sutter\",\"start\":5.97,\"end\":6.31}]"}
{"utterance_id": "2078-142845-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[18.43, 19.09, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":18.43,\"end\":19.09}]"}
{"utterance_id": "5536-43358-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0003.flac", "task_type": "lexical_detection", "query": "confess", "prompt": "After the audio ends, extract all occurrences of the word \"confess\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.49, "confess"]], "response_mode": "offline", "answer": "[{\"word\":\"confess\",\"start\":1.8,\"end\":2.49}]"}
{"utterance_id": "6345-93306-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0004.flac", "task_type": "lexical_detection", "query": "act", "prompt": "Process the full audio offline. Find every instance of the word \"act\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.6, "act"]], "response_mode": "offline", "answer": "[{\"word\":\"act\",\"start\":1.26,\"end\":1.6}]"}
{"utterance_id": "1993-147966-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0004.flac", "task_type": "lexical_detection", "query": "traits", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"traits\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.82, 3.14, "traits"]], "response_mode": "offline", "answer": "[{\"word\":\"traits\",\"start\":2.82,\"end\":3.14}]"}
{"utterance_id": "1673-143397-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.07, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":2.65,\"end\":3.07}]"}
{"utterance_id": "1988-24833-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0006.flac", "task_type": "lexical_detection", "query": "pop", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"pop\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.51, 0.91, "pop"]], "response_mode": "offline", "answer": "[{\"word\":\"pop\",\"start\":0.51,\"end\":0.91}]"}
{"utterance_id": "2035-147961-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0036.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[2.17, 2.4, "one"], [3.62, 4.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.17,\"end\":2.4},{\"word\":\"two\",\"start\":3.62,\"end\":4.13}]"}
{"utterance_id": "3081-166546-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.9, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":0.51,\"end\":0.9}]"}
{"utterance_id": "2078-142845-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0017.flac", "task_type": "lexical_detection", "query": "nourishment", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"nourishment\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.97, 5.54, "nourishment"]], "response_mode": "offline", "answer": "[{\"word\":\"nourishment\",\"start\":4.97,\"end\":5.54}]"}
{"utterance_id": "1673-143397-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0019.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[6.38, 6.84, "bishops"], [14.17, 14.68, "confessor"]], "response_mode": "offline", "answer": "[{\"word\":\"bishops\",\"start\":6.38,\"end\":6.84},{\"word\":\"confessor\",\"start\":14.17,\"end\":14.68}]"}
{"utterance_id": "5694-64029-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0030.flac", "task_type": "lexical_detection", "query": "stone", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"stone\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.5, 0.95, "stone"]], "response_mode": "offline", "answer": "[{\"word\":\"stone\",\"start\":0.5,\"end\":0.95}]"}
{"utterance_id": "6319-64726-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0005.flac", "task_type": "lexical_detection", "query": "hasty", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hasty\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.35, "hasty"]], "response_mode": "offline", "answer": "[{\"word\":\"hasty\",\"start\":3.86,\"end\":4.35}]"}
{"utterance_id": "3752-4944-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0062.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.6, 3.11, "water"]], "response_mode": "offline", "answer": "[{\"word\":\"water\",\"start\":2.6,\"end\":3.11}]"}
{"utterance_id": "2902-9006-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0017.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.55, 14.95, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":14.55,\"end\":14.95}]"}
{"utterance_id": "8297-275155-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0030.flac", "task_type": "lexical_detection", "query": "dear", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"dear\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.1, 1.44, "dear"]], "response_mode": "offline", "answer": "[{\"word\":\"dear\",\"start\":1.1,\"end\":1.44}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[3.62, 3.97, "one"], [9.11, 9.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.62,\"end\":3.97},{\"word\":\"two\",\"start\":9.11,\"end\":9.57}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "lexical_detection", "query": "sterling", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sterling\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.34, "sterling"]], "response_mode": "offline", "answer": "[{\"word\":\"sterling\",\"start\":1.83,\"end\":2.34}]"}
{"utterance_id": "652-130726-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0024.flac", "task_type": "lexical_counting", "query": "two", "prompt": "Process the full audio offline. For every occurrence of the word \"two\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.68, "one"], [3.77, 3.96, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.54,\"end\":1.68},{\"word\":\"two\",\"start\":3.77,\"end\":3.96}]"}
{"utterance_id": "3000-15664-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0001.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[19.81, 20.51, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":19.81,\"end\":20.51}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.14, "sons"]], "response_mode": "offline", "answer": "[{\"word\":\"sons\",\"start\":2.74,\"end\":3.14}]"}
{"utterance_id": "6295-64301-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0015.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Process the full audio offline. For every occurrence of a body part, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.83, "one"], [2.79, 3.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.45,\"end\":0.83},{\"word\":\"two\",\"start\":2.79,\"end\":3.47}]"}
{"utterance_id": "2902-9006-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0013.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 1.09, "huns"], [13.83, 14.18, "roman"]], "response_mode": "offline", "answer": "[{\"word\":\"huns\",\"start\":0.49,\"end\":1.09},{\"word\":\"roman\",\"start\":13.83,\"end\":14.18}]"}
{"utterance_id": "2412-153947-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0010.flac", "task_type": "semantic_counting", "query": "date", "prompt": "After the audio ends, tally every instance of a date. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[4.44, 6.24, "one"], [15.69, 16.72, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.44,\"end\":6.24},{\"word\":\"two\",\"start\":15.69,\"end\":16.72}]"}
{"utterance_id": "2035-147961-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0040.flac", "task_type": "lexical_detection", "query": "cave", "prompt": "When the utterance is complete, report all instances of the word \"cave\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.94, 4.48, "cave"]], "response_mode": "offline", "answer": "[{\"word\":\"cave\",\"start\":3.94,\"end\":4.48}]"}
{"utterance_id": "3000-15664-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0031.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.92, "modocs"]], "response_mode": "offline", "answer": "[{\"word\":\"modocs\",\"start\":3.26,\"end\":3.92}]"}
{"utterance_id": "6313-76958-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.32, 1.16, "stacy brown"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy brown\",\"start\":0.32,\"end\":1.16}]"}
{"utterance_id": "6313-66129-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0007.flac", "task_type": "lexical_detection", "query": "hungry", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hungry\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.02, "hungry"]], "response_mode": "offline", "answer": "[{\"word\":\"hungry\",\"start\":1.69,\"end\":2.02}]"}
{"utterance_id": "6313-66129-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0016.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Once the audio has finished, identify every instance of a date. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.08, 4.42, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":4.08,\"end\":4.42}]"}
{"utterance_id": "422-122949-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0025.flac", "task_type": "lexical_counting", "query": "profound", "prompt": "Listen to the entire clip, then report the count sequence for the word \"profound\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.03, "one"], [16.69, 17.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.03},{\"word\":\"two\",\"start\":16.69,\"end\":17.37}]"}
{"utterance_id": "1988-147956-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0019.flac", "task_type": "lexical_detection", "query": "edge", "prompt": "After the audio ends, extract all occurrences of the word \"edge\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 1.82, "edge"]], "response_mode": "offline", "answer": "[{\"word\":\"edge\",\"start\":1.57,\"end\":1.82}]"}
{"utterance_id": "2035-147961-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.5, 3.95, "ghosts"]], "response_mode": "offline", "answer": "[{\"word\":\"ghosts\",\"start\":3.5,\"end\":3.95}]"}
{"utterance_id": "2428-83699-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0041.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.97, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":1.42,\"end\":1.97}]"}
{"utterance_id": "7850-111771-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0001.flac", "task_type": "lexical_detection", "query": "interest", "prompt": "Once the audio has finished, identify every instance of the word \"interest\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.77, 5.27, "interest"]], "response_mode": "offline", "answer": "[{\"word\":\"interest\",\"start\":4.77,\"end\":5.27}]"}
{"utterance_id": "7976-105575-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0015.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.57, 7.0, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":6.57,\"end\":7.0}]"}
{"utterance_id": "5895-34615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0006.flac", "task_type": "lexical_detection", "query": "platform", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"platform\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 2.2, "platform"]], "response_mode": "offline", "answer": "[{\"word\":\"platform\",\"start\":1.45,\"end\":2.2}]"}
{"utterance_id": "1462-170145-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.37, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":0.69,\"end\":1.37}]"}
{"utterance_id": "1988-148538-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0011.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Once the audio has finished, identify every instance of a country name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.31, 1.84, "countries"]], "response_mode": "offline", "answer": "[{\"word\":\"countries\",\"start\":1.31,\"end\":1.84}]"}
{"utterance_id": "3081-166546-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0003.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "When the utterance is complete, report how a family relation accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.53, 1.86, "one"], [3.42, 3.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.53,\"end\":1.86},{\"word\":\"two\",\"start\":3.42,\"end\":3.92}]"}
{"utterance_id": "7976-110523-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0017.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.34, 1.79, "witch"]], "response_mode": "offline", "answer": "[{\"word\":\"witch\",\"start\":1.34,\"end\":1.79}]"}
{"utterance_id": "3853-163249-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0028.flac", "task_type": "lexical_detection", "query": "missus", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"missus\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.19, 1.51, "missus"]], "response_mode": "offline", "answer": "[{\"word\":\"missus\",\"start\":1.19,\"end\":1.51}]"}
{"utterance_id": "777-126732-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0077.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.67, 0.91, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":0.67,\"end\":0.91}]"}
{"utterance_id": "6241-61943-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.13, 1.51, "scholar"]], "response_mode": "offline", "answer": "[{\"word\":\"scholar\",\"start\":1.13,\"end\":1.51}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.12, 8.49, "knee"], [20.54, 20.92, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"knee\",\"start\":8.12,\"end\":8.49},{\"word\":\"body\",\"start\":20.54,\"end\":20.92}]"}
{"utterance_id": "1993-147149-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0014.flac", "task_type": "lexical_detection", "query": "broke", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"broke\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.1, 1.33, "broke"]], "response_mode": "offline", "answer": "[{\"word\":\"broke\",\"start\":1.1,\"end\":1.33}]"}
{"utterance_id": "1673-143397-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.16, 3.61, "theon"]], "response_mode": "offline", "answer": "[{\"word\":\"theon\",\"start\":3.16,\"end\":3.61}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "After the audio ends, tally every instance of an ordinal number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.14, 5.44, "one"], [7.74, 8.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.14,\"end\":5.44},{\"word\":\"two\",\"start\":7.74,\"end\":8.08}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "lexical_counting", "query": "april", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"april\" with word/start/end. If none, reply [].", "labels": [[6.52, 6.79, "one"], [8.88, 9.14, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.52,\"end\":6.79},{\"word\":\"two\",\"start\":8.88,\"end\":9.14}]"}
{"utterance_id": "6295-244435-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0006.flac", "task_type": "lexical_detection", "query": "completed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"completed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.44, "completed"]], "response_mode": "offline", "answer": "[{\"word\":\"completed\",\"start\":1.79,\"end\":2.44}]"}
{"utterance_id": "1993-147964-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0008.flac", "task_type": "lexical_detection", "query": "smelling", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"smelling\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.86, "smelling"]], "response_mode": "offline", "answer": "[{\"word\":\"smelling\",\"start\":2.4,\"end\":2.86}]"}
{"utterance_id": "1993-147965-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0004.flac", "task_type": "lexical_detection", "query": "dining", "prompt": "When the utterance is complete, report all instances of the word \"dining\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.23, "dining"]], "response_mode": "offline", "answer": "[{\"word\":\"dining\",\"start\":1.89,\"end\":2.23}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.7, "merchant's"]], "response_mode": "offline", "answer": "[{\"word\":\"merchant's\",\"start\":3.24,\"end\":3.7}]"}
{"utterance_id": "2086-149220-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0029.flac", "task_type": "lexical_detection", "query": "assure", "prompt": "After the audio ends, extract all occurrences of the word \"assure\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.08, "assure"]], "response_mode": "offline", "answer": "[{\"word\":\"assure\",\"start\":0.77,\"end\":1.08}]"}
{"utterance_id": "5536-43358-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0012.flac", "task_type": "lexical_detection", "query": "favor", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"favor\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.87, "favor"]], "response_mode": "offline", "answer": "[{\"word\":\"favor\",\"start\":2.38,\"end\":2.87}]"}
{"utterance_id": "2086-149214-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0001.flac", "task_type": "lexical_detection", "query": "flatter", "prompt": "When the utterance is complete, report all instances of the word \"flatter\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.1, 4.41, "flatter"]], "response_mode": "offline", "answer": "[{\"word\":\"flatter\",\"start\":4.1,\"end\":4.41}]"}
{"utterance_id": "6313-66125-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[4.28, 4.75, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":4.28,\"end\":4.75}]"}
{"utterance_id": "174-168635-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.83, 8.16, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":7.83,\"end\":8.16}]"}
{"utterance_id": "174-168635-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0019.flac", "task_type": "lexical_detection", "query": "eve", "prompt": "Once the audio has finished, identify every instance of the word \"eve\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.96, 4.38, "eve"]], "response_mode": "offline", "answer": "[{\"word\":\"eve\",\"start\":3.96,\"end\":4.38}]"}
{"utterance_id": "2428-83705-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0033.flac", "task_type": "lexical_detection", "query": "ann", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"ann\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.08, "ann"]], "response_mode": "offline", "answer": "[{\"word\":\"ann\",\"start\":1.83,\"end\":2.08}]"}
{"utterance_id": "7850-281318-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0020.flac", "task_type": "lexical_detection", "query": "luck", "prompt": "After listening to the whole recording, output every instance of the word \"luck\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.14, "luck"]], "response_mode": "offline", "answer": "[{\"word\":\"luck\",\"start\":0.79,\"end\":1.14}]"}
{"utterance_id": "6241-61946-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.17, "back"]], "response_mode": "offline", "answer": "[{\"word\":\"back\",\"start\":1.87,\"end\":2.17}]"}
{"utterance_id": "251-136532-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0012.flac", "task_type": "lexical_detection", "query": "lattimer", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"lattimer\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.15, 3.65, "lattimer"]], "response_mode": "offline", "answer": "[{\"word\":\"lattimer\",\"start\":3.15,\"end\":3.65}]"}
{"utterance_id": "3752-4944-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0017.flac", "task_type": "lexical_detection", "query": "grows", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"grows\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.14, 3.48, "grows"]], "response_mode": "offline", "answer": "[{\"word\":\"grows\",\"start\":3.14,\"end\":3.48}]"}
{"utterance_id": "1272-128104-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0010.flac", "task_type": "semantic_detection", "query": "country", "prompt": "When the utterance is complete, report all instances of a country name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 4.03, "india"]], "response_mode": "offline", "answer": "[{\"word\":\"india\",\"start\":3.5,\"end\":4.03}]"}
{"utterance_id": "6313-66129-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0021.flac", "task_type": "lexical_detection", "query": "stretching", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"stretching\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.78, 6.22, "stretching"]], "response_mode": "offline", "answer": "[{\"word\":\"stretching\",\"start\":5.78,\"end\":6.22}]"}
{"utterance_id": "1988-24833-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0025.flac", "task_type": "lexical_detection", "query": "anyway", "prompt": "Once the audio has finished, identify every instance of the word \"anyway\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.07, 2.45, "anyway"]], "response_mode": "offline", "answer": "[{\"word\":\"anyway\",\"start\":2.07,\"end\":2.45}]"}
{"utterance_id": "3853-163249-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0007.flac", "task_type": "lexical_detection", "query": "you'll", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"you'll\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 0.92, "you'll"]], "response_mode": "offline", "answer": "[{\"word\":\"you'll\",\"start\":0.76,\"end\":0.92}]"}
{"utterance_id": "3536-8226-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0029.flac", "task_type": "lexical_detection", "query": "watched", "prompt": "When the utterance is complete, report all instances of the word \"watched\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.89, "watched"]], "response_mode": "offline", "answer": "[{\"word\":\"watched\",\"start\":0.48,\"end\":0.89}]"}
{"utterance_id": "1919-142785-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0015.flac", "task_type": "lexical_detection", "query": "sufficiently", "prompt": "Process the full audio offline. Find every instance of the word \"sufficiently\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.43, "sufficiently"]], "response_mode": "offline", "answer": "[{\"word\":\"sufficiently\",\"start\":0.74,\"end\":1.43}]"}
{"utterance_id": "8842-302203-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.78, 13.17, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":12.78,\"end\":13.17}]"}
{"utterance_id": "5338-24615-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0008.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.67, 5.39, "architects"]], "response_mode": "offline", "answer": "[{\"word\":\"architects\",\"start\":4.67,\"end\":5.39}]"}
{"utterance_id": "174-84280-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0009.flac", "task_type": "lexical_detection", "query": "met", "prompt": "Once the audio has finished, identify every instance of the word \"met\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.48, 2.77, "met"]], "response_mode": "offline", "answer": "[{\"word\":\"met\",\"start\":2.48,\"end\":2.77}]"}
{"utterance_id": "6345-93306-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0023.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.68, 7.99, "noon"]], "response_mode": "offline", "answer": "[{\"word\":\"noon\",\"start\":7.68,\"end\":7.99}]"}
{"utterance_id": "1462-170145-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0007.flac", "task_type": "lexical_counting", "query": "myself", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"myself\" (word, start, end). If none, reply [].", "labels": [[0.67, 1.25, "one"], [3.48, 4.1, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.67,\"end\":1.25},{\"word\":\"two\",\"start\":3.48,\"end\":4.1}]"}
{"utterance_id": "2078-142845-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0033.flac", "task_type": "lexical_detection", "query": "swimming", "prompt": "After listening to the whole recording, output every instance of the word \"swimming\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.6, 10.15, "swimming"]], "response_mode": "offline", "answer": "[{\"word\":\"swimming\",\"start\":9.6,\"end\":10.15}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.02, 4.92, "soldier officer"], [6.47, 6.99, "recruit"]], "response_mode": "offline", "answer": "[{\"word\":\"soldier officer\",\"start\":4.02,\"end\":4.92},{\"word\":\"recruit\",\"start\":6.47,\"end\":6.99}]"}
{"utterance_id": "1919-142785-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.21, "one quarter"]], "response_mode": "offline", "answer": "[{\"word\":\"one quarter\",\"start\":1.67,\"end\":2.21}]"}
{"utterance_id": "3000-15664-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0043.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.57, 8.93, "sixty"]], "response_mode": "offline", "answer": "[{\"word\":\"sixty\",\"start\":8.57,\"end\":8.93}]"}
{"utterance_id": "6345-93302-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0007.flac", "task_type": "lexical_detection", "query": "railway", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"railway\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.27, 3.61, "railway"]], "response_mode": "offline", "answer": "[{\"word\":\"railway\",\"start\":3.27,\"end\":3.61}]"}
{"utterance_id": "2803-154320-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.32, "john"]], "response_mode": "offline", "answer": "[{\"word\":\"john\",\"start\":2.9,\"end\":3.32}]"}
{"utterance_id": "84-121550-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.44, 6.92, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":6.44,\"end\":6.92}]"}
{"utterance_id": "3170-137482-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0046.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.53, 4.03, "servant"]], "response_mode": "offline", "answer": "[{\"word\":\"servant\",\"start\":3.53,\"end\":4.03}]"}
{"utterance_id": "7976-110523-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0015.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 2.79, "water"]], "response_mode": "offline", "answer": "[{\"word\":\"water\",\"start\":2.54,\"end\":2.79}]"}
{"utterance_id": "5338-24615-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.04, 3.32, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":3.04,\"end\":3.32}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.0, 10.38, "hand"], [13.59, 14.05, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":10.0,\"end\":10.38},{\"word\":\"head\",\"start\":13.59,\"end\":14.05}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 2.07, "phil tyson"]], "response_mode": "offline", "answer": "[{\"word\":\"phil tyson\",\"start\":1.16,\"end\":2.07}]"}
{"utterance_id": "3536-23268-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0030.flac", "task_type": "lexical_detection", "query": "issue", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"issue\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.05, "issue"]], "response_mode": "offline", "answer": "[{\"word\":\"issue\",\"start\":2.74,\"end\":3.05}]"}
{"utterance_id": "3853-163249-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0021.flac", "task_type": "lexical_detection", "query": "follow", "prompt": "After the audio ends, extract all occurrences of the word \"follow\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.37, "follow"]], "response_mode": "offline", "answer": "[{\"word\":\"follow\",\"start\":3.81,\"end\":4.37}]"}
{"utterance_id": "1462-170142-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0001.flac", "task_type": "lexical_detection", "query": "drove", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"drove\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.3, 7.6, "drove"]], "response_mode": "offline", "answer": "[{\"word\":\"drove\",\"start\":7.3,\"end\":7.6}]"}
{"utterance_id": "5536-43358-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0000.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.51, 3.0, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":2.51,\"end\":3.0}]"}
{"utterance_id": "6241-66616-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0015.flac", "task_type": "lexical_detection", "query": "forests", "prompt": "Process the full audio offline. Find every instance of the word \"forests\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.33, "forests"]], "response_mode": "offline", "answer": "[{\"word\":\"forests\",\"start\":3.7,\"end\":4.33}]"}
{"utterance_id": "1462-170138-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0007.flac", "task_type": "lexical_detection", "query": "alexander's", "prompt": "After listening to the whole recording, output every instance of the word \"alexander's\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.62, "alexander's"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander's\",\"start\":2.98,\"end\":3.62}]"}
{"utterance_id": "6313-66129-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0019.flac", "task_type": "lexical_detection", "query": "pitched", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pitched\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.23, "pitched"]], "response_mode": "offline", "answer": "[{\"word\":\"pitched\",\"start\":1.76,\"end\":2.23}]"}
{"utterance_id": "5536-43358-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0010.flac", "task_type": "lexical_detection", "query": "material", "prompt": "Once the audio has finished, identify every instance of the word \"material\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.64, 7.09, "material"]], "response_mode": "offline", "answer": "[{\"word\":\"material\",\"start\":6.64,\"end\":7.09}]"}
{"utterance_id": "2035-147961-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0006.flac", "task_type": "lexical_detection", "query": "take", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"take\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.14, "take"]], "response_mode": "offline", "answer": "[{\"word\":\"take\",\"start\":0.98,\"end\":1.14}]"}
{"utterance_id": "2412-153947-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0004.flac", "task_type": "lexical_detection", "query": "inclined", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"inclined\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.04, 3.48, "inclined"]], "response_mode": "offline", "answer": "[{\"word\":\"inclined\",\"start\":3.04,\"end\":3.48}]"}
{"utterance_id": "3536-8226-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0028.flac", "task_type": "lexical_detection", "query": "town", "prompt": "After the audio ends, extract all occurrences of the word \"town\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.01, "town"]], "response_mode": "offline", "answer": "[{\"word\":\"town\",\"start\":0.66,\"end\":1.01}]"}
{"utterance_id": "3170-137482-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "When the utterance is complete, report all instances of an ordinal number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.76, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.44,\"end\":1.76}]"}
{"utterance_id": "1462-170138-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0004.flac", "task_type": "lexical_detection", "query": "myself", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"myself\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.3, 0.82, "myself"]], "response_mode": "offline", "answer": "[{\"word\":\"myself\",\"start\":0.3,\"end\":0.82}]"}
{"utterance_id": "6295-244435-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0028.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.57, "charleston"]], "response_mode": "offline", "answer": "[{\"word\":\"charleston\",\"start\":1.05,\"end\":1.57}]"}
{"utterance_id": "8842-302201-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0012.flac", "task_type": "lexical_detection", "query": "speak", "prompt": "When the utterance is complete, report all instances of the word \"speak\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.89, 4.16, "speak"]], "response_mode": "offline", "answer": "[{\"word\":\"speak\",\"start\":3.89,\"end\":4.16}]"}
{"utterance_id": "8842-302196-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.74, 6.43, "translator"], [7.87, 8.3, "poet"]], "response_mode": "offline", "answer": "[{\"word\":\"translator\",\"start\":5.74,\"end\":6.43},{\"word\":\"poet\",\"start\":7.87,\"end\":8.3}]"}
{"utterance_id": "3536-8226-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0018.flac", "task_type": "lexical_detection", "query": "domestic", "prompt": "After listening to the whole recording, output every instance of the word \"domestic\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.83, 8.39, "domestic"]], "response_mode": "offline", "answer": "[{\"word\":\"domestic\",\"start\":7.83,\"end\":8.39}]"}
{"utterance_id": "5895-34629-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.22, 0.75, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.22,\"end\":0.75}]"}
{"utterance_id": "6313-66129-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0020.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.45, 3.97, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":3.45,\"end\":3.97}]"}
{"utterance_id": "3853-163249-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0029.flac", "task_type": "lexical_detection", "query": "afford", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"afford\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.15, "afford"]], "response_mode": "offline", "answer": "[{\"word\":\"afford\",\"start\":0.78,\"end\":1.15}]"}
{"utterance_id": "6295-64301-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.7, 3.41, "joseph"]], "response_mode": "offline", "answer": "[{\"word\":\"joseph\",\"start\":2.7,\"end\":3.41}]"}
{"utterance_id": "3170-137482-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0015.flac", "task_type": "lexical_detection", "query": "thanking", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"thanking\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.75, 7.24, "thanking"]], "response_mode": "offline", "answer": "[{\"word\":\"thanking\",\"start\":6.75,\"end\":7.24}]"}
{"utterance_id": "3752-4943-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.39, 5.86, "faces"]], "response_mode": "offline", "answer": "[{\"word\":\"faces\",\"start\":5.39,\"end\":5.86}]"}
{"utterance_id": "1993-147149-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0001.flac", "task_type": "lexical_detection", "query": "feel", "prompt": "Once the audio has finished, identify every instance of the word \"feel\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.72, 2.08, "feel"]], "response_mode": "offline", "answer": "[{\"word\":\"feel\",\"start\":1.72,\"end\":2.08}]"}
{"utterance_id": "3000-15664-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0012.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.01, "leaves"], [10.91, 11.64, "chaparral"]], "response_mode": "offline", "answer": "[{\"word\":\"leaves\",\"start\":1.62,\"end\":2.01},{\"word\":\"chaparral\",\"start\":10.91,\"end\":11.64}]"}
{"utterance_id": "3853-163249-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0050.flac", "task_type": "lexical_detection", "query": "waiting", "prompt": "Process the full audio offline. Find every instance of the word \"waiting\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.8, "waiting"]], "response_mode": "offline", "answer": "[{\"word\":\"waiting\",\"start\":1.4,\"end\":1.8}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.14, 0.64, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.14,\"end\":0.64}]"}
{"utterance_id": "3853-163249-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0005.flac", "task_type": "lexical_detection", "query": "soaring", "prompt": "Once the audio has finished, identify every instance of the word \"soaring\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.08, 4.48, "soaring"]], "response_mode": "offline", "answer": "[{\"word\":\"soaring\",\"start\":4.08,\"end\":4.48}]"}
{"utterance_id": "1919-142785-0063", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0063.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.74, 2.34, "sage"]], "response_mode": "offline", "answer": "[{\"word\":\"sage\",\"start\":1.74,\"end\":2.34}]"}
{"utterance_id": "84-121550-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.02, 6.49, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":6.02,\"end\":6.49}]"}
{"utterance_id": "5694-64029-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 2.31, "shoulder"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulder\",\"start\":1.66,\"end\":2.31}]"}
{"utterance_id": "1993-147965-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0000.flac", "task_type": "lexical_detection", "query": "grandfather", "prompt": "After listening to the whole recording, output every instance of the word \"grandfather\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 0.89, "grandfather"]], "response_mode": "offline", "answer": "[{\"word\":\"grandfather\",\"start\":0.31,\"end\":0.89}]"}
{"utterance_id": "7976-105575-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0006.flac", "task_type": "lexical_detection", "query": "rebel", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"rebel\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.57, "rebel"]], "response_mode": "offline", "answer": "[{\"word\":\"rebel\",\"start\":1.15,\"end\":1.57}]"}
{"utterance_id": "3536-23268-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0011.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[0.67, 0.99, "one"], [11.31, 11.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.67,\"end\":0.99},{\"word\":\"two\",\"start\":11.31,\"end\":11.65}]"}
{"utterance_id": "5536-43359-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0009.flac", "task_type": "lexical_detection", "query": "anywhere", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"anywhere\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.02, 3.49, "anywhere"]], "response_mode": "offline", "answer": "[{\"word\":\"anywhere\",\"start\":3.02,\"end\":3.49}]"}
{"utterance_id": "3752-4944-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0043.flac", "task_type": "lexical_detection", "query": "prisoners", "prompt": "Once the audio has finished, identify every instance of the word \"prisoners\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.1, 1.66, "prisoners"]], "response_mode": "offline", "answer": "[{\"word\":\"prisoners\",\"start\":1.1,\"end\":1.66}]"}
{"utterance_id": "3752-4943-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0004.flac", "task_type": "lexical_detection", "query": "dawn", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"dawn\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.13, "dawn"]], "response_mode": "offline", "answer": "[{\"word\":\"dawn\",\"start\":0.86,\"end\":1.13}]"}
{"utterance_id": "2277-149897-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0014.flac", "task_type": "lexical_detection", "query": "therein", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"therein\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.43, 3.73, "therein"]], "response_mode": "offline", "answer": "[{\"word\":\"therein\",\"start\":3.43,\"end\":3.73}]"}
{"utterance_id": "1919-142785-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0024.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.39, "ginger"]], "response_mode": "offline", "answer": "[{\"word\":\"ginger\",\"start\":1.87,\"end\":2.39}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.84, 4.15, "rhein"]], "response_mode": "offline", "answer": "[{\"word\":\"rhein\",\"start\":3.84,\"end\":4.15}]"}
{"utterance_id": "3853-163249-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0019.flac", "task_type": "lexical_detection", "query": "garden", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"garden\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.52, "garden"]], "response_mode": "offline", "answer": "[{\"word\":\"garden\",\"start\":1.06,\"end\":1.52}]"}
{"utterance_id": "2803-154320-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0004.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Do not answer until the audio ends. Then list each instance of a country name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.38, 6.28, "australia"], [8.01, 8.83, "new zealand"]], "response_mode": "offline", "answer": "[{\"word\":\"australia\",\"start\":5.38,\"end\":6.28},{\"word\":\"new zealand\",\"start\":8.01,\"end\":8.83}]"}
{"utterance_id": "2035-152373-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0010.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.41, 5.87, "humber"]], "response_mode": "offline", "answer": "[{\"word\":\"humber\",\"start\":5.41,\"end\":5.87}]"}
{"utterance_id": "6295-64301-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0032.flac", "task_type": "lexical_detection", "query": "murmured", "prompt": "After the audio ends, extract all occurrences of the word \"murmured\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.51, "murmured"]], "response_mode": "offline", "answer": "[{\"word\":\"murmured\",\"start\":1.97,\"end\":2.51}]"}
{"utterance_id": "6295-64301-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0006.flac", "task_type": "lexical_detection", "query": "homes", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"homes\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.45, "homes"]], "response_mode": "offline", "answer": "[{\"word\":\"homes\",\"start\":2.95,\"end\":3.45}]"}
{"utterance_id": "5536-43363-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.01, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.78,\"end\":1.01}]"}
{"utterance_id": "2078-142845-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0008.flac", "task_type": "lexical_detection", "query": "knife", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"knife\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.25, 6.78, "knife"]], "response_mode": "offline", "answer": "[{\"word\":\"knife\",\"start\":6.25,\"end\":6.78}]"}
{"utterance_id": "6241-61946-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0008.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.81, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":3.31,\"end\":3.81}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "lexical_detection", "query": "restore", "prompt": "Process the full audio offline. Find every instance of the word \"restore\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.19, "restore"]], "response_mode": "offline", "answer": "[{\"word\":\"restore\",\"start\":1.79,\"end\":2.19}]"}
{"utterance_id": "1919-142785-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0036.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.8, 5.29, "leaves"]], "response_mode": "offline", "answer": "[{\"word\":\"leaves\",\"start\":4.8,\"end\":5.29}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.52, "legs"], [6.67, 7.05, "legs"], [8.68, 8.93, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"legs\",\"start\":3.15,\"end\":3.52},{\"word\":\"legs\",\"start\":6.67,\"end\":7.05},{\"word\":\"feet\",\"start\":8.68,\"end\":8.93}]"}
{"utterance_id": "1988-24833-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.25, 5.63, "mom"]], "response_mode": "offline", "answer": "[{\"word\":\"mom\",\"start\":5.25,\"end\":5.63}]"}
{"utterance_id": "3576-138058-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0022.flac", "task_type": "lexical_detection", "query": "gentleman", "prompt": "Once the audio has finished, identify every instance of the word \"gentleman\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.09, 1.67, "gentleman"]], "response_mode": "offline", "answer": "[{\"word\":\"gentleman\",\"start\":1.09,\"end\":1.67}]"}
{"utterance_id": "7976-105575-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0021.flac", "task_type": "lexical_detection", "query": "rebels", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"rebels\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.57, "rebels"]], "response_mode": "offline", "answer": "[{\"word\":\"rebels\",\"start\":1.16,\"end\":1.57}]"}
{"utterance_id": "652-130737-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0001.flac", "task_type": "lexical_detection", "query": "sauterne", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sauterne\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.07, 4.72, "sauterne"]], "response_mode": "offline", "answer": "[{\"word\":\"sauterne\",\"start\":4.07,\"end\":4.72}]"}
{"utterance_id": "2428-83699-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.62, "bone"]], "response_mode": "offline", "answer": "[{\"word\":\"bone\",\"start\":1.08,\"end\":1.62}]"}
{"utterance_id": "174-168635-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0000.flac", "task_type": "lexical_detection", "query": "lover", "prompt": "Process the full audio offline. Find every instance of the word \"lover\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.55, "lover"]], "response_mode": "offline", "answer": "[{\"word\":\"lover\",\"start\":2.01,\"end\":2.55}]"}
{"utterance_id": "3081-166546-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0008.flac", "task_type": "lexical_detection", "query": "george", "prompt": "Process the full audio offline. Find every instance of the word \"george\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.25, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":0.52,\"end\":1.25}]"}
{"utterance_id": "6241-66616-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0020.flac", "task_type": "lexical_detection", "query": "summer", "prompt": "Once the audio has finished, identify every instance of the word \"summer\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.13, 2.57, "summer"]], "response_mode": "offline", "answer": "[{\"word\":\"summer\",\"start\":2.13,\"end\":2.57}]"}
{"utterance_id": "2086-149220-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 2.95, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.72,\"end\":2.95}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.84, 6.44, "tad butler"]], "response_mode": "offline", "answer": "[{\"word\":\"tad butler\",\"start\":5.84,\"end\":6.44}]"}
{"utterance_id": "5694-64038-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0008.flac", "task_type": "lexical_detection", "query": "man", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"man\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 0.86, "man"]], "response_mode": "offline", "answer": "[{\"word\":\"man\",\"start\":0.64,\"end\":0.86}]"}
{"utterance_id": "1919-142785-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0003.flac", "task_type": "lexical_detection", "query": "aromatic", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"aromatic\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.68, 2.23, "aromatic"]], "response_mode": "offline", "answer": "[{\"word\":\"aromatic\",\"start\":1.68,\"end\":2.23}]"}
{"utterance_id": "1993-147966-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.27, "mama"]], "response_mode": "offline", "answer": "[{\"word\":\"mama\",\"start\":0.84,\"end\":1.27}]"}
{"utterance_id": "2035-147960-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0015.flac", "task_type": "lexical_detection", "query": "otto", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"otto\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.74, "otto"]], "response_mode": "offline", "answer": "[{\"word\":\"otto\",\"start\":0.42,\"end\":0.74}]"}
{"utterance_id": "6313-76958-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0001.flac", "task_type": "lexical_detection", "query": "beat", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"beat\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.1, "beat"]], "response_mode": "offline", "answer": "[{\"word\":\"beat\",\"start\":1.9,\"end\":2.1}]"}
{"utterance_id": "2035-147961-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0028.flac", "task_type": "lexical_detection", "query": "thirty", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"thirty\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.1, 1.61, "thirty"]], "response_mode": "offline", "answer": "[{\"word\":\"thirty\",\"start\":1.1,\"end\":1.61}]"}
{"utterance_id": "1988-148538-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0010.flac", "task_type": "lexical_detection", "query": "applaud", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"applaud\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.66, 1.14, "applaud"]], "response_mode": "offline", "answer": "[{\"word\":\"applaud\",\"start\":0.66,\"end\":1.14}]"}
{"utterance_id": "6319-275224-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0003.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.08, 1.72, "plants"]], "response_mode": "offline", "answer": "[{\"word\":\"plants\",\"start\":1.08,\"end\":1.72}]"}
{"utterance_id": "1673-143396-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0020.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a number with word/start/end. If none, reply [].", "labels": [[8.42, 8.71, "one"], [10.74, 11.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.42,\"end\":8.71},{\"word\":\"two\",\"start\":10.74,\"end\":11.0}]"}
{"utterance_id": "2277-149897-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0033.flac", "task_type": "lexical_detection", "query": "hay", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hay\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.62, "hay"]], "response_mode": "offline", "answer": "[{\"word\":\"hay\",\"start\":3.29,\"end\":3.62}]"}
{"utterance_id": "3081-166546-0085", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0085.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.83, 8.58, "miss clarke"]], "response_mode": "offline", "answer": "[{\"word\":\"miss clarke\",\"start\":7.83,\"end\":8.58}]"}
{"utterance_id": "2035-147961-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0003.flac", "task_type": "lexical_detection", "query": "lay", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"lay\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.63, 0.89, "lay"]], "response_mode": "offline", "answer": "[{\"word\":\"lay\",\"start\":0.63,\"end\":0.89}]"}
{"utterance_id": "6345-64257-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.92, 1.11, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.92,\"end\":1.11}]"}
{"utterance_id": "2277-149897-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 3.67, "four five six"]], "response_mode": "offline", "answer": "[{\"word\":\"four five six\",\"start\":1.66,\"end\":3.67}]"}
{"utterance_id": "3752-4944-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 1.11, "captain frere"]], "response_mode": "offline", "answer": "[{\"word\":\"captain frere\",\"start\":0.37,\"end\":1.11}]"}
{"utterance_id": "1988-24833-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.55, 4.76, "limb"]], "response_mode": "offline", "answer": "[{\"word\":\"limb\",\"start\":4.55,\"end\":4.76}]"}
{"utterance_id": "5694-64038-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.58, 1.88, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":1.58,\"end\":1.88}]"}
{"utterance_id": "652-130726-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.68, "two"], [3.77, 3.96, "two"], [5.39, 5.75, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.54,\"end\":1.68},{\"word\":\"two\",\"start\":3.77,\"end\":3.96},{\"word\":\"one\",\"start\":5.39,\"end\":5.75}]"}
{"utterance_id": "1272-141231-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0025.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.85, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":4.45,\"end\":4.85}]"}
{"utterance_id": "5536-43363-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 2.04, "seventy five"]], "response_mode": "offline", "answer": "[{\"word\":\"seventy five\",\"start\":1.43,\"end\":2.04}]"}
{"utterance_id": "652-129742-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0016.flac", "task_type": "lexical_counting", "query": "baskets", "prompt": "Process the full audio offline. For every occurrence of the word \"baskets\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.54, "one"], [3.48, 4.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.88,\"end\":1.54},{\"word\":\"two\",\"start\":3.48,\"end\":4.13}]"}
{"utterance_id": "2428-83705-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0017.flac", "task_type": "lexical_detection", "query": "expectations", "prompt": "Process the full audio offline. Find every instance of the word \"expectations\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.62, "expectations"]], "response_mode": "offline", "answer": "[{\"word\":\"expectations\",\"start\":3.58,\"end\":4.62}]"}
{"utterance_id": "5694-64025-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0003.flac", "task_type": "lexical_detection", "query": "brigade", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"brigade\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.43, "brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"brigade\",\"start\":2.87,\"end\":3.43}]"}
{"utterance_id": "3536-8226-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[2.62, 3.01, "husband"], [8.87, 9.48, "children"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":2.62,\"end\":3.01},{\"word\":\"children\",\"start\":8.87,\"end\":9.48}]"}
{"utterance_id": "3853-163249-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0004.flac", "task_type": "lexical_detection", "query": "school", "prompt": "After the audio ends, extract all occurrences of the word \"school\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.91, "school"]], "response_mode": "offline", "answer": "[{\"word\":\"school\",\"start\":1.36,\"end\":1.91}]"}
{"utterance_id": "2035-147961-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[3.06, 3.43, "legs"]], "response_mode": "offline", "answer": "[{\"word\":\"legs\",\"start\":3.06,\"end\":3.43}]"}
{"utterance_id": "2277-149897-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0019.flac", "task_type": "lexical_detection", "query": "o'clock", "prompt": "After listening to the whole recording, output every instance of the word \"o'clock\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.08, "o'clock"]], "response_mode": "offline", "answer": "[{\"word\":\"o'clock\",\"start\":0.76,\"end\":1.08}]"}
{"utterance_id": "5338-284437-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0029.flac", "task_type": "lexical_counting", "query": "right", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"right\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[2.2, 2.53, "one"], [7.18, 7.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.2,\"end\":2.53},{\"word\":\"two\",\"start\":7.18,\"end\":7.45}]"}
{"utterance_id": "3000-15664-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0016.flac", "task_type": "lexical_detection", "query": "emerald", "prompt": "Once the audio has finished, identify every instance of the word \"emerald\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.28, 1.68, "emerald"]], "response_mode": "offline", "answer": "[{\"word\":\"emerald\",\"start\":1.28,\"end\":1.68}]"}
{"utterance_id": "5536-43359-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0010.flac", "task_type": "lexical_detection", "query": "necessary", "prompt": "Once the audio has finished, identify every instance of the word \"necessary\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.68, 2.3, "necessary"]], "response_mode": "offline", "answer": "[{\"word\":\"necessary\",\"start\":1.68,\"end\":2.3}]"}
{"utterance_id": "84-121123-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0022.flac", "task_type": "lexical_detection", "query": "entered", "prompt": "Process the full audio offline. Find every instance of the word \"entered\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.72, "entered"]], "response_mode": "offline", "answer": "[{\"word\":\"entered\",\"start\":1.47,\"end\":1.72}]"}
{"utterance_id": "1919-142785-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0055.flac", "task_type": "lexical_detection", "query": "cookery", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cookery\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.09, 3.49, "cookery"]], "response_mode": "offline", "answer": "[{\"word\":\"cookery\",\"start\":3.09,\"end\":3.49}]"}
{"utterance_id": "1462-170145-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.22, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander\",\"start\":0.52,\"end\":1.22}]"}
{"utterance_id": "5895-34629-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.93, "saint paul"]], "response_mode": "offline", "answer": "[{\"word\":\"saint paul\",\"start\":0.28,\"end\":0.93}]"}
{"utterance_id": "2035-147961-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0037.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.63, "peter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":1.28,\"end\":1.63}]"}
{"utterance_id": "6295-244435-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0027.flac", "task_type": "lexical_detection", "query": "victory", "prompt": "Process the full audio offline. Find every instance of the word \"victory\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.3, 2.01, "victory"]], "response_mode": "offline", "answer": "[{\"word\":\"victory\",\"start\":1.3,\"end\":2.01}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.69, 4.12, "bride"], [6.97, 7.35, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"bride\",\"start\":3.69,\"end\":4.12},{\"word\":\"brother\",\"start\":6.97,\"end\":7.35}]"}
{"utterance_id": "2277-149896-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0011.flac", "task_type": "lexical_detection", "query": "restless", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"restless\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.16, "restless"]], "response_mode": "offline", "answer": "[{\"word\":\"restless\",\"start\":0.67,\"end\":1.16}]"}
{"utterance_id": "3576-138058-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0030.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.91, 3.21, "one"], [4.96, 5.28, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.91,\"end\":3.21},{\"word\":\"two\",\"start\":4.96,\"end\":5.28}]"}
{"utterance_id": "5694-64029-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0011.flac", "task_type": "lexical_detection", "query": "killed", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"killed\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.93, 1.2, "killed"]], "response_mode": "offline", "answer": "[{\"word\":\"killed\",\"start\":0.93,\"end\":1.2}]"}
{"utterance_id": "3853-163249-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.73, 1.13, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"christie\",\"start\":0.73,\"end\":1.13}]"}
{"utterance_id": "6313-66125-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0026.flac", "task_type": "lexical_detection", "query": "sure", "prompt": "When the utterance is complete, report all instances of the word \"sure\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 0.71, "sure"]], "response_mode": "offline", "answer": "[{\"word\":\"sure\",\"start\":0.26,\"end\":0.71}]"}
{"utterance_id": "2412-153947-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0001.flac", "task_type": "lexical_detection", "query": "perfectly", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"perfectly\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.67, 2.13, "perfectly"]], "response_mode": "offline", "answer": "[{\"word\":\"perfectly\",\"start\":1.67,\"end\":2.13}]"}
{"utterance_id": "6345-93302-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0006.flac", "task_type": "lexical_detection", "query": "setting", "prompt": "After the audio ends, extract all occurrences of the word \"setting\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.44, 0.75, "setting"]], "response_mode": "offline", "answer": "[{\"word\":\"setting\",\"start\":0.44,\"end\":0.75}]"}
{"utterance_id": "3081-166546-0086", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0086.flac", "task_type": "lexical_detection", "query": "lady's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lady's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.58, "lady's"]], "response_mode": "offline", "answer": "[{\"word\":\"lady's\",\"start\":2.2,\"end\":2.58}]"}
{"utterance_id": "652-130726-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0026.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.76, 5.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":4.76,\"end\":5.01}]"}
{"utterance_id": "7976-110523-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0005.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.76, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":3.4,\"end\":3.76}]"}
{"utterance_id": "3536-8226-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.86, "hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"hundred\",\"start\":1.43,\"end\":1.86}]"}
{"utterance_id": "3853-163249-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0022.flac", "task_type": "lexical_detection", "query": "marry", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"marry\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.17, 3.58, "marry"]], "response_mode": "offline", "answer": "[{\"word\":\"marry\",\"start\":3.17,\"end\":3.58}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.07, "last"], [14.06, 14.45, "tenth"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.75,\"end\":1.07},{\"word\":\"tenth\",\"start\":14.06,\"end\":14.45}]"}
{"utterance_id": "1462-170145-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0015.flac", "task_type": "lexical_detection", "query": "york", "prompt": "After listening to the whole recording, output every instance of the word \"york\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.14, 4.51, "york"]], "response_mode": "offline", "answer": "[{\"word\":\"york\",\"start\":4.14,\"end\":4.51}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.92, "warrior"]], "response_mode": "offline", "answer": "[{\"word\":\"warrior\",\"start\":3.4,\"end\":3.92}]"}
{"utterance_id": "3752-4944-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0008.flac", "task_type": "lexical_detection", "query": "friend", "prompt": "Process the full audio offline. Find every instance of the word \"friend\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.13, "friend"]], "response_mode": "offline", "answer": "[{\"word\":\"friend\",\"start\":0.86,\"end\":1.13}]"}
{"utterance_id": "5536-43358-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0018.flac", "task_type": "lexical_detection", "query": "task", "prompt": "After listening to the whole recording, output every instance of the word \"task\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.63, 6.05, "task"]], "response_mode": "offline", "answer": "[{\"word\":\"task\",\"start\":5.63,\"end\":6.05}]"}
{"utterance_id": "5694-64025-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0003.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the full audio. After it finishes, find every instance of a time of day. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.22, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.72,\"end\":2.22}]"}
{"utterance_id": "652-130726-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0027.flac", "task_type": "lexical_detection", "query": "cook", "prompt": "Once the audio has finished, identify every instance of the word \"cook\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.57, 1.79, "cook"]], "response_mode": "offline", "answer": "[{\"word\":\"cook\",\"start\":1.57,\"end\":1.79}]"}
{"utterance_id": "6313-76958-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0021.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Once the audio has finished, list each occurrence of an animal name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[6.37, 6.81, "one"], [10.93, 11.24, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.37,\"end\":6.81},{\"word\":\"two\",\"start\":10.93,\"end\":11.24}]"}
{"utterance_id": "2086-149220-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0024.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a weather term with fields word, start, end (seconds). If none, reply [].", "labels": [[2.89, 3.45, "sunshine"]], "response_mode": "offline", "answer": "[{\"word\":\"sunshine\",\"start\":2.89,\"end\":3.45}]"}
{"utterance_id": "7850-111771-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.26, 5.72, "jesse"]], "response_mode": "offline", "answer": "[{\"word\":\"jesse\",\"start\":5.26,\"end\":5.72}]"}
{"utterance_id": "7976-105575-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.8, 3.41, "hamilton's"]], "response_mode": "offline", "answer": "[{\"word\":\"hamilton's\",\"start\":2.8,\"end\":3.41}]"}
{"utterance_id": "8842-302203-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.71, 8.24, "kindred"]], "response_mode": "offline", "answer": "[{\"word\":\"kindred\",\"start\":7.71,\"end\":8.24}]"}
{"utterance_id": "3000-15664-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0022.flac", "task_type": "lexical_detection", "query": "strike", "prompt": "Process the full audio offline. Find every instance of the word \"strike\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 0.98, "strike"]], "response_mode": "offline", "answer": "[{\"word\":\"strike\",\"start\":0.59,\"end\":0.98}]"}
{"utterance_id": "3081-166546-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0057.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.3, 1.61, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.3,\"end\":1.61}]"}
{"utterance_id": "5338-284437-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0008.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.75, 8.07, "pink"]], "response_mode": "offline", "answer": "[{\"word\":\"pink\",\"start\":7.75,\"end\":8.07}]"}
{"utterance_id": "6345-93306-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0023.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.76, 11.48, "charwoman"]], "response_mode": "offline", "answer": "[{\"word\":\"charwoman\",\"start\":10.76,\"end\":11.48}]"}
{"utterance_id": "422-122949-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0008.flac", "task_type": "lexical_detection", "query": "slave", "prompt": "After the audio ends, extract all occurrences of the word \"slave\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.39, "slave"]], "response_mode": "offline", "answer": "[{\"word\":\"slave\",\"start\":1.01,\"end\":1.39}]"}
{"utterance_id": "3576-138058-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0034.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.14, 7.67, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":7.14,\"end\":7.67}]"}
{"utterance_id": "1988-24833-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0000.flac", "task_type": "lexical_detection", "query": "stray", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"stray\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.77, 1.11, "stray"]], "response_mode": "offline", "answer": "[{\"word\":\"stray\",\"start\":0.77,\"end\":1.11}]"}
{"utterance_id": "7976-110124-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0021.flac", "task_type": "lexical_detection", "query": "robber", "prompt": "When the utterance is complete, report all instances of the word \"robber\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 3.11, "robber"]], "response_mode": "offline", "answer": "[{\"word\":\"robber\",\"start\":2.62,\"end\":3.11}]"}
{"utterance_id": "6241-66616-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0011.flac", "task_type": "lexical_detection", "query": "port", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"port\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.12, 7.37, "port"]], "response_mode": "offline", "answer": "[{\"word\":\"port\",\"start\":7.12,\"end\":7.37}]"}
{"utterance_id": "7976-105575-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.57, 4.27, "hamilton's"]], "response_mode": "offline", "answer": "[{\"word\":\"hamilton's\",\"start\":3.57,\"end\":4.27}]"}
{"utterance_id": "5694-64038-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0023.flac", "task_type": "lexical_detection", "query": "third", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"third\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.79, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"third\",\"start\":0.54,\"end\":0.79}]"}
{"utterance_id": "8297-275154-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.43, 1.95, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":1.43,\"end\":1.95}]"}
{"utterance_id": "777-126732-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0004.flac", "task_type": "lexical_detection", "query": "ounce", "prompt": "After listening to the whole recording, output every instance of the word \"ounce\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.34, 3.83, "ounce"]], "response_mode": "offline", "answer": "[{\"word\":\"ounce\",\"start\":3.34,\"end\":3.83}]"}
{"utterance_id": "1272-135031-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0005.flac", "task_type": "lexical_detection", "query": "ruggedo", "prompt": "When the utterance is complete, report all instances of the word \"ruggedo\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.39, "ruggedo"]], "response_mode": "offline", "answer": "[{\"word\":\"ruggedo\",\"start\":0.9,\"end\":1.39}]"}
{"utterance_id": "7850-111771-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0008.flac", "task_type": "lexical_detection", "query": "finished", "prompt": "When the utterance is complete, report all instances of the word \"finished\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.45, "finished"]], "response_mode": "offline", "answer": "[{\"word\":\"finished\",\"start\":1.9,\"end\":2.45}]"}
{"utterance_id": "7976-110523-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.43, 3.64, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":3.43,\"end\":3.64}]"}
{"utterance_id": "6345-93302-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0021.flac", "task_type": "lexical_detection", "query": "sorry", "prompt": "Process the full audio offline. Find every instance of the word \"sorry\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.4, "sorry"]], "response_mode": "offline", "answer": "[{\"word\":\"sorry\",\"start\":1.12,\"end\":1.4}]"}
{"utterance_id": "5694-64029-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0023.flac", "task_type": "lexical_counting", "query": "time", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"time\" with word/start/end. If none, reply [].", "labels": [[1.11, 1.37, "one"], [4.98, 5.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.11,\"end\":1.37},{\"word\":\"two\",\"start\":4.98,\"end\":5.32}]"}
{"utterance_id": "6313-66125-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.29, 1.68, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":1.29,\"end\":1.68}]"}
{"utterance_id": "6313-76958-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.22, 5.15, "professor zepplin"]], "response_mode": "offline", "answer": "[{\"word\":\"professor zepplin\",\"start\":4.22,\"end\":5.15}]"}
{"utterance_id": "174-84280-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0011.flac", "task_type": "lexical_detection", "query": "seem", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"seem\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 1.87, "seem"]], "response_mode": "offline", "answer": "[{\"word\":\"seem\",\"start\":1.48,\"end\":1.87}]"}
{"utterance_id": "6241-66616-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0013.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.29, 2.64, "woongas"]], "response_mode": "offline", "answer": "[{\"word\":\"woongas\",\"start\":2.29,\"end\":2.64}]"}
{"utterance_id": "7976-110523-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0015.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After the audio ends, extract all occurrences of a time of day. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.67, 6.03, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":5.67,\"end\":6.03}]"}
{"utterance_id": "652-130726-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0008.flac", "task_type": "lexical_counting", "query": "order", "prompt": "Listen to the full audio. Count every occurrence of the word \"order\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.93, "one"], [3.84, 4.17, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.34,\"end\":1.93},{\"word\":\"two\",\"start\":3.84,\"end\":4.17}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.46, "two"], [2.46, 2.74, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.21,\"end\":1.46},{\"word\":\"one\",\"start\":2.46,\"end\":2.74}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[22.68, 23.36, "minerva"]], "response_mode": "offline", "answer": "[{\"word\":\"minerva\",\"start\":22.68,\"end\":23.36}]"}
{"utterance_id": "6345-93306-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0015.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Process the full audio offline. Find every instance of a money or currency term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.05, 7.32, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":7.05,\"end\":7.32}]"}
{"utterance_id": "8297-275154-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0009.flac", "task_type": "lexical_detection", "query": "wine", "prompt": "Process the full audio offline. Find every instance of the word \"wine\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.47, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":1.16,\"end\":1.47}]"}
{"utterance_id": "8297-275155-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.67, 6.16, "wife"], [7.22, 7.82, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":5.67,\"end\":6.16},{\"word\":\"husband\",\"start\":7.22,\"end\":7.82}]"}
{"utterance_id": "2078-142845-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0019.flac", "task_type": "lexical_detection", "query": "preparations", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"preparations\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.83, 1.75, "preparations"]], "response_mode": "offline", "answer": "[{\"word\":\"preparations\",\"start\":0.83,\"end\":1.75}]"}
{"utterance_id": "3752-4943-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0012.flac", "task_type": "lexical_detection", "query": "foot", "prompt": "Process the full audio offline. Find every instance of the word \"foot\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.85, 3.2, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":2.85,\"end\":3.2}]"}
{"utterance_id": "2412-153954-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0001.flac", "task_type": "lexical_detection", "query": "politeness", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"politeness\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.8, 3.34, "politeness"]], "response_mode": "offline", "answer": "[{\"word\":\"politeness\",\"start\":2.8,\"end\":3.34}]"}
{"utterance_id": "1993-147149-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0022.flac", "task_type": "lexical_counting", "query": "kind", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"kind\" (word, start, end). If none, reply [].", "labels": [[0.96, 1.28, "one"], [6.07, 6.39, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.96,\"end\":1.28},{\"word\":\"two\",\"start\":6.07,\"end\":6.39}]"}
{"utterance_id": "5895-34629-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0012.flac", "task_type": "lexical_detection", "query": "southwark", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"southwark\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.55, 1.15, "southwark"]], "response_mode": "offline", "answer": "[{\"word\":\"southwark\",\"start\":0.55,\"end\":1.15}]"}
{"utterance_id": "652-130726-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0015.flac", "task_type": "lexical_detection", "query": "idea", "prompt": "After listening to the whole recording, output every instance of the word \"idea\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.07, "idea"]], "response_mode": "offline", "answer": "[{\"word\":\"idea\",\"start\":0.66,\"end\":1.07}]"}
{"utterance_id": "5694-64029-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0012.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.98, 4.35, "regiment"], [6.56, 7.01, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":3.98,\"end\":4.35},{\"word\":\"regiment\",\"start\":6.56,\"end\":7.01}]"}
{"utterance_id": "2803-154328-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0019.flac", "task_type": "lexical_detection", "query": "poniard", "prompt": "Process the full audio offline. Find every instance of the word \"poniard\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.04, 3.74, "poniard"]], "response_mode": "offline", "answer": "[{\"word\":\"poniard\",\"start\":3.04,\"end\":3.74}]"}
{"utterance_id": "1272-135031-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0016.flac", "task_type": "lexical_detection", "query": "hesitated", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hesitated\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.64, "hesitated"]], "response_mode": "offline", "answer": "[{\"word\":\"hesitated\",\"start\":0.9,\"end\":1.64}]"}
{"utterance_id": "2277-149897-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0004.flac", "task_type": "lexical_detection", "query": "panacea", "prompt": "After listening to the whole recording, output every instance of the word \"panacea\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.55, 4.04, "panacea"]], "response_mode": "offline", "answer": "[{\"word\":\"panacea\",\"start\":3.55,\"end\":4.04}]"}
{"utterance_id": "3000-15664-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0002.flac", "task_type": "lexical_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"profession\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.97, 1.61, "profession"]], "response_mode": "offline", "answer": "[{\"word\":\"profession\",\"start\":0.97,\"end\":1.61}]"}
{"utterance_id": "6295-244435-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0039.flac", "task_type": "lexical_detection", "query": "fashion", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fashion\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.26, "fashion"]], "response_mode": "offline", "answer": "[{\"word\":\"fashion\",\"start\":0.83,\"end\":1.26}]"}
{"utterance_id": "6241-61946-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0008.flac", "task_type": "lexical_detection", "query": "guide", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"guide\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.31, 3.81, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":3.31,\"end\":3.81}]"}
{"utterance_id": "1272-128104-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0005.flac", "task_type": "lexical_counting", "query": "how", "prompt": "Once the audio has finished, list each occurrence of the word \"how\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[3.46, 3.71, "one"], [6.34, 6.62, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.46,\"end\":3.71},{\"word\":\"two\",\"start\":6.34,\"end\":6.62}]"}
{"utterance_id": "3853-163249-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0023.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.21, 2.8, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":2.21,\"end\":2.8}]"}
{"utterance_id": "3576-138058-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0039.flac", "task_type": "lexical_counting", "query": "ours", "prompt": "After the audio ends, tally every instance of the word \"ours\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.64, 6.04, "one"], [13.37, 14.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.64,\"end\":6.04},{\"word\":\"two\",\"start\":13.37,\"end\":14.01}]"}
{"utterance_id": "777-126732-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0036.flac", "task_type": "lexical_detection", "query": "visions", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"visions\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.26, 5.66, "visions"]], "response_mode": "offline", "answer": "[{\"word\":\"visions\",\"start\":5.26,\"end\":5.66}]"}
{"utterance_id": "251-137823-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.37, 4.27, "mark faber"]], "response_mode": "offline", "answer": "[{\"word\":\"mark faber\",\"start\":3.37,\"end\":4.27}]"}
{"utterance_id": "7976-105575-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0015.flac", "task_type": "lexical_detection", "query": "six", "prompt": "After listening to the whole recording, output every instance of the word \"six\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 2.9, "six"]], "response_mode": "offline", "answer": "[{\"word\":\"six\",\"start\":2.61,\"end\":2.9}]"}
{"utterance_id": "3752-4944-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0068.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.62, 13.06, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":12.62,\"end\":13.06}]"}
{"utterance_id": "5694-64029-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.23, 7.59, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":7.23,\"end\":7.59}]"}
{"utterance_id": "174-168635-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0006.flac", "task_type": "lexical_detection", "query": "felt", "prompt": "After the audio ends, extract all occurrences of the word \"felt\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.17, "felt"], [2.42, 2.77, "felt"]], "response_mode": "offline", "answer": "[{\"word\":\"felt\",\"start\":0.75,\"end\":1.17},{\"word\":\"felt\",\"start\":2.42,\"end\":2.77}]"}
{"utterance_id": "3853-163249-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0037.flac", "task_type": "lexical_detection", "query": "burden", "prompt": "After listening to the whole recording, output every instance of the word \"burden\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.48, 2.85, "burden"]], "response_mode": "offline", "answer": "[{\"word\":\"burden\",\"start\":2.48,\"end\":2.85}]"}
{"utterance_id": "5694-64025-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0016.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.65, "side"]], "response_mode": "offline", "answer": "[{\"word\":\"side\",\"start\":4.13,\"end\":4.65}]"}
{"utterance_id": "174-50561-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0001.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.63, 6.12, "flower"]], "response_mode": "offline", "answer": "[{\"word\":\"flower\",\"start\":5.63,\"end\":6.12}]"}
{"utterance_id": "7850-286674-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0002.flac", "task_type": "lexical_detection", "query": "bringing", "prompt": "Process the full audio offline. Find every instance of the word \"bringing\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.67, "bringing"], [4.12, 4.53, "bringing"]], "response_mode": "offline", "answer": "[{\"word\":\"bringing\",\"start\":2.3,\"end\":2.67},{\"word\":\"bringing\",\"start\":4.12,\"end\":4.53}]"}
{"utterance_id": "174-168635-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0015.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.44, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":1.85,\"end\":2.44}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Once the audio has finished, list each occurrence of a profession as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.24, 1.84, "one"], [11.57, 12.05, "two"], [16.03, 17.05, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.24,\"end\":1.84},{\"word\":\"two\",\"start\":11.57,\"end\":12.05},{\"word\":\"three\",\"start\":16.03,\"end\":17.05}]"}
{"utterance_id": "6319-275224-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0014.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.64, 1.37, "honeysuckle"], [3.3, 4.12, "carnation"]], "response_mode": "offline", "answer": "[{\"word\":\"honeysuckle\",\"start\":0.64,\"end\":1.37},{\"word\":\"carnation\",\"start\":3.3,\"end\":4.12}]"}
{"utterance_id": "174-168635-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0011.flac", "task_type": "lexical_detection", "query": "chosen", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"chosen\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.12, "chosen"]], "response_mode": "offline", "answer": "[{\"word\":\"chosen\",\"start\":2.65,\"end\":3.12}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.16, 5.56, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":5.16,\"end\":5.56}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "After the audio ends, tally every instance of a number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.21, 1.46, "one"], [2.46, 2.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.21,\"end\":1.46},{\"word\":\"two\",\"start\":2.46,\"end\":2.74}]"}
{"utterance_id": "1993-147964-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0004.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the full audio. After it finishes, find every instance of a date. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.09, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":1.73,\"end\":2.09}]"}
{"utterance_id": "6313-66129-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.87, 6.34, "burros"]], "response_mode": "offline", "answer": "[{\"word\":\"burros\",\"start\":5.87,\"end\":6.34}]"}
{"utterance_id": "777-126732-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0032.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.15, "eyes"], [4.34, 4.64, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":0.84,\"end\":1.15},{\"word\":\"mouth\",\"start\":4.34,\"end\":4.64}]"}
{"utterance_id": "3853-163249-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0027.flac", "task_type": "lexical_detection", "query": "enthusiasm", "prompt": "After listening to the whole recording, output every instance of the word \"enthusiasm\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.94, "enthusiasm"]], "response_mode": "offline", "answer": "[{\"word\":\"enthusiasm\",\"start\":1.97,\"end\":2.94}]"}
{"utterance_id": "652-129742-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0003.flac", "task_type": "lexical_detection", "query": "chop", "prompt": "Process the full audio offline. Find every instance of the word \"chop\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.57, "chop"]], "response_mode": "offline", "answer": "[{\"word\":\"chop\",\"start\":2.16,\"end\":2.57}]"}
{"utterance_id": "1919-142785-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0029.flac", "task_type": "lexical_detection", "query": "juice", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"juice\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 0.99, "juice"]], "response_mode": "offline", "answer": "[{\"word\":\"juice\",\"start\":0.71,\"end\":0.99}]"}
{"utterance_id": "5895-34615-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0012.flac", "task_type": "lexical_detection", "query": "absolute", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"absolute\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.74, "absolute"]], "response_mode": "offline", "answer": "[{\"word\":\"absolute\",\"start\":2.06,\"end\":2.74}]"}
{"utterance_id": "1919-142785-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0062.flac", "task_type": "lexical_detection", "query": "quantity", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"quantity\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 2.14, "quantity"]], "response_mode": "offline", "answer": "[{\"word\":\"quantity\",\"start\":1.54,\"end\":2.14}]"}
{"utterance_id": "3853-163249-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.83, 2.22, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":1.83,\"end\":2.22}]"}
{"utterance_id": "8297-275154-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0015.flac", "task_type": "lexical_counting", "query": "alone", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"alone\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.64, 1.2, "one"], [2.35, 2.93, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.64,\"end\":1.2},{\"word\":\"two\",\"start\":2.35,\"end\":2.93}]"}
{"utterance_id": "1919-142785-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0058.flac", "task_type": "lexical_detection", "query": "fried", "prompt": "Once the audio has finished, identify every instance of the word \"fried\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.56, 0.95, "fried"]], "response_mode": "offline", "answer": "[{\"word\":\"fried\",\"start\":0.56,\"end\":0.95}]"}
{"utterance_id": "2277-149874-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.38, 6.93, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":6.38,\"end\":6.93}]"}
{"utterance_id": "2412-153947-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.14, 4.44, "twelve"]], "response_mode": "offline", "answer": "[{\"word\":\"twelve\",\"start\":4.14,\"end\":4.44}]"}
{"utterance_id": "7976-105575-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0006.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.63, 5.5, "company b"], [8.24, 8.93, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"company b\",\"start\":4.63,\"end\":5.5},{\"word\":\"regiment\",\"start\":8.24,\"end\":8.93}]"}
{"utterance_id": "2428-83705-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0025.flac", "task_type": "lexical_detection", "query": "list", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"list\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.07, 1.47, "list"]], "response_mode": "offline", "answer": "[{\"word\":\"list\",\"start\":1.07,\"end\":1.47}]"}
{"utterance_id": "7976-105575-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.46, 2.75, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.46,\"end\":2.75}]"}
{"utterance_id": "2803-154328-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0001.flac", "task_type": "lexical_detection", "query": "prepare", "prompt": "Process the full audio offline. Find every instance of the word \"prepare\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.97, "prepare"]], "response_mode": "offline", "answer": "[{\"word\":\"prepare\",\"start\":2.35,\"end\":2.97}]"}
{"utterance_id": "84-121550-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0025.flac", "task_type": "lexical_counting", "query": "weep", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"weep\" with word/start/end. If none, reply [].", "labels": [[3.97, 4.22, "one"], [5.42, 5.67, "two"], [8.97, 9.37, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.97,\"end\":4.22},{\"word\":\"two\",\"start\":5.42,\"end\":5.67},{\"word\":\"three\",\"start\":8.97,\"end\":9.37}]"}
{"utterance_id": "2803-154328-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0002.flac", "task_type": "lexical_detection", "query": "ended", "prompt": "Once the audio has finished, identify every instance of the word \"ended\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.02, 1.54, "ended"]], "response_mode": "offline", "answer": "[{\"word\":\"ended\",\"start\":1.02,\"end\":1.54}]"}
{"utterance_id": "8842-304647-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.34, 5.58, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"blood\",\"start\":5.34,\"end\":5.58}]"}
{"utterance_id": "1272-141231-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.9, "brion"]], "response_mode": "offline", "answer": "[{\"word\":\"brion\",\"start\":0.52,\"end\":0.9}]"}
{"utterance_id": "1673-143397-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.24, 1.9, "arcadius"]], "response_mode": "offline", "answer": "[{\"word\":\"arcadius\",\"start\":1.24,\"end\":1.9}]"}
{"utterance_id": "5895-34629-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0011.flac", "task_type": "lexical_detection", "query": "recall", "prompt": "Process the full audio offline. Find every instance of the word \"recall\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.34, 5.87, "recall"]], "response_mode": "offline", "answer": "[{\"word\":\"recall\",\"start\":5.34,\"end\":5.87}]"}
{"utterance_id": "1673-143397-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0015.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "When the utterance is complete, report all instances of a state or province name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.57, "provinces"]], "response_mode": "offline", "answer": "[{\"word\":\"provinces\",\"start\":0.91,\"end\":1.57}]"}
{"utterance_id": "3853-163249-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0036.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.66, "wife"], [4.47, 4.9, "children"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":0.42,\"end\":0.66},{\"word\":\"children\",\"start\":4.47,\"end\":4.9}]"}
{"utterance_id": "1988-147956-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.35, 7.16, "wine colored"]], "response_mode": "offline", "answer": "[{\"word\":\"wine colored\",\"start\":6.35,\"end\":7.16}]"}
{"utterance_id": "3081-166546-0086", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0086.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.83, "next"]], "response_mode": "offline", "answer": "[{\"word\":\"next\",\"start\":0.46,\"end\":0.83}]"}
{"utterance_id": "2412-153948-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0003.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.42, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":2.08,\"end\":2.42}]"}
{"utterance_id": "7850-286674-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0010.flac", "task_type": "lexical_detection", "query": "dragon", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"dragon\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.17, "dragon"]], "response_mode": "offline", "answer": "[{\"word\":\"dragon\",\"start\":1.75,\"end\":2.17}]"}
{"utterance_id": "7976-110124-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0016.flac", "task_type": "lexical_detection", "query": "robber", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"robber\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.16, 3.5, "robber"]], "response_mode": "offline", "answer": "[{\"word\":\"robber\",\"start\":3.16,\"end\":3.5}]"}
{"utterance_id": "3536-23268-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.28, 0.93, "miss woodley"]], "response_mode": "offline", "answer": "[{\"word\":\"miss woodley\",\"start\":0.28,\"end\":0.93}]"}
{"utterance_id": "3576-138058-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0012.flac", "task_type": "lexical_detection", "query": "afraid", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"afraid\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.28, 4.68, "afraid"]], "response_mode": "offline", "answer": "[{\"word\":\"afraid\",\"start\":4.28,\"end\":4.68}]"}
{"utterance_id": "1462-170145-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.62,\"end\":1.79}]"}
{"utterance_id": "652-129742-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0004.flac", "task_type": "lexical_detection", "query": "tablespoons", "prompt": "Once the audio has finished, identify every instance of the word \"tablespoons\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.72, 1.36, "tablespoons"], [2.92, 3.51, "tablespoons"]], "response_mode": "offline", "answer": "[{\"word\":\"tablespoons\",\"start\":0.72,\"end\":1.36},{\"word\":\"tablespoons\",\"start\":2.92,\"end\":3.51}]"}
{"utterance_id": "3536-23268-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0015.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.13, 3.61, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":3.13,\"end\":3.61}]"}
{"utterance_id": "422-122949-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0006.flac", "task_type": "lexical_detection", "query": "origin", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"origin\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 2.08, "origin"]], "response_mode": "offline", "answer": "[{\"word\":\"origin\",\"start\":1.57,\"end\":2.08}]"}
{"utterance_id": "2035-147961-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0021.flac", "task_type": "lexical_counting", "query": "sledge", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"sledge\" with word/start/end. If none, reply [].", "labels": [[1.97, 2.48, "one"], [7.47, 7.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.97,\"end\":2.48},{\"word\":\"two\",\"start\":7.47,\"end\":7.77}]"}
{"utterance_id": "3536-23268-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.24, 8.49, "eye"], [16.26, 16.69, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":8.24,\"end\":8.49},{\"word\":\"hand\",\"start\":16.26,\"end\":16.69}]"}
{"utterance_id": "5895-34629-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0023.flac", "task_type": "lexical_detection", "query": "inn", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"inn\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.12, 2.44, "inn"]], "response_mode": "offline", "answer": "[{\"word\":\"inn\",\"start\":2.12,\"end\":2.44}]"}
{"utterance_id": "3000-15664-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0019.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "After the audio ends, extract all occurrences of a weather term. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.22, 9.71, "flood"]], "response_mode": "offline", "answer": "[{\"word\":\"flood\",\"start\":9.22,\"end\":9.71}]"}
{"utterance_id": "8297-275154-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.51, 1.95, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":1.51,\"end\":1.95}]"}
{"utterance_id": "422-122949-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0025.flac", "task_type": "lexical_detection", "query": "disguise", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"disguise\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.86, 6.63, "disguise"]], "response_mode": "offline", "answer": "[{\"word\":\"disguise\",\"start\":5.86,\"end\":6.63}]"}
{"utterance_id": "3081-166546-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0051.flac", "task_type": "lexical_detection", "query": "sank", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sank\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.57, 5.99, "sank"]], "response_mode": "offline", "answer": "[{\"word\":\"sank\",\"start\":5.57,\"end\":5.99}]"}
{"utterance_id": "3081-166546-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.95, 4.19, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":3.95,\"end\":4.19}]"}
{"utterance_id": "6313-66125-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0012.flac", "task_type": "lexical_detection", "query": "ready", "prompt": "Process the full audio offline. Find every instance of the word \"ready\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.92, "ready"]], "response_mode": "offline", "answer": "[{\"word\":\"ready\",\"start\":0.55,\"end\":0.92}]"}
{"utterance_id": "1462-170145-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0016.flac", "task_type": "lexical_detection", "query": "weeks", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"weeks\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.54, "weeks"]], "response_mode": "offline", "answer": "[{\"word\":\"weeks\",\"start\":2.01,\"end\":2.54}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[21.03, 21.17, "one"], [26.29, 26.9, "thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":21.03,\"end\":21.17},{\"word\":\"thousand\",\"start\":26.29,\"end\":26.9}]"}
{"utterance_id": "2078-142845-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0018.flac", "task_type": "lexical_detection", "query": "stored", "prompt": "Once the audio has finished, identify every instance of the word \"stored\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.39, 2.84, "stored"]], "response_mode": "offline", "answer": "[{\"word\":\"stored\",\"start\":2.39,\"end\":2.84}]"}
{"utterance_id": "1919-142785-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0040.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.94, "portugal"]], "response_mode": "offline", "answer": "[{\"word\":\"portugal\",\"start\":1.31,\"end\":1.94}]"}
{"utterance_id": "5895-34615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0004.flac", "task_type": "lexical_detection", "query": "countenance", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"countenance\" with fields word, start, end (seconds). If none, reply [].", "labels": [[13.69, 14.41, "countenance"]], "response_mode": "offline", "answer": "[{\"word\":\"countenance\",\"start\":13.69,\"end\":14.41}]"}
{"utterance_id": "251-136532-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0006.flac", "task_type": "lexical_detection", "query": "martians", "prompt": "After the audio ends, extract all occurrences of the word \"martians\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 2.26, "martians"]], "response_mode": "offline", "answer": "[{\"word\":\"martians\",\"start\":1.31,\"end\":2.26}]"}
{"utterance_id": "3752-4944-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0059.flac", "task_type": "lexical_detection", "query": "hadn't", "prompt": "After the audio ends, extract all occurrences of the word \"hadn't\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.29, "hadn't"]], "response_mode": "offline", "answer": "[{\"word\":\"hadn't\",\"start\":0.95,\"end\":1.29}]"}
{"utterance_id": "3752-4944-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.0, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":1.61,\"end\":2.0}]"}
{"utterance_id": "6313-66129-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0011.flac", "task_type": "lexical_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"country\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.81, 1.13, "country"]], "response_mode": "offline", "answer": "[{\"word\":\"country\",\"start\":0.81,\"end\":1.13}]"}
{"utterance_id": "652-130726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0001.flac", "task_type": "lexical_detection", "query": "restaurant", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"restaurant\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.17, "restaurant"]], "response_mode": "offline", "answer": "[{\"word\":\"restaurant\",\"start\":1.63,\"end\":2.17}]"}
{"utterance_id": "652-130726-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.63, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.47,\"end\":0.63}]"}
{"utterance_id": "251-118436-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0014.flac", "task_type": "lexical_detection", "query": "dome", "prompt": "After the audio ends, extract all occurrences of the word \"dome\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.59, "dome"]], "response_mode": "offline", "answer": "[{\"word\":\"dome\",\"start\":2.14,\"end\":2.59}]"}
{"utterance_id": "7850-281318-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0001.flac", "task_type": "lexical_detection", "query": "beginning", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"beginning\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.09, 3.62, "beginning"]], "response_mode": "offline", "answer": "[{\"word\":\"beginning\",\"start\":3.09,\"end\":3.62}]"}
{"utterance_id": "1462-170145-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0007.flac", "task_type": "lexical_detection", "query": "really", "prompt": "Once the audio has finished, identify every instance of the word \"really\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.94, 2.23, "really"]], "response_mode": "offline", "answer": "[{\"word\":\"really\",\"start\":1.94,\"end\":2.23}]"}
{"utterance_id": "6313-66125-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.57, "professor"]], "response_mode": "offline", "answer": "[{\"word\":\"professor\",\"start\":2.01,\"end\":2.57}]"}
{"utterance_id": "3000-15664-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0000.flac", "task_type": "lexical_detection", "query": "memories", "prompt": "Once the audio has finished, identify every instance of the word \"memories\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.04, 2.71, "memories"]], "response_mode": "offline", "answer": "[{\"word\":\"memories\",\"start\":2.04,\"end\":2.71}]"}
{"utterance_id": "7850-73752-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0014.flac", "task_type": "lexical_detection", "query": "ended", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ended\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.66, "ended"]], "response_mode": "offline", "answer": "[{\"word\":\"ended\",\"start\":1.28,\"end\":1.66}]"}
{"utterance_id": "3752-4944-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0050.flac", "task_type": "lexical_detection", "query": "officer", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"officer\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.07, 1.54, "officer"]], "response_mode": "offline", "answer": "[{\"word\":\"officer\",\"start\":1.07,\"end\":1.54}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 4.13, "theodosius"]], "response_mode": "offline", "answer": "[{\"word\":\"theodosius\",\"start\":3.24,\"end\":4.13}]"}
{"utterance_id": "6319-275224-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0010.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 1.28, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":0.65,\"end\":1.28}]"}
{"utterance_id": "1993-147149-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0007.flac", "task_type": "lexical_detection", "query": "breathed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"breathed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.24, "breathed"]], "response_mode": "offline", "answer": "[{\"word\":\"breathed\",\"start\":1.8,\"end\":2.24}]"}
{"utterance_id": "1988-147956-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0020.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a weather term with fields word, start, end (seconds). If none, reply [].", "labels": [[0.49, 0.79, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":0.49,\"end\":0.79}]"}
{"utterance_id": "174-84280-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0004.flac", "task_type": "lexical_detection", "query": "related", "prompt": "After listening to the whole recording, output every instance of the word \"related\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.84, 9.34, "related"]], "response_mode": "offline", "answer": "[{\"word\":\"related\",\"start\":8.84,\"end\":9.34}]"}
{"utterance_id": "2035-147961-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0013.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.4, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.04,\"end\":1.4}]"}
{"utterance_id": "3752-4944-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0066.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.26, 0.69, "twice"], [3.35, 3.86, "twice"]], "response_mode": "offline", "answer": "[{\"word\":\"twice\",\"start\":0.26,\"end\":0.69},{\"word\":\"twice\",\"start\":3.35,\"end\":3.86}]"}
{"utterance_id": "1462-170142-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0037.flac", "task_type": "lexical_detection", "query": "ask", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"ask\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.84, "ask"]], "response_mode": "offline", "answer": "[{\"word\":\"ask\",\"start\":0.46,\"end\":0.84}]"}
{"utterance_id": "777-126732-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0012.flac", "task_type": "lexical_detection", "query": "condition", "prompt": "Process the full audio offline. Find every instance of the word \"condition\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.42, "condition"]], "response_mode": "offline", "answer": "[{\"word\":\"condition\",\"start\":1.84,\"end\":2.42}]"}
{"utterance_id": "6345-93306-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0020.flac", "task_type": "lexical_detection", "query": "sighed", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sighed\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.56, 6.22, "sighed"]], "response_mode": "offline", "answer": "[{\"word\":\"sighed\",\"start\":5.56,\"end\":6.22}]"}
{"utterance_id": "6295-244435-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0007.flac", "task_type": "lexical_detection", "query": "between", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"between\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.83, 2.29, "between"]], "response_mode": "offline", "answer": "[{\"word\":\"between\",\"start\":1.83,\"end\":2.29}]"}
{"utterance_id": "6313-66125-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0021.flac", "task_type": "lexical_detection", "query": "branches", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"branches\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.36, "branches"]], "response_mode": "offline", "answer": "[{\"word\":\"branches\",\"start\":0.88,\"end\":1.36}]"}
{"utterance_id": "251-137823-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0010.flac", "task_type": "lexical_detection", "query": "supporting", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"supporting\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.56, 7.05, "supporting"]], "response_mode": "offline", "answer": "[{\"word\":\"supporting\",\"start\":6.56,\"end\":7.05}]"}
{"utterance_id": "3752-4943-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0016.flac", "task_type": "lexical_detection", "query": "good", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"good\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.7, 1.18, "good"]], "response_mode": "offline", "answer": "[{\"word\":\"good\",\"start\":0.7,\"end\":1.18}]"}
{"utterance_id": "1272-128104-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0010.flac", "task_type": "lexical_detection", "query": "ornaments", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"ornaments\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.37, "ornaments"]], "response_mode": "offline", "answer": "[{\"word\":\"ornaments\",\"start\":1.8,\"end\":2.37}]"}
{"utterance_id": "1462-170142-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0008.flac", "task_type": "lexical_detection", "query": "happen", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"happen\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.73, "happen"]], "response_mode": "offline", "answer": "[{\"word\":\"happen\",\"start\":2.26,\"end\":2.73}]"}
{"utterance_id": "6319-275224-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0014.flac", "task_type": "lexical_detection", "query": "carnation", "prompt": "Once the audio has finished, identify every instance of the word \"carnation\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.3, 4.12, "carnation"]], "response_mode": "offline", "answer": "[{\"word\":\"carnation\",\"start\":3.3,\"end\":4.12}]"}
{"utterance_id": "3081-166546-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0042.flac", "task_type": "lexical_detection", "query": "spot", "prompt": "After the audio ends, extract all occurrences of the word \"spot\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.37, 3.83, "spot"]], "response_mode": "offline", "answer": "[{\"word\":\"spot\",\"start\":3.37,\"end\":3.83}]"}
{"utterance_id": "2078-142845-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0022.flac", "task_type": "lexical_detection", "query": "rolls", "prompt": "After listening to the whole recording, output every instance of the word \"rolls\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 2.04, "rolls"]], "response_mode": "offline", "answer": "[{\"word\":\"rolls\",\"start\":1.22,\"end\":2.04}]"}
{"utterance_id": "3000-15664-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0015.flac", "task_type": "lexical_detection", "query": "spotted", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"spotted\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.25, 3.78, "spotted"]], "response_mode": "offline", "answer": "[{\"word\":\"spotted\",\"start\":3.25,\"end\":3.78}]"}
{"utterance_id": "5536-43359-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0003.flac", "task_type": "lexical_detection", "query": "seeks", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"seeks\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.53, 7.86, "seeks"]], "response_mode": "offline", "answer": "[{\"word\":\"seeks\",\"start\":7.53,\"end\":7.86}]"}
{"utterance_id": "3000-15664-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0008.flac", "task_type": "lexical_detection", "query": "invitation", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"invitation\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.19, 10.9, "invitation"]], "response_mode": "offline", "answer": "[{\"word\":\"invitation\",\"start\":10.19,\"end\":10.9}]"}
{"utterance_id": "7850-281318-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0013.flac", "task_type": "lexical_detection", "query": "show", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"show\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.2, "show"]], "response_mode": "offline", "answer": "[{\"word\":\"show\",\"start\":0.89,\"end\":1.2}]"}
{"utterance_id": "3853-163249-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0041.flac", "task_type": "lexical_detection", "query": "supported", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"supported\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.38, 4.93, "supported"]], "response_mode": "offline", "answer": "[{\"word\":\"supported\",\"start\":4.38,\"end\":4.93}]"}
{"utterance_id": "2428-83705-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.9, "mary ann"]], "response_mode": "offline", "answer": "[{\"word\":\"mary ann\",\"start\":2.06,\"end\":2.9}]"}
{"utterance_id": "652-130737-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0010.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.85, 2.18, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.85,\"end\":2.18}]"}
{"utterance_id": "251-136532-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.64, 3.03, "lunch"]], "response_mode": "offline", "answer": "[{\"word\":\"lunch\",\"start\":2.64,\"end\":3.03}]"}
{"utterance_id": "6313-66125-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0006.flac", "task_type": "lexical_detection", "query": "urged", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"urged\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.15, 2.48, "urged"]], "response_mode": "offline", "answer": "[{\"word\":\"urged\",\"start\":2.15,\"end\":2.48}]"}
{"utterance_id": "7976-105575-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0028.flac", "task_type": "lexical_detection", "query": "buried", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"buried\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.17, 3.61, "buried"]], "response_mode": "offline", "answer": "[{\"word\":\"buried\",\"start\":3.17,\"end\":3.61}]"}
{"utterance_id": "3752-4944-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.51, "meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"meekin\",\"start\":1.21,\"end\":1.51}]"}
{"utterance_id": "3081-166546-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.12, 3.67, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":3.12,\"end\":3.67}]"}
{"utterance_id": "6345-93302-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0023.flac", "task_type": "lexical_detection", "query": "decent", "prompt": "When the utterance is complete, report all instances of the word \"decent\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.09, "decent"]], "response_mode": "offline", "answer": "[{\"word\":\"decent\",\"start\":0.76,\"end\":1.09}]"}
{"utterance_id": "3081-166546-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0002.flac", "task_type": "lexical_detection", "query": "respectable", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"respectable\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.55, 2.32, "respectable"]], "response_mode": "offline", "answer": "[{\"word\":\"respectable\",\"start\":1.55,\"end\":2.32}]"}
{"utterance_id": "2902-9008-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0004.flac", "task_type": "lexical_counting", "query": "ceased", "prompt": "Process the full audio offline. For every occurrence of the word \"ceased\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.25, "one"], [3.12, 3.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.74,\"end\":1.25},{\"word\":\"two\",\"start\":3.12,\"end\":3.59}]"}
{"utterance_id": "3576-138058-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.76, "squires"]], "response_mode": "offline", "answer": "[{\"word\":\"squires\",\"start\":1.06,\"end\":1.76}]"}
{"utterance_id": "2428-83699-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0012.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After the audio ends, extract all occurrences of a date. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.85, 3.26, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":2.85,\"end\":3.26}]"}
{"utterance_id": "3170-137482-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0027.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.63, 1.15, "physician"]], "response_mode": "offline", "answer": "[{\"word\":\"physician\",\"start\":0.63,\"end\":1.15}]"}
{"utterance_id": "3081-166546-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0059.flac", "task_type": "lexical_detection", "query": "suppose", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"suppose\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.31, "suppose"]], "response_mode": "offline", "answer": "[{\"word\":\"suppose\",\"start\":0.82,\"end\":1.31}]"}
{"utterance_id": "3000-15664-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0036.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.36, 6.69, "half"], [11.16, 11.38, "few"]], "response_mode": "offline", "answer": "[{\"word\":\"half\",\"start\":6.36,\"end\":6.69},{\"word\":\"few\",\"start\":11.16,\"end\":11.38}]"}
{"utterance_id": "7976-105575-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.23, 3.63, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":3.23,\"end\":3.63}]"}
{"utterance_id": "2078-142845-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0044.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 2.02, "rusks"]], "response_mode": "offline", "answer": "[{\"word\":\"rusks\",\"start\":1.32,\"end\":2.02}]"}
{"utterance_id": "8842-304647-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0008.flac", "task_type": "lexical_detection", "query": "shame", "prompt": "Once the audio has finished, identify every instance of the word \"shame\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.85, 5.36, "shame"]], "response_mode": "offline", "answer": "[{\"word\":\"shame\",\"start\":4.85,\"end\":5.36}]"}
{"utterance_id": "6313-76958-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0026.flac", "task_type": "lexical_detection", "query": "trouble", "prompt": "When the utterance is complete, report all instances of the word \"trouble\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.29, "trouble"]], "response_mode": "offline", "answer": "[{\"word\":\"trouble\",\"start\":1.85,\"end\":2.29}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "lexical_detection", "query": "german", "prompt": "After listening to the whole recording, output every instance of the word \"german\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.49, 3.85, "german"]], "response_mode": "offline", "answer": "[{\"word\":\"german\",\"start\":3.49,\"end\":3.85}]"}
{"utterance_id": "5338-24615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0006.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[1.91, 2.46, "painter"]], "response_mode": "offline", "answer": "[{\"word\":\"painter\",\"start\":1.91,\"end\":2.46}]"}
{"utterance_id": "3170-137482-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0046.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.71, 7.96, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":7.71,\"end\":7.96}]"}
{"utterance_id": "6345-64257-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0015.flac", "task_type": "lexical_detection", "query": "waking", "prompt": "After the audio ends, extract all occurrences of the word \"waking\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 2.02, "waking"]], "response_mode": "offline", "answer": "[{\"word\":\"waking\",\"start\":1.4,\"end\":2.02}]"}
{"utterance_id": "3000-15664-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0028.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 3.05, "hand"], [5.85, 6.22, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.61,\"end\":3.05},{\"word\":\"eye\",\"start\":5.85,\"end\":6.22}]"}
{"utterance_id": "5338-24640-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0006.flac", "task_type": "lexical_detection", "query": "donald", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"donald\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.11, "donald"]], "response_mode": "offline", "answer": "[{\"word\":\"donald\",\"start\":5.59,\"end\":6.11}]"}
{"utterance_id": "3081-166546-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0050.flac", "task_type": "lexical_detection", "query": "asking", "prompt": "Process the full audio offline. Find every instance of the word \"asking\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.76, 3.37, "asking"]], "response_mode": "offline", "answer": "[{\"word\":\"asking\",\"start\":2.76,\"end\":3.37}]"}
{"utterance_id": "6295-244435-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0020.flac", "task_type": "lexical_detection", "query": "hoped", "prompt": "After listening to the whole recording, output every instance of the word \"hoped\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 5.92, "hoped"]], "response_mode": "offline", "answer": "[{\"word\":\"hoped\",\"start\":5.55,\"end\":5.92}]"}
{"utterance_id": "6241-61946-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.41, 0.62, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.41,\"end\":0.62}]"}
{"utterance_id": "777-126732-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0007.flac", "task_type": "lexical_detection", "query": "skinny", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"skinny\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.27, "skinny"]], "response_mode": "offline", "answer": "[{\"word\":\"skinny\",\"start\":2.84,\"end\":3.27}]"}
{"utterance_id": "6313-66129-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0018.flac", "task_type": "lexical_detection", "query": "start", "prompt": "Process the full audio offline. Find every instance of the word \"start\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.03, "start"]], "response_mode": "offline", "answer": "[{\"word\":\"start\",\"start\":0.75,\"end\":1.03}]"}
{"utterance_id": "1919-142785-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0032.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.82, 1.22, "wine"], [3.26, 3.75, "cayenne"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":0.82,\"end\":1.22},{\"word\":\"cayenne\",\"start\":3.26,\"end\":3.75}]"}
{"utterance_id": "8297-275155-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0010.flac", "task_type": "lexical_detection", "query": "report", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"report\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.52, 0.93, "report"]], "response_mode": "offline", "answer": "[{\"word\":\"report\",\"start\":0.52,\"end\":0.93}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.67, 3.16, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":2.67,\"end\":3.16}]"}
{"utterance_id": "1272-141231-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0022.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[5.35, 5.71, "one"], [7.47, 7.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.35,\"end\":5.71},{\"word\":\"two\",\"start\":7.47,\"end\":7.85}]"}
{"utterance_id": "1462-170142-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0039.flac", "task_type": "lexical_detection", "query": "loving", "prompt": "When the utterance is complete, report all instances of the word \"loving\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.56, "loving"]], "response_mode": "offline", "answer": "[{\"word\":\"loving\",\"start\":1.19,\"end\":1.56}]"}
{"utterance_id": "2902-9006-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0003.flac", "task_type": "lexical_detection", "query": "craving", "prompt": "After listening to the whole recording, output every instance of the word \"craving\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.56, 9.06, "craving"]], "response_mode": "offline", "answer": "[{\"word\":\"craving\",\"start\":8.56,\"end\":9.06}]"}
{"utterance_id": "5338-24615-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0014.flac", "task_type": "lexical_detection", "query": "maintained", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"maintained\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.36, 7.91, "maintained"]], "response_mode": "offline", "answer": "[{\"word\":\"maintained\",\"start\":7.36,\"end\":7.91}]"}
{"utterance_id": "1919-142785-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0041.flac", "task_type": "lexical_detection", "query": "finely", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"finely\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.58, 6.13, "finely"]], "response_mode": "offline", "answer": "[{\"word\":\"finely\",\"start\":5.58,\"end\":6.13}]"}
{"utterance_id": "84-121123-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0003.flac", "task_type": "lexical_counting", "query": "cry", "prompt": "Once the audio has finished, list each occurrence of the word \"cry\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.56, 1.06, "one"], [4.29, 4.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.56,\"end\":1.06},{\"word\":\"two\",\"start\":4.29,\"end\":4.66}]"}
{"utterance_id": "7976-110523-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.29, 6.7, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":6.29,\"end\":6.7}]"}
{"utterance_id": "6319-64726-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 2.76, "hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"hundred\",\"start\":2.34,\"end\":2.76}]"}
{"utterance_id": "1462-170138-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0017.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a person's name with word/start/end. If none, reply [].", "labels": [[0.47, 1.49, "one"], [3.32, 4.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.47,\"end\":1.49},{\"word\":\"two\",\"start\":3.32,\"end\":4.29}]"}
{"utterance_id": "3170-137482-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.47, 7.94, "side"]], "response_mode": "offline", "answer": "[{\"word\":\"side\",\"start\":7.47,\"end\":7.94}]"}
{"utterance_id": "5694-64038-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0014.flac", "task_type": "lexical_detection", "query": "gun", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"gun\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.01, "gun"]], "response_mode": "offline", "answer": "[{\"word\":\"gun\",\"start\":0.72,\"end\":1.01}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.37, "leaven"]], "response_mode": "offline", "answer": "[{\"word\":\"leaven\",\"start\":0.94,\"end\":1.37}]"}
{"utterance_id": "2412-153948-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0010.flac", "task_type": "lexical_detection", "query": "imagined", "prompt": "Once the audio has finished, identify every instance of the word \"imagined\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.11, 2.74, "imagined"]], "response_mode": "offline", "answer": "[{\"word\":\"imagined\",\"start\":2.11,\"end\":2.74}]"}
{"utterance_id": "5338-284437-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0005.flac", "task_type": "lexical_detection", "query": "peculiar", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"peculiar\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.21, 2.86, "peculiar"]], "response_mode": "offline", "answer": "[{\"word\":\"peculiar\",\"start\":2.21,\"end\":2.86}]"}
{"utterance_id": "3576-138058-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.54, "sancho panza"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho panza\",\"start\":5.59,\"end\":6.54}]"}
{"utterance_id": "652-129742-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0004.flac", "task_type": "lexical_counting", "query": "tablespoons", "prompt": "When the utterance is complete, report how the word \"tablespoons\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.36, "one"], [2.92, 3.51, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.72,\"end\":1.36},{\"word\":\"two\",\"start\":2.92,\"end\":3.51}]"}
{"utterance_id": "2803-161169-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.79, 6.17, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.79,\"end\":6.17}]"}
{"utterance_id": "8842-304647-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0002.flac", "task_type": "lexical_counting", "query": "dead", "prompt": "After listening to the whole recording, output timed count words for the word \"dead\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[4.78, 5.13, "one"], [23.45, 23.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.78,\"end\":5.13},{\"word\":\"two\",\"start\":23.45,\"end\":23.69}]"}
{"utterance_id": "1462-170138-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0001.flac", "task_type": "lexical_detection", "query": "inexpressible", "prompt": "Once the audio has finished, identify every instance of the word \"inexpressible\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.82, 3.71, "inexpressible"]], "response_mode": "offline", "answer": "[{\"word\":\"inexpressible\",\"start\":2.82,\"end\":3.71}]"}
{"utterance_id": "6295-244435-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0036.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.97, 1.31, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.97,\"end\":1.31}]"}
{"utterance_id": "6241-61943-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0013.flac", "task_type": "lexical_detection", "query": "remark", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"remark\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.94, "remark"]], "response_mode": "offline", "answer": "[{\"word\":\"remark\",\"start\":1.35,\"end\":1.94}]"}
{"utterance_id": "1919-142785-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0061.flac", "task_type": "lexical_counting", "query": "onion", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"onion\" with word/start/end. If none, reply [].", "labels": [[2.35, 2.65, "one"], [15.83, 16.14, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.35,\"end\":2.65},{\"word\":\"two\",\"start\":15.83,\"end\":16.14}]"}
{"utterance_id": "1272-141231-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0014.flac", "task_type": "lexical_detection", "query": "fool", "prompt": "After the audio ends, extract all occurrences of the word \"fool\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.44, 5.87, "fool"]], "response_mode": "offline", "answer": "[{\"word\":\"fool\",\"start\":5.44,\"end\":5.87}]"}
{"utterance_id": "1988-147956-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0001.flac", "task_type": "lexical_detection", "query": "tear", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"tear\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.0, "tear"]], "response_mode": "offline", "answer": "[{\"word\":\"tear\",\"start\":2.72,\"end\":3.0}]"}
{"utterance_id": "5338-284437-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.4, "trot"]], "response_mode": "offline", "answer": "[{\"word\":\"trot\",\"start\":1.02,\"end\":1.4}]"}
{"utterance_id": "7976-110124-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0000.flac", "task_type": "lexical_detection", "query": "ago", "prompt": "Once the audio has finished, identify every instance of the word \"ago\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.87, 1.23, "ago"]], "response_mode": "offline", "answer": "[{\"word\":\"ago\",\"start\":0.87,\"end\":1.23}]"}
{"utterance_id": "1462-170142-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0013.flac", "task_type": "lexical_detection", "query": "miserable", "prompt": "Process the full audio offline. Find every instance of the word \"miserable\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.96, "miserable"]], "response_mode": "offline", "answer": "[{\"word\":\"miserable\",\"start\":2.38,\"end\":2.96}]"}
{"utterance_id": "5536-43358-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0014.flac", "task_type": "lexical_detection", "query": "prayer", "prompt": "When the utterance is complete, report all instances of the word \"prayer\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.4, "prayer"]], "response_mode": "offline", "answer": "[{\"word\":\"prayer\",\"start\":1.9,\"end\":2.4}]"}
{"utterance_id": "3081-166546-0074", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0074.flac", "task_type": "lexical_detection", "query": "running", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"running\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.93, 9.28, "running"]], "response_mode": "offline", "answer": "[{\"word\":\"running\",\"start\":8.93,\"end\":9.28}]"}
{"utterance_id": "6295-64301-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.8, 5.44, "musician"]], "response_mode": "offline", "answer": "[{\"word\":\"musician\",\"start\":4.8,\"end\":5.44}]"}
{"utterance_id": "3853-163249-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0039.flac", "task_type": "lexical_detection", "query": "clasped", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"clasped\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.84, 3.3, "clasped"]], "response_mode": "offline", "answer": "[{\"word\":\"clasped\",\"start\":2.84,\"end\":3.3}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "lexical_counting", "query": "divide", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"divide\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[2.06, 2.48, "one"], [21.13, 21.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.48},{\"word\":\"two\",\"start\":21.13,\"end\":21.57}]"}
{"utterance_id": "251-137823-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0023.flac", "task_type": "lexical_detection", "query": "great", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"great\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 1.03, "great"]], "response_mode": "offline", "answer": "[{\"word\":\"great\",\"start\":0.65,\"end\":1.03}]"}
{"utterance_id": "2428-83705-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0041.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.22, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":0.76,\"end\":1.22}]"}
{"utterance_id": "8842-302196-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0006.flac", "task_type": "lexical_detection", "query": "poems", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"poems\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.21, 4.55, "poems"]], "response_mode": "offline", "answer": "[{\"word\":\"poems\",\"start\":4.21,\"end\":4.55}]"}
{"utterance_id": "3576-138058-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 3.88, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":3.48,\"end\":3.88}]"}
{"utterance_id": "8297-275154-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0015.flac", "task_type": "lexical_detection", "query": "hear", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hear\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 1.79, "hear"]], "response_mode": "offline", "answer": "[{\"word\":\"hear\",\"start\":1.56,\"end\":1.79}]"}
{"utterance_id": "3576-138058-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.16, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":0.51,\"end\":1.16}]"}
{"utterance_id": "3536-8226-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0005.flac", "task_type": "lexical_detection", "query": "exertions", "prompt": "When the utterance is complete, report all instances of the word \"exertions\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.84, 4.5, "exertions"]], "response_mode": "offline", "answer": "[{\"word\":\"exertions\",\"start\":3.84,\"end\":4.5}]"}
{"utterance_id": "6313-66129-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0031.flac", "task_type": "lexical_detection", "query": "chunky", "prompt": "Process the full audio offline. Find every instance of the word \"chunky\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.85, 3.23, "chunky"]], "response_mode": "offline", "answer": "[{\"word\":\"chunky\",\"start\":2.85,\"end\":3.23}]"}
{"utterance_id": "1919-142785-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0040.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "After the audio ends, tally every instance of a plant name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.81, 3.16, "one"], [4.61, 5.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.81,\"end\":3.16},{\"word\":\"two\",\"start\":4.61,\"end\":5.0}]"}
{"utterance_id": "7850-281318-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0023.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.21, "magpie"]], "response_mode": "offline", "answer": "[{\"word\":\"magpie\",\"start\":3.78,\"end\":4.21}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Once the audio has finished, list each occurrence of a profession as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.62, 1.06, "one"], [2.14, 2.6, "two"], [6.97, 7.46, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.62,\"end\":1.06},{\"word\":\"two\",\"start\":2.14,\"end\":2.6},{\"word\":\"three\",\"start\":6.97,\"end\":7.46}]"}
{"utterance_id": "7850-73752-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.07, 4.66, "ferdinand"]], "response_mode": "offline", "answer": "[{\"word\":\"ferdinand\",\"start\":4.07,\"end\":4.66}]"}
{"utterance_id": "6319-57405-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0007.flac", "task_type": "lexical_detection", "query": "anything", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"anything\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.89, 1.32, "anything"]], "response_mode": "offline", "answer": "[{\"word\":\"anything\",\"start\":0.89,\"end\":1.32}]"}
{"utterance_id": "1462-170138-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0002.flac", "task_type": "lexical_detection", "query": "mac", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mac\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.15, 1.25, "mac"]], "response_mode": "offline", "answer": "[{\"word\":\"mac\",\"start\":1.15,\"end\":1.25}]"}
{"utterance_id": "251-118436-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0019.flac", "task_type": "lexical_detection", "query": "summoned", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"summoned\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.65, 9.08, "summoned"]], "response_mode": "offline", "answer": "[{\"word\":\"summoned\",\"start\":8.65,\"end\":9.08}]"}
{"utterance_id": "5895-34615-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.72, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":2.18,\"end\":2.72}]"}
{"utterance_id": "3853-163249-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0047.flac", "task_type": "lexical_detection", "query": "braver", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"braver\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.19, 7.64, "braver"]], "response_mode": "offline", "answer": "[{\"word\":\"braver\",\"start\":7.19,\"end\":7.64}]"}
{"utterance_id": "2277-149897-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0013.flac", "task_type": "lexical_detection", "query": "conversation", "prompt": "After listening to the whole recording, output every instance of the word \"conversation\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.69, "conversation"]], "response_mode": "offline", "answer": "[{\"word\":\"conversation\",\"start\":1.96,\"end\":2.69}]"}
{"utterance_id": "1272-135031-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0010.flac", "task_type": "lexical_detection", "query": "dominions", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"dominions\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 3.3, "dominions"]], "response_mode": "offline", "answer": "[{\"word\":\"dominions\",\"start\":2.66,\"end\":3.3}]"}
{"utterance_id": "6295-64301-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0014.flac", "task_type": "lexical_detection", "query": "sat", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sat\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 0.9, "sat"]], "response_mode": "offline", "answer": "[{\"word\":\"sat\",\"start\":0.64,\"end\":0.9}]"}
{"utterance_id": "3853-163249-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.66, 7.13, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":6.66,\"end\":7.13}]"}
{"utterance_id": "1919-142785-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0062.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a number with word/start/end. If none, reply [].", "labels": [[1.06, 1.34, "one"], [2.41, 2.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.06,\"end\":1.34},{\"word\":\"two\",\"start\":2.41,\"end\":2.65}]"}
{"utterance_id": "3853-163249-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0052.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[11.05, 11.6, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":11.05,\"end\":11.6}]"}
{"utterance_id": "1988-24833-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.09,\"end\":2.31}]"}
{"utterance_id": "6313-76958-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0007.flac", "task_type": "lexical_detection", "query": "we've", "prompt": "Once the audio has finished, identify every instance of the word \"we've\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.22, 0.4, "we've"]], "response_mode": "offline", "answer": "[{\"word\":\"we've\",\"start\":0.22,\"end\":0.4}]"}
{"utterance_id": "6313-66125-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0027.flac", "task_type": "lexical_detection", "query": "circles", "prompt": "Process the full audio offline. Find every instance of the word \"circles\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.13, 11.63, "circles"]], "response_mode": "offline", "answer": "[{\"word\":\"circles\",\"start\":11.13,\"end\":11.63}]"}
{"utterance_id": "5536-43363-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0016.flac", "task_type": "lexical_detection", "query": "christian", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"christian\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.51, 2.95, "christian"]], "response_mode": "offline", "answer": "[{\"word\":\"christian\",\"start\":2.51,\"end\":2.95}]"}
{"utterance_id": "8297-275155-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0022.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.41, "mother's"], [5.43, 5.91, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"mother's\",\"start\":2.01,\"end\":2.41},{\"word\":\"father\",\"start\":5.43,\"end\":5.91}]"}
{"utterance_id": "1993-147149-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0007.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.31, "child"], [2.32, 2.65, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"child\",\"start\":0.89,\"end\":1.31},{\"word\":\"father\",\"start\":2.32,\"end\":2.65}]"}
{"utterance_id": "8842-302196-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0002.flac", "task_type": "lexical_detection", "query": "turned", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"turned\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.36, 5.6, "turned"]], "response_mode": "offline", "answer": "[{\"word\":\"turned\",\"start\":5.36,\"end\":5.6}]"}
{"utterance_id": "1272-141231-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0004.flac", "task_type": "lexical_detection", "query": "buzzer", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"buzzer\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.75, 4.05, "buzzer"]], "response_mode": "offline", "answer": "[{\"word\":\"buzzer\",\"start\":3.75,\"end\":4.05}]"}
{"utterance_id": "6313-76958-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0021.flac", "task_type": "lexical_detection", "query": "sleepy", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sleepy\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.01, 5.43, "sleepy"]], "response_mode": "offline", "answer": "[{\"word\":\"sleepy\",\"start\":5.01,\"end\":5.43}]"}
{"utterance_id": "8297-275155-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0001.flac", "task_type": "lexical_detection", "query": "published", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"published\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.99, 4.54, "published"]], "response_mode": "offline", "answer": "[{\"word\":\"published\",\"start\":3.99,\"end\":4.54}]"}
{"utterance_id": "3081-166546-0060", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0060.flac", "task_type": "lexical_detection", "query": "speaks", "prompt": "Process the full audio offline. Find every instance of the word \"speaks\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.46, "speaks"]], "response_mode": "offline", "answer": "[{\"word\":\"speaks\",\"start\":1.06,\"end\":1.46}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "lexical_detection", "query": "special", "prompt": "When the utterance is complete, report all instances of the word \"special\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.08, "special"]], "response_mode": "offline", "answer": "[{\"word\":\"special\",\"start\":0.71,\"end\":1.08}]"}
{"utterance_id": "2902-9008-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0012.flac", "task_type": "lexical_counting", "query": "men", "prompt": "Listen to the entire clip, then report the count sequence for the word \"men\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.64, "one"], [3.55, 3.98, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.39,\"end\":1.64},{\"word\":\"two\",\"start\":3.55,\"end\":3.98}]"}
{"utterance_id": "5536-43358-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 2.42, "sons"]], "response_mode": "offline", "answer": "[{\"word\":\"sons\",\"start\":2.13,\"end\":2.42}]"}
{"utterance_id": "2902-9008-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.09, 17.52, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":17.09,\"end\":17.52}]"}
{"utterance_id": "1993-147965-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.11, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":1.67,\"end\":2.11}]"}
{"utterance_id": "2035-147961-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0022.flac", "task_type": "lexical_detection", "query": "shrieks", "prompt": "Process the full audio offline. Find every instance of the word \"shrieks\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 1.1, "shrieks"]], "response_mode": "offline", "answer": "[{\"word\":\"shrieks\",\"start\":0.56,\"end\":1.1}]"}
{"utterance_id": "8297-275156-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0000.flac", "task_type": "lexical_detection", "query": "doing", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"doing\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.59, "doing"]], "response_mode": "offline", "answer": "[{\"word\":\"doing\",\"start\":1.21,\"end\":1.59}]"}
{"utterance_id": "1462-170142-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0031.flac", "task_type": "lexical_detection", "query": "listened", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"listened\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.02, "listened"]], "response_mode": "offline", "answer": "[{\"word\":\"listened\",\"start\":0.66,\"end\":1.02}]"}
{"utterance_id": "6295-244435-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0030.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.85, 4.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.85,\"end\":4.08}]"}
{"utterance_id": "3536-23268-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.17, 0.81, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss milner\",\"start\":0.17,\"end\":0.81}]"}
{"utterance_id": "6313-66125-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.24,\"end\":3.43}]"}
{"utterance_id": "6313-66129-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.99, 2.37, "ned"]], "response_mode": "offline", "answer": "[{\"word\":\"ned\",\"start\":1.99,\"end\":2.37}]"}
{"utterance_id": "251-137823-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0002.flac", "task_type": "lexical_detection", "query": "isn't", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"isn't\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.45, 0.7, "isn't"]], "response_mode": "offline", "answer": "[{\"word\":\"isn't\",\"start\":0.45,\"end\":0.7}]"}
{"utterance_id": "2086-149220-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0014.flac", "task_type": "lexical_detection", "query": "begun", "prompt": "Process the full audio offline. Find every instance of the word \"begun\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.83, 5.15, "begun"]], "response_mode": "offline", "answer": "[{\"word\":\"begun\",\"start\":4.83,\"end\":5.15}]"}
{"utterance_id": "2086-149220-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0036.flac", "task_type": "lexical_detection", "query": "nonsense", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"nonsense\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.75, 1.33, "nonsense"]], "response_mode": "offline", "answer": "[{\"word\":\"nonsense\",\"start\":0.75,\"end\":1.33}]"}
{"utterance_id": "3853-163249-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0055.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.76, 3.02, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.76,\"end\":3.02}]"}
{"utterance_id": "6313-76958-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0031.flac", "task_type": "lexical_detection", "query": "drive", "prompt": "Process the full audio offline. Find every instance of the word \"drive\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.46, "drive"]], "response_mode": "offline", "answer": "[{\"word\":\"drive\",\"start\":1.16,\"end\":1.46}]"}
{"utterance_id": "2428-83705-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0023.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.39, "mamma"]], "response_mode": "offline", "answer": "[{\"word\":\"mamma\",\"start\":1.03,\"end\":1.39}]"}
{"utterance_id": "1993-147964-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0003.flac", "task_type": "lexical_detection", "query": "cotton", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cotton\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.68, "cotton"]], "response_mode": "offline", "answer": "[{\"word\":\"cotton\",\"start\":1.34,\"end\":1.68}]"}
{"utterance_id": "2035-152373-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0016.flac", "task_type": "lexical_detection", "query": "tribute", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"tribute\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 1.0, "tribute"]], "response_mode": "offline", "answer": "[{\"word\":\"tribute\",\"start\":0.58,\"end\":1.0}]"}
{"utterance_id": "2086-149220-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0048.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.0, 6.43, "heart"], [7.98, 8.34, "bosom"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":6.0,\"end\":6.43},{\"word\":\"bosom\",\"start\":7.98,\"end\":8.34}]"}
{"utterance_id": "6313-76958-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0002.flac", "task_type": "lexical_detection", "query": "demanded", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"demanded\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.9, 2.3, "demanded"]], "response_mode": "offline", "answer": "[{\"word\":\"demanded\",\"start\":1.9,\"end\":2.3}]"}
{"utterance_id": "251-137823-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0012.flac", "task_type": "lexical_detection", "query": "friend's", "prompt": "After the audio ends, extract all occurrences of the word \"friend's\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.94, "friend's"]], "response_mode": "offline", "answer": "[{\"word\":\"friend's\",\"start\":0.51,\"end\":0.94}]"}
{"utterance_id": "2035-147961-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0035.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.73, 3.54, "chicago des moines fort wayne"]], "response_mode": "offline", "answer": "[{\"word\":\"chicago des moines fort wayne\",\"start\":0.73,\"end\":3.54}]"}
{"utterance_id": "6319-275224-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0002.flac", "task_type": "lexical_detection", "query": "convolvulus", "prompt": "After the audio ends, extract all occurrences of the word \"convolvulus\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.99, 2.87, "convolvulus"]], "response_mode": "offline", "answer": "[{\"word\":\"convolvulus\",\"start\":1.99,\"end\":2.87}]"}
{"utterance_id": "2277-149874-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0010.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Once the audio has finished, identify every instance of a country name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.64, 5.06, "swedish"]], "response_mode": "offline", "answer": "[{\"word\":\"swedish\",\"start\":4.64,\"end\":5.06}]"}
{"utterance_id": "251-137823-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0026.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.57, "mister swift's"]], "response_mode": "offline", "answer": "[{\"word\":\"mister swift's\",\"start\":0.51,\"end\":1.57}]"}
{"utterance_id": "2902-9008-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0004.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.67, 2.3, "nations"]], "response_mode": "offline", "answer": "[{\"word\":\"nations\",\"start\":1.67,\"end\":2.3}]"}
{"utterance_id": "652-129742-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0010.flac", "task_type": "lexical_detection", "query": "while", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"while\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.9, 2.06, "while"]], "response_mode": "offline", "answer": "[{\"word\":\"while\",\"start\":1.9,\"end\":2.06}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "lexical_counting", "query": "leaf", "prompt": "After listening to the whole recording, output timed count words for the word \"leaf\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[8.81, 9.14, "one"], [21.07, 21.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.81,\"end\":9.14},{\"word\":\"two\",\"start\":21.07,\"end\":21.3}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "lexical_detection", "query": "apples", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"apples\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.34, "apples"]], "response_mode": "offline", "answer": "[{\"word\":\"apples\",\"start\":2.87,\"end\":3.34}]"}
{"utterance_id": "3752-4944-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0068.flac", "task_type": "lexical_counting", "query": "man", "prompt": "When the utterance is complete, report how the word \"man\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.61, 7.97, "one"], [10.2, 10.52, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.61,\"end\":7.97},{\"word\":\"two\",\"start\":10.2,\"end\":10.52}]"}
{"utterance_id": "6345-64257-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0001.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.93, 6.28, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.93,\"end\":6.28}]"}
{"utterance_id": "422-122949-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0009.flac", "task_type": "lexical_counting", "query": "others", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"others\" with word/start/end. If none, reply [].", "labels": [[4.89, 5.44, "one"], [16.41, 16.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.89,\"end\":5.44},{\"word\":\"two\",\"start\":16.41,\"end\":16.79}]"}
{"utterance_id": "652-129742-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0019.flac", "task_type": "lexical_detection", "query": "cool", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"cool\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.89, 2.28, "cool"]], "response_mode": "offline", "answer": "[{\"word\":\"cool\",\"start\":1.89,\"end\":2.28}]"}
{"utterance_id": "7850-111771-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.97, 1.5, "commander"]], "response_mode": "offline", "answer": "[{\"word\":\"commander\",\"start\":0.97,\"end\":1.5}]"}
{"utterance_id": "6319-57405-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.86, 3.55, "deucalion"]], "response_mode": "offline", "answer": "[{\"word\":\"deucalion\",\"start\":2.86,\"end\":3.55}]"}
{"utterance_id": "7976-110124-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0003.flac", "task_type": "lexical_detection", "query": "merchant's", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"merchant's\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.27, 6.7, "merchant's"]], "response_mode": "offline", "answer": "[{\"word\":\"merchant's\",\"start\":6.27,\"end\":6.7}]"}
{"utterance_id": "2902-9008-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0008.flac", "task_type": "lexical_detection", "query": "care", "prompt": "When the utterance is complete, report all instances of the word \"care\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.27, "care"]], "response_mode": "offline", "answer": "[{\"word\":\"care\",\"start\":0.97,\"end\":1.27}]"}
{"utterance_id": "1462-170142-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0040.flac", "task_type": "lexical_detection", "query": "difference", "prompt": "When the utterance is complete, report all instances of the word \"difference\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.46, 4.11, "difference"]], "response_mode": "offline", "answer": "[{\"word\":\"difference\",\"start\":3.46,\"end\":4.11}]"}
{"utterance_id": "2803-161169-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0016.flac", "task_type": "lexical_detection", "query": "refuse", "prompt": "When the utterance is complete, report all instances of the word \"refuse\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 4.37, "refuse"]], "response_mode": "offline", "answer": "[{\"word\":\"refuse\",\"start\":3.59,\"end\":4.37}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[27.83, 28.36, "brethren"]], "response_mode": "offline", "answer": "[{\"word\":\"brethren\",\"start\":27.83,\"end\":28.36}]"}
{"utterance_id": "7850-286674-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 2.63, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":2.34,\"end\":2.63}]"}
{"utterance_id": "6313-76958-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.59, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.14,\"end\":1.59}]"}
{"utterance_id": "2078-142845-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0032.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.87, "soyer"]], "response_mode": "offline", "answer": "[{\"word\":\"soyer\",\"start\":0.33,\"end\":0.87}]"}
{"utterance_id": "84-121550-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0000.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.71, 6.08, "leaves"]], "response_mode": "offline", "answer": "[{\"word\":\"leaves\",\"start\":5.71,\"end\":6.08}]"}
{"utterance_id": "6313-76958-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0004.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.94, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":0.57,\"end\":0.94}]"}
{"utterance_id": "7850-111771-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 3.9, "hon thomas l hamer"]], "response_mode": "offline", "answer": "[{\"word\":\"hon thomas l hamer\",\"start\":1.76,\"end\":3.9}]"}
{"utterance_id": "251-118436-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0004.flac", "task_type": "lexical_detection", "query": "confused", "prompt": "Process the full audio offline. Find every instance of the word \"confused\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.48, "confused"]], "response_mode": "offline", "answer": "[{\"word\":\"confused\",\"start\":0.86,\"end\":1.48}]"}
{"utterance_id": "6313-66129-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0034.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Once the audio has finished, list each occurrence of a body part as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.69, 1.89, "one"], [3.05, 3.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.69,\"end\":1.89},{\"word\":\"two\",\"start\":3.05,\"end\":3.35}]"}
{"utterance_id": "1673-143397-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0009.flac", "task_type": "lexical_detection", "query": "egypt", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"egypt\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.89, 3.35, "egypt"]], "response_mode": "offline", "answer": "[{\"word\":\"egypt\",\"start\":2.89,\"end\":3.35}]"}
{"utterance_id": "6241-61946-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0019.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 0.95, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":0.66,\"end\":0.95}]"}
{"utterance_id": "422-122949-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0026.flac", "task_type": "lexical_detection", "query": "purity", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"purity\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.91, "purity"]], "response_mode": "offline", "answer": "[{\"word\":\"purity\",\"start\":1.42,\"end\":1.91}]"}
{"utterance_id": "3536-8226-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0013.flac", "task_type": "lexical_detection", "query": "won't", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"won't\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.03, "won't"]], "response_mode": "offline", "answer": "[{\"word\":\"won't\",\"start\":1.75,\"end\":2.03}]"}
{"utterance_id": "251-136532-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 2.2, "zoologist"]], "response_mode": "offline", "answer": "[{\"word\":\"zoologist\",\"start\":1.45,\"end\":2.2}]"}
{"utterance_id": "6319-64726-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0012.flac", "task_type": "lexical_detection", "query": "drawn", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"drawn\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.12, 4.45, "drawn"]], "response_mode": "offline", "answer": "[{\"word\":\"drawn\",\"start\":4.12,\"end\":4.45}]"}
{"utterance_id": "3170-137482-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0042.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[10.11, 10.84, "jesuits"]], "response_mode": "offline", "answer": "[{\"word\":\"jesuits\",\"start\":10.11,\"end\":10.84}]"}
{"utterance_id": "6295-244435-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0005.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.38, "sumter"]], "response_mode": "offline", "answer": "[{\"word\":\"sumter\",\"start\":2.95,\"end\":3.38}]"}
{"utterance_id": "1988-24833-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0017.flac", "task_type": "lexical_detection", "query": "goes", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"goes\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.1, 4.34, "goes"]], "response_mode": "offline", "answer": "[{\"word\":\"goes\",\"start\":4.1,\"end\":4.34}]"}
{"utterance_id": "2086-149220-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0007.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.48, "wives"]], "response_mode": "offline", "answer": "[{\"word\":\"wives\",\"start\":2.95,\"end\":3.48}]"}
{"utterance_id": "2428-83699-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0008.flac", "task_type": "lexical_counting", "query": "name", "prompt": "After listening to the whole recording, output timed count words for the word \"name\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[4.93, 5.15, "one"], [7.72, 8.17, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.93,\"end\":5.15},{\"word\":\"two\",\"start\":7.72,\"end\":8.17}]"}
{"utterance_id": "5536-43358-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0015.flac", "task_type": "lexical_detection", "query": "beast", "prompt": "After listening to the whole recording, output every instance of the word \"beast\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.45, "beast"]], "response_mode": "offline", "answer": "[{\"word\":\"beast\",\"start\":3.15,\"end\":3.45}]"}
{"utterance_id": "1272-128104-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 3.16, "mister quilter"]], "response_mode": "offline", "answer": "[{\"word\":\"mister quilter\",\"start\":2.3,\"end\":3.16}]"}
{"utterance_id": "174-168635-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0002.flac", "task_type": "lexical_counting", "query": "find", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"find\" (word, start, end). If none, reply [].", "labels": [[10.91, 11.24, "one"], [13.48, 13.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.91,\"end\":11.24},{\"word\":\"two\",\"start\":13.48,\"end\":13.74}]"}
{"utterance_id": "422-122949-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0015.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "When the utterance is complete, report how a family relation accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.56, 2.91, "one"], [10.69, 11.22, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.56,\"end\":2.91},{\"word\":\"two\",\"start\":10.69,\"end\":11.22}]"}
{"utterance_id": "2277-149896-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0005.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the entire clip, then report the count sequence for a body part as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.59, "one"], [4.13, 4.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.25,\"end\":2.59},{\"word\":\"two\",\"start\":4.13,\"end\":4.37}]"}
{"utterance_id": "3752-4943-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0006.flac", "task_type": "lexical_detection", "query": "north", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"north\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.53, "north"]], "response_mode": "offline", "answer": "[{\"word\":\"north\",\"start\":2.04,\"end\":2.53}]"}
{"utterance_id": "8842-302196-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0010.flac", "task_type": "lexical_detection", "query": "remote", "prompt": "Once the audio has finished, identify every instance of the word \"remote\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.16, 5.46, "remote"]], "response_mode": "offline", "answer": "[{\"word\":\"remote\",\"start\":5.16,\"end\":5.46}]"}
{"utterance_id": "6319-275224-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0008.flac", "task_type": "lexical_detection", "query": "extended", "prompt": "After listening to the whole recording, output every instance of the word \"extended\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.34, 8.03, "extended"]], "response_mode": "offline", "answer": "[{\"word\":\"extended\",\"start\":7.34,\"end\":8.03}]"}
{"utterance_id": "1272-135031-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0023.flac", "task_type": "lexical_detection", "query": "gong", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"gong\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.27, "gong"]], "response_mode": "offline", "answer": "[{\"word\":\"gong\",\"start\":1.61,\"end\":2.27}]"}
{"utterance_id": "1988-147956-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.43, "antonia"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia\",\"start\":0.88,\"end\":1.43}]"}
{"utterance_id": "2035-147960-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0014.flac", "task_type": "lexical_detection", "query": "fight", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"fight\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.37, 3.6, "fight"]], "response_mode": "offline", "answer": "[{\"word\":\"fight\",\"start\":3.37,\"end\":3.6}]"}
{"utterance_id": "174-50561-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0005.flac", "task_type": "lexical_detection", "query": "roundel", "prompt": "After the audio ends, extract all occurrences of the word \"roundel\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.56, 4.99, "roundel"]], "response_mode": "offline", "answer": "[{\"word\":\"roundel\",\"start\":4.56,\"end\":4.99}]"}
{"utterance_id": "2902-9008-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.7, 5.16, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":4.7,\"end\":5.16}]"}
{"utterance_id": "3170-137482-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0039.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Process the full audio offline. Find every instance of a country name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[19.28, 19.83, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"europe\",\"start\":19.28,\"end\":19.83}]"}
{"utterance_id": "5338-24640-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.74, 2.64, "mister waverley"]], "response_mode": "offline", "answer": "[{\"word\":\"mister waverley\",\"start\":1.74,\"end\":2.64}]"}
{"utterance_id": "3000-15664-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0024.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[6.68, 7.01, "sage"]], "response_mode": "offline", "answer": "[{\"word\":\"sage\",\"start\":6.68,\"end\":7.01}]"}
{"utterance_id": "2086-149214-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0000.flac", "task_type": "lexical_detection", "query": "narrative", "prompt": "After the audio ends, extract all occurrences of the word \"narrative\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.17, "narrative"]], "response_mode": "offline", "answer": "[{\"word\":\"narrative\",\"start\":0.5,\"end\":1.17}]"}
{"utterance_id": "2078-142845-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0041.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.53, 2.28, "buns"]], "response_mode": "offline", "answer": "[{\"word\":\"buns\",\"start\":1.53,\"end\":2.28}]"}
{"utterance_id": "3853-163249-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.08, "sister"]], "response_mode": "offline", "answer": "[{\"word\":\"sister\",\"start\":1.69,\"end\":2.08}]"}
{"utterance_id": "3170-137482-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0010.flac", "task_type": "lexical_detection", "query": "contain", "prompt": "Once the audio has finished, identify every instance of the word \"contain\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.52, 2.91, "contain"]], "response_mode": "offline", "answer": "[{\"word\":\"contain\",\"start\":2.52,\"end\":2.91}]"}
{"utterance_id": "5694-64025-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0019.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.67, 4.13, "mule"]], "response_mode": "offline", "answer": "[{\"word\":\"mule\",\"start\":3.67,\"end\":4.13}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[6.19, 6.67, "professor's"]], "response_mode": "offline", "answer": "[{\"word\":\"professor's\",\"start\":6.19,\"end\":6.67}]"}
{"utterance_id": "2277-149896-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0007.flac", "task_type": "lexical_detection", "query": "morrow", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"morrow\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.26, "morrow"]], "response_mode": "offline", "answer": "[{\"word\":\"morrow\",\"start\":1.8,\"end\":2.26}]"}
{"utterance_id": "3000-15664-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0045.flac", "task_type": "lexical_detection", "query": "invaded", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"invaded\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.3, 6.86, "invaded"]], "response_mode": "offline", "answer": "[{\"word\":\"invaded\",\"start\":6.3,\"end\":6.86}]"}
{"utterance_id": "6241-61943-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0023.flac", "task_type": "lexical_detection", "query": "likely", "prompt": "Process the full audio offline. Find every instance of the word \"likely\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.41, 2.74, "likely"]], "response_mode": "offline", "answer": "[{\"word\":\"likely\",\"start\":2.41,\"end\":2.74}]"}
{"utterance_id": "2412-153947-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.38, 5.55, "two"], [8.45, 8.6, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":5.38,\"end\":5.55},{\"word\":\"one\",\"start\":8.45,\"end\":8.6}]"}
{"utterance_id": "6313-66129-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0000.flac", "task_type": "lexical_detection", "query": "bring", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"bring\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.06, 1.32, "bring"]], "response_mode": "offline", "answer": "[{\"word\":\"bring\",\"start\":1.06,\"end\":1.32}]"}
{"utterance_id": "2902-9006-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0000.flac", "task_type": "lexical_detection", "query": "era", "prompt": "Once the audio has finished, identify every instance of the word \"era\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.71, 2.04, "era"]], "response_mode": "offline", "answer": "[{\"word\":\"era\",\"start\":1.71,\"end\":2.04}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.62, 1.06, "regent's"], [2.14, 2.6, "servants"], [6.97, 7.46, "captains"]], "response_mode": "offline", "answer": "[{\"word\":\"regent's\",\"start\":0.62,\"end\":1.06},{\"word\":\"servants\",\"start\":2.14,\"end\":2.6},{\"word\":\"captains\",\"start\":6.97,\"end\":7.46}]"}
{"utterance_id": "7850-73752-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0012.flac", "task_type": "lexical_detection", "query": "prove", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"prove\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.65, "prove"]], "response_mode": "offline", "answer": "[{\"word\":\"prove\",\"start\":2.43,\"end\":2.65}]"}
{"utterance_id": "5536-43363-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.25, 6.58, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":6.25,\"end\":6.58}]"}
{"utterance_id": "5536-43359-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0017.flac", "task_type": "lexical_detection", "query": "mark", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mark\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.08, 4.4, "mark"]], "response_mode": "offline", "answer": "[{\"word\":\"mark\",\"start\":4.08,\"end\":4.4}]"}
{"utterance_id": "3081-166546-0086", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0086.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.58, 3.31, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":2.58,\"end\":3.31}]"}
{"utterance_id": "1993-147964-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0010.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After listening to the whole recording, output timed count words for a person's name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[6.79, 7.15, "one"], [13.12, 13.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.79,\"end\":7.15},{\"word\":\"two\",\"start\":13.12,\"end\":13.48}]"}
{"utterance_id": "7850-281318-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0012.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.37, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":1.14,\"end\":1.37}]"}
{"utterance_id": "8297-275155-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0004.flac", "task_type": "lexical_detection", "query": "suspicions", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"suspicions\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.81, "suspicions"]], "response_mode": "offline", "answer": "[{\"word\":\"suspicions\",\"start\":1.12,\"end\":1.81}]"}
{"utterance_id": "2277-149874-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0014.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.12, "one"], [6.94, 7.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.81,\"end\":1.12},{\"word\":\"two\",\"start\":6.94,\"end\":7.3}]"}
{"utterance_id": "3752-4944-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0057.flac", "task_type": "lexical_detection", "query": "boy", "prompt": "When the utterance is complete, report all instances of the word \"boy\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 1.89, "boy"]], "response_mode": "offline", "answer": "[{\"word\":\"boy\",\"start\":1.66,\"end\":1.89}]"}
{"utterance_id": "8842-302201-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0006.flac", "task_type": "lexical_detection", "query": "deep", "prompt": "After listening to the whole recording, output every instance of the word \"deep\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.42, 8.64, "deep"]], "response_mode": "offline", "answer": "[{\"word\":\"deep\",\"start\":8.42,\"end\":8.64}]"}
{"utterance_id": "6241-66616-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 1.88, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.72,\"end\":1.88}]"}
{"utterance_id": "5694-64038-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0023.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.29, "yankees"]], "response_mode": "offline", "answer": "[{\"word\":\"yankees\",\"start\":1.85,\"end\":2.29}]"}
{"utterance_id": "1988-24833-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0007.flac", "task_type": "lexical_detection", "query": "does", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"does\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.67, "does"]], "response_mode": "offline", "answer": "[{\"word\":\"does\",\"start\":0.3,\"end\":0.67}]"}
{"utterance_id": "5338-24615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.75, 10.91, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.75,\"end\":10.91}]"}
{"utterance_id": "2078-142845-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0031.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Process the full audio offline. For every occurrence of a food name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.62, 11.99, "one"], [15.56, 16.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":11.62,\"end\":11.99},{\"word\":\"two\",\"start\":15.56,\"end\":16.2}]"}
{"utterance_id": "2428-83699-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0018.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.65, 5.5, "hangar dene"]], "response_mode": "offline", "answer": "[{\"word\":\"hangar dene\",\"start\":4.65,\"end\":5.5}]"}
{"utterance_id": "84-121550-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0011.flac", "task_type": "lexical_detection", "query": "needs", "prompt": "Once the audio has finished, identify every instance of the word \"needs\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.54, 1.86, "needs"]], "response_mode": "offline", "answer": "[{\"word\":\"needs\",\"start\":1.54,\"end\":1.86}]"}
{"utterance_id": "251-118436-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 1.07, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":0.48,\"end\":1.07}]"}
{"utterance_id": "6313-66129-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0013.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.93, 1.29, "settler"]], "response_mode": "offline", "answer": "[{\"word\":\"settler\",\"start\":0.93,\"end\":1.29}]"}
{"utterance_id": "1462-170142-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0033.flac", "task_type": "lexical_detection", "query": "leaned", "prompt": "Process the full audio offline. Find every instance of the word \"leaned\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.11, "leaned"]], "response_mode": "offline", "answer": "[{\"word\":\"leaned\",\"start\":0.85,\"end\":1.11}]"}
{"utterance_id": "3853-163249-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0033.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[1.53, 2.12, "trainer"]], "response_mode": "offline", "answer": "[{\"word\":\"trainer\",\"start\":1.53,\"end\":2.12}]"}
{"utterance_id": "8297-275154-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0002.flac", "task_type": "lexical_detection", "query": "captain's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"captain's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.31, "captain's"]], "response_mode": "offline", "answer": "[{\"word\":\"captain's\",\"start\":1.74,\"end\":2.31}]"}
{"utterance_id": "6241-61946-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0000.flac", "task_type": "lexical_detection", "query": "exhilarating", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"exhilarating\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 2.03, "exhilarating"]], "response_mode": "offline", "answer": "[{\"word\":\"exhilarating\",\"start\":1.45,\"end\":2.03}]"}
{"utterance_id": "2428-83699-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0003.flac", "task_type": "lexical_detection", "query": "terms", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"terms\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.37, "terms"]], "response_mode": "offline", "answer": "[{\"word\":\"terms\",\"start\":0.86,\"end\":1.37}]"}
{"utterance_id": "3170-137482-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0000.flac", "task_type": "lexical_counting", "query": "rightly", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"rightly\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[23.74, 24.09, "one"], [26.48, 26.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":23.74,\"end\":24.09},{\"word\":\"two\",\"start\":26.48,\"end\":26.83}]"}
{"utterance_id": "1272-135031-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0017.flac", "task_type": "lexical_detection", "query": "discover", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"discover\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.24, 4.75, "discover"]], "response_mode": "offline", "answer": "[{\"word\":\"discover\",\"start\":4.24,\"end\":4.75}]"}
{"utterance_id": "777-126732-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0064.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.51, 4.74, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":4.51,\"end\":4.74}]"}
{"utterance_id": "1462-170138-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0006.flac", "task_type": "lexical_detection", "query": "interior", "prompt": "Process the full audio offline. Find every instance of the word \"interior\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.8, 6.35, "interior"]], "response_mode": "offline", "answer": "[{\"word\":\"interior\",\"start\":5.8,\"end\":6.35}]"}
{"utterance_id": "2086-149220-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0015.flac", "task_type": "lexical_detection", "query": "passes", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"passes\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.49, 6.01, "passes"]], "response_mode": "offline", "answer": "[{\"word\":\"passes\",\"start\":5.49,\"end\":6.01}]"}
{"utterance_id": "2086-149220-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.41, 1.82, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.41,\"end\":1.82}]"}
{"utterance_id": "2412-153948-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0011.flac", "task_type": "lexical_detection", "query": "lonely", "prompt": "Process the full audio offline. Find every instance of the word \"lonely\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.97, "lonely"]], "response_mode": "offline", "answer": "[{\"word\":\"lonely\",\"start\":0.48,\"end\":0.97}]"}
{"utterance_id": "5338-284437-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0023.flac", "task_type": "lexical_detection", "query": "best", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"best\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.1, 2.69, "best"]], "response_mode": "offline", "answer": "[{\"word\":\"best\",\"start\":2.1,\"end\":2.69}]"}
{"utterance_id": "7850-111771-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.01, 9.17, "one hundred thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"one hundred thousand\",\"start\":8.01,\"end\":9.17}]"}
{"utterance_id": "5536-43363-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0006.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.39, 4.68, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":4.39,\"end\":4.68}]"}
{"utterance_id": "2412-153947-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0006.flac", "task_type": "lexical_detection", "query": "complain", "prompt": "Process the full audio offline. Find every instance of the word \"complain\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.91, 3.48, "complain"]], "response_mode": "offline", "answer": "[{\"word\":\"complain\",\"start\":2.91,\"end\":3.48}]"}
{"utterance_id": "2035-147960-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0010.flac", "task_type": "lexical_detection", "query": "explained", "prompt": "Once the audio has finished, identify every instance of the word \"explained\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.54, 1.02, "explained"]], "response_mode": "offline", "answer": "[{\"word\":\"explained\",\"start\":0.54,\"end\":1.02}]"}
{"utterance_id": "6313-76958-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.85, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":0.5,\"end\":0.85}]"}
{"utterance_id": "5694-64025-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0003.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.31, 3.43, "chalmers brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"chalmers brigade\",\"start\":2.31,\"end\":3.43}]"}
{"utterance_id": "5895-34622-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0007.flac", "task_type": "lexical_detection", "query": "regarded", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"regarded\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 2.75, "regarded"]], "response_mode": "offline", "answer": "[{\"word\":\"regarded\",\"start\":2.32,\"end\":2.75}]"}
{"utterance_id": "3576-138058-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.57, 4.96, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":4.57,\"end\":4.96}]"}
{"utterance_id": "1673-143396-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0016.flac", "task_type": "lexical_detection", "query": "julian", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"julian\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.42, 3.97, "julian"]], "response_mode": "offline", "answer": "[{\"word\":\"julian\",\"start\":3.42,\"end\":3.97}]"}
{"utterance_id": "2086-149220-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0036.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.9, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":1.51,\"end\":1.9}]"}
{"utterance_id": "1272-128104-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[14.59, 15.53, "michael angelo"]], "response_mode": "offline", "answer": "[{\"word\":\"michael angelo\",\"start\":14.59,\"end\":15.53}]"}
{"utterance_id": "2803-161169-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0009.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[13.17, 13.66, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"black\",\"start\":13.17,\"end\":13.66}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.15, 2.45, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":2.15,\"end\":2.45}]"}
{"utterance_id": "3000-15664-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0011.flac", "task_type": "lexical_detection", "query": "shelter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"shelter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.04, 6.56, "shelter"]], "response_mode": "offline", "answer": "[{\"word\":\"shelter\",\"start\":6.04,\"end\":6.56}]"}
{"utterance_id": "1272-128104-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0001.flac", "task_type": "lexical_detection", "query": "interesting", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"interesting\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 3.28, "interesting"]], "response_mode": "offline", "answer": "[{\"word\":\"interesting\",\"start\":2.61,\"end\":3.28}]"}
{"utterance_id": "3536-23268-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.79, 1.32, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss milner\",\"start\":0.79,\"end\":1.32}]"}
{"utterance_id": "5694-64029-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0023.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen carefully to the entire clip, then report every instance of an organization name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.98, "brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"brigade\",\"start\":3.31,\"end\":3.98}]"}
{"utterance_id": "8842-302201-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0000.flac", "task_type": "lexical_detection", "query": "verse", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"verse\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.35, 3.66, "verse"]], "response_mode": "offline", "answer": "[{\"word\":\"verse\",\"start\":3.35,\"end\":3.66}]"}
{"utterance_id": "3000-15664-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0011.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.84, 8.46, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":7.84,\"end\":8.46}]"}
{"utterance_id": "8842-304647-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0011.flac", "task_type": "lexical_detection", "query": "ears", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"ears\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.95, 4.3, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":3.95,\"end\":4.3}]"}
{"utterance_id": "2078-142845-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0030.flac", "task_type": "lexical_detection", "query": "household", "prompt": "Process the full audio offline. Find every instance of the word \"household\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.54, "household"]], "response_mode": "offline", "answer": "[{\"word\":\"household\",\"start\":0.98,\"end\":1.54}]"}
{"utterance_id": "3536-8226-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.44, 8.76, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":8.44,\"end\":8.76}]"}
{"utterance_id": "5338-24615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0002.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[22.93, 23.21, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":22.93,\"end\":23.21}]"}
{"utterance_id": "1988-147956-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0002.flac", "task_type": "lexical_detection", "query": "hole", "prompt": "After listening to the whole recording, output every instance of the word \"hole\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.11, "hole"]], "response_mode": "offline", "answer": "[{\"word\":\"hole\",\"start\":1.77,\"end\":2.11}]"}
{"utterance_id": "6241-66616-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0004.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.28, 3.82, "hudson bay"]], "response_mode": "offline", "answer": "[{\"word\":\"hudson bay\",\"start\":3.28,\"end\":3.82}]"}
{"utterance_id": "652-130737-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0001.flac", "task_type": "lexical_counting", "query": "white", "prompt": "After the audio ends, tally every instance of the word \"white\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.81, 2.12, "one"], [4.95, 5.21, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.81,\"end\":2.12},{\"word\":\"two\",\"start\":4.95,\"end\":5.21}]"}
{"utterance_id": "3000-15664-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0031.flac", "task_type": "lexical_detection", "query": "soldiers", "prompt": "After listening to the whole recording, output every instance of the word \"soldiers\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.62, 8.17, "soldiers"]], "response_mode": "offline", "answer": "[{\"word\":\"soldiers\",\"start\":7.62,\"end\":8.17}]"}
{"utterance_id": "251-118436-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0013.flac", "task_type": "lexical_detection", "query": "space", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"space\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.08, 5.74, "space"]], "response_mode": "offline", "answer": "[{\"word\":\"space\",\"start\":5.08,\"end\":5.74}]"}
{"utterance_id": "652-130737-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.19, 5.02, "bordeaux"]], "response_mode": "offline", "answer": "[{\"word\":\"bordeaux\",\"start\":4.19,\"end\":5.02}]"}
{"utterance_id": "2277-149896-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0005.flac", "task_type": "lexical_detection", "query": "contemplated", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"contemplated\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.8, 3.52, "contemplated"]], "response_mode": "offline", "answer": "[{\"word\":\"contemplated\",\"start\":2.8,\"end\":3.52}]"}
{"utterance_id": "422-122949-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0020.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Once the audio has finished, list each occurrence of a number as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.03, 2.26, "one"], [3.44, 3.62, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.03,\"end\":2.26},{\"word\":\"two\",\"start\":3.44,\"end\":3.62}]"}
{"utterance_id": "652-130726-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0017.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.27, "dishes"]], "response_mode": "offline", "answer": "[{\"word\":\"dishes\",\"start\":3.83,\"end\":4.27}]"}
{"utterance_id": "2086-149220-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.49, 0.82, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":0.49,\"end\":0.82}]"}
{"utterance_id": "7976-110523-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.46, 0.89, "hansel"], [5.59, 5.93, "grethel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":0.46,\"end\":0.89},{\"word\":\"grethel\",\"start\":5.59,\"end\":5.93}]"}
{"utterance_id": "2086-149220-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0048.flac", "task_type": "lexical_detection", "query": "maiden", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"maiden\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.48, "maiden"]], "response_mode": "offline", "answer": "[{\"word\":\"maiden\",\"start\":2.16,\"end\":2.48}]"}
{"utterance_id": "3536-23268-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0009.flac", "task_type": "lexical_detection", "query": "leave", "prompt": "Once the audio has finished, identify every instance of the word \"leave\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.69, 1.89, "leave"]], "response_mode": "offline", "answer": "[{\"word\":\"leave\",\"start\":1.69,\"end\":1.89}]"}
{"utterance_id": "3752-4944-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0046.flac", "task_type": "lexical_detection", "query": "orders", "prompt": "After listening to the whole recording, output every instance of the word \"orders\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.34, "orders"]], "response_mode": "offline", "answer": "[{\"word\":\"orders\",\"start\":0.99,\"end\":1.34}]"}
{"utterance_id": "251-136532-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.09, 7.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":7.09,\"end\":7.35}]"}
{"utterance_id": "2078-142845-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0039.flac", "task_type": "lexical_counting", "query": "buns", "prompt": "Process the full audio offline. For every occurrence of the word \"buns\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.12, "one"], [11.12, 11.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.6,\"end\":1.12},{\"word\":\"two\",\"start\":11.12,\"end\":11.69}]"}
{"utterance_id": "2902-9008-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0001.flac", "task_type": "lexical_detection", "query": "sofa", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sofa\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.49, 5.83, "sofa"]], "response_mode": "offline", "answer": "[{\"word\":\"sofa\",\"start\":5.49,\"end\":5.83}]"}
{"utterance_id": "1993-147149-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.59, 0.88, "jem"]], "response_mode": "offline", "answer": "[{\"word\":\"jem\",\"start\":0.59,\"end\":0.88}]"}
{"utterance_id": "3000-15664-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0031.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.62, 8.17, "soldiers"]], "response_mode": "offline", "answer": "[{\"word\":\"soldiers\",\"start\":7.62,\"end\":8.17}]"}
{"utterance_id": "1919-142785-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0011.flac", "task_type": "lexical_counting", "query": "springs", "prompt": "Listen to the entire clip, then report the count sequence for the word \"springs\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.33, "one"], [5.33, 5.98, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.87,\"end\":1.33},{\"word\":\"two\",\"start\":5.33,\"end\":5.98}]"}
{"utterance_id": "652-129742-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0001.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.88, "salad"], [4.67, 5.1, "lettuce"]], "response_mode": "offline", "answer": "[{\"word\":\"salad\",\"start\":1.34,\"end\":1.88},{\"word\":\"lettuce\",\"start\":4.67,\"end\":5.1}]"}
{"utterance_id": "5694-64025-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0019.flac", "task_type": "lexical_detection", "query": "pleasant", "prompt": "Once the audio has finished, identify every instance of the word \"pleasant\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.77, 2.27, "pleasant"]], "response_mode": "offline", "answer": "[{\"word\":\"pleasant\",\"start\":1.77,\"end\":2.27}]"}
{"utterance_id": "1462-170145-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.19, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander\",\"start\":0.51,\"end\":1.19}]"}
{"utterance_id": "7850-281318-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.63, 1.19, "magpie"], [6.74, 7.36, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"magpie\",\"start\":0.63,\"end\":1.19},{\"word\":\"birds\",\"start\":6.74,\"end\":7.36}]"}
{"utterance_id": "6241-61943-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0002.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the full audio. After it finishes, find every instance of a weather term. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.57, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":2.14,\"end\":2.57}]"}
{"utterance_id": "777-126732-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0036.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.52, "michaelis"]], "response_mode": "offline", "answer": "[{\"word\":\"michaelis\",\"start\":3.78,\"end\":4.52}]"}
{"utterance_id": "3536-23268-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0023.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.81, 4.52, "one"], [9.98, 10.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.81,\"end\":4.52},{\"word\":\"two\",\"start\":9.98,\"end\":10.57}]"}
{"utterance_id": "3536-23268-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0029.flac", "task_type": "lexical_detection", "query": "chair", "prompt": "After the audio ends, extract all occurrences of the word \"chair\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.05, "chair"]], "response_mode": "offline", "answer": "[{\"word\":\"chair\",\"start\":1.64,\"end\":2.05}]"}
{"utterance_id": "1919-142785-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0030.flac", "task_type": "lexical_detection", "query": "shaking", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"shaking\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.62, 8.11, "shaking"]], "response_mode": "offline", "answer": "[{\"word\":\"shaking\",\"start\":7.62,\"end\":8.11}]"}
{"utterance_id": "1272-141231-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0032.flac", "task_type": "lexical_detection", "query": "aside", "prompt": "When the utterance is complete, report all instances of the word \"aside\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 2.99, "aside"]], "response_mode": "offline", "answer": "[{\"word\":\"aside\",\"start\":2.34,\"end\":2.99}]"}
{"utterance_id": "2078-142845-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.64, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":1.97,\"end\":2.64}]"}
{"utterance_id": "3752-4944-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0062.flac", "task_type": "lexical_detection", "query": "teaspoonful", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"teaspoonful\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.88, 1.45, "teaspoonful"]], "response_mode": "offline", "answer": "[{\"word\":\"teaspoonful\",\"start\":0.88,\"end\":1.45}]"}
{"utterance_id": "3536-23268-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 1.1, "missus horton"]], "response_mode": "offline", "answer": "[{\"word\":\"missus horton\",\"start\":0.4,\"end\":1.1}]"}
{"utterance_id": "1673-143396-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[16.66, 17.31, "christ"]], "response_mode": "offline", "answer": "[{\"word\":\"christ\",\"start\":16.66,\"end\":17.31}]"}
{"utterance_id": "422-122949-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0032.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.6, 4.08, "teeth"]], "response_mode": "offline", "answer": "[{\"word\":\"teeth\",\"start\":3.6,\"end\":4.08}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "lexical_detection", "query": "cedars", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cedars\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[21.95, 22.42, "cedars"]], "response_mode": "offline", "answer": "[{\"word\":\"cedars\",\"start\":21.95,\"end\":22.42}]"}
{"utterance_id": "3170-137482-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0048.flac", "task_type": "lexical_detection", "query": "embraced", "prompt": "After listening to the whole recording, output every instance of the word \"embraced\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.69, 5.19, "embraced"]], "response_mode": "offline", "answer": "[{\"word\":\"embraced\",\"start\":4.69,\"end\":5.19}]"}
{"utterance_id": "2428-83699-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0038.flac", "task_type": "lexical_detection", "query": "lives", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lives\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 1.07, "lives"]], "response_mode": "offline", "answer": "[{\"word\":\"lives\",\"start\":0.64,\"end\":1.07}]"}
{"utterance_id": "2078-142845-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0019.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.28, 2.83, "one"], [6.95, 7.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.28,\"end\":2.83},{\"word\":\"two\",\"start\":6.95,\"end\":7.38}]"}
{"utterance_id": "3081-166546-0084", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0084.flac", "task_type": "lexical_detection", "query": "sweetwater", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sweetwater\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.26, "sweetwater"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":0.51,\"end\":1.26}]"}
{"utterance_id": "7976-110124-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0002.flac", "task_type": "lexical_counting", "query": "know", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"know\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.95, 1.32, "one"], [4.92, 5.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.95,\"end\":1.32},{\"word\":\"two\",\"start\":4.92,\"end\":5.08}]"}
{"utterance_id": "6241-66616-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0025.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.52, 3.92, "fourth"]], "response_mode": "offline", "answer": "[{\"word\":\"fourth\",\"start\":3.52,\"end\":3.92}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "lexical_detection", "query": "eat", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"eat\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.9, 4.08, "eat"]], "response_mode": "offline", "answer": "[{\"word\":\"eat\",\"start\":3.9,\"end\":4.08}]"}
{"utterance_id": "1988-24833-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0001.flac", "task_type": "lexical_detection", "query": "taught", "prompt": "Once the audio has finished, identify every instance of the word \"taught\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.96, 2.27, "taught"]], "response_mode": "offline", "answer": "[{\"word\":\"taught\",\"start\":1.96,\"end\":2.27}]"}
{"utterance_id": "2086-149220-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0047.flac", "task_type": "lexical_detection", "query": "awhile", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"awhile\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.92, "awhile"]], "response_mode": "offline", "answer": "[{\"word\":\"awhile\",\"start\":1.43,\"end\":1.92}]"}
{"utterance_id": "652-130737-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0013.flac", "task_type": "lexical_detection", "query": "eight", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"eight\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.53, 2.92, "eight"]], "response_mode": "offline", "answer": "[{\"word\":\"eight\",\"start\":2.53,\"end\":2.92}]"}
{"utterance_id": "5536-43363-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0017.flac", "task_type": "lexical_detection", "query": "fort", "prompt": "After the audio ends, extract all occurrences of the word \"fort\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 3.73, "fort"]], "response_mode": "offline", "answer": "[{\"word\":\"fort\",\"start\":3.5,\"end\":3.73}]"}
{"utterance_id": "2078-142845-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0030.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.27, 2.49, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.27,\"end\":2.49}]"}
{"utterance_id": "6295-64301-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0028.flac", "task_type": "lexical_detection", "query": "crash", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"crash\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.64, 5.15, "crash"]], "response_mode": "offline", "answer": "[{\"word\":\"crash\",\"start\":4.64,\"end\":5.15}]"}
{"utterance_id": "8297-275155-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0032.flac", "task_type": "semantic_counting", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a city name (word, start, end). If none, reply [].", "labels": [[7.03, 7.67, "one"], [8.86, 9.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.03,\"end\":7.67},{\"word\":\"two\",\"start\":8.86,\"end\":9.27}]"}
{"utterance_id": "8842-302203-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0011.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "After listening to the whole recording, output timed count words for an ordinal number (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.43, 0.71, "one"], [3.91, 4.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.43,\"end\":0.71},{\"word\":\"two\",\"start\":3.91,\"end\":4.2}]"}
{"utterance_id": "652-129742-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0015.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.34, "one"], [4.07, 4.27, "two"], [6.35, 6.62, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.34},{\"word\":\"two\",\"start\":4.07,\"end\":4.27},{\"word\":\"three\",\"start\":6.35,\"end\":6.62}]"}
{"utterance_id": "5895-34622-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0005.flac", "task_type": "lexical_counting", "query": "green", "prompt": "Process the full audio offline. For every occurrence of the word \"green\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.82, "one"], [6.23, 6.56, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.55,\"end\":0.82},{\"word\":\"two\",\"start\":6.23,\"end\":6.56}]"}
{"utterance_id": "2086-149220-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0016.flac", "task_type": "lexical_detection", "query": "hepzibah", "prompt": "When the utterance is complete, report all instances of the word \"hepzibah\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.15, "hepzibah"]], "response_mode": "offline", "answer": "[{\"word\":\"hepzibah\",\"start\":0.69,\"end\":1.15}]"}
{"utterance_id": "1919-142785-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0019.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Once the audio has finished, identify every instance of a country name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.11, 2.62, "england"]], "response_mode": "offline", "answer": "[{\"word\":\"england\",\"start\":2.11,\"end\":2.62}]"}
{"utterance_id": "3752-4944-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0052.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.15, 0.49, "north"]], "response_mode": "offline", "answer": "[{\"word\":\"north\",\"start\":0.15,\"end\":0.49}]"}
{"utterance_id": "777-126732-0070", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0070.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[0.96, 1.36, "one"], [3.36, 3.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.96,\"end\":1.36},{\"word\":\"two\",\"start\":3.36,\"end\":3.88}]"}
{"utterance_id": "5895-34622-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0010.flac", "task_type": "lexical_detection", "query": "surprising", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"surprising\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.21, 2.99, "surprising"]], "response_mode": "offline", "answer": "[{\"word\":\"surprising\",\"start\":2.21,\"end\":2.99}]"}
{"utterance_id": "8842-302201-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0012.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.95, "fourth"]], "response_mode": "offline", "answer": "[{\"word\":\"fourth\",\"start\":0.47,\"end\":0.95}]"}
{"utterance_id": "1272-128104-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0000.flac", "task_type": "lexical_detection", "query": "gospel", "prompt": "Once the audio has finished, identify every instance of the word \"gospel\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.84, 5.51, "gospel"]], "response_mode": "offline", "answer": "[{\"word\":\"gospel\",\"start\":4.84,\"end\":5.51}]"}
{"utterance_id": "5895-34629-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0000.flac", "task_type": "lexical_detection", "query": "established", "prompt": "After listening to the whole recording, output every instance of the word \"established\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 1.33, "established"]], "response_mode": "offline", "answer": "[{\"word\":\"established\",\"start\":0.68,\"end\":1.33}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.55, 0.93, "next"]], "response_mode": "offline", "answer": "[{\"word\":\"next\",\"start\":0.55,\"end\":0.93}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[19.15, 19.6, "church's"]], "response_mode": "offline", "answer": "[{\"word\":\"church's\",\"start\":19.15,\"end\":19.6}]"}
{"utterance_id": "652-130726-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0034.flac", "task_type": "lexical_detection", "query": "ordinary", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"ordinary\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.54, 1.02, "ordinary"]], "response_mode": "offline", "answer": "[{\"word\":\"ordinary\",\"start\":0.54,\"end\":1.02}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "lexical_counting", "query": "silver", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"silver\" with word/start/end. If none, reply [].", "labels": [[0.36, 0.8, "one"], [6.67, 7.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.36,\"end\":0.8},{\"word\":\"two\",\"start\":6.67,\"end\":7.0}]"}
{"utterance_id": "652-129742-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0020.flac", "task_type": "lexical_detection", "query": "shake", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"shake\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.47, 2.79, "shake"]], "response_mode": "offline", "answer": "[{\"word\":\"shake\",\"start\":2.47,\"end\":2.79}]"}
{"utterance_id": "2412-153948-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0006.flac", "task_type": "lexical_counting", "query": "see", "prompt": "After listening to the whole recording, output timed count words for the word \"see\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.78, 1.01, "one"], [5.26, 5.42, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.78,\"end\":1.01},{\"word\":\"two\",\"start\":5.26,\"end\":5.42}]"}
{"utterance_id": "251-137823-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0019.flac", "task_type": "lexical_detection", "query": "repaired", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"repaired\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.55, "repaired"]], "response_mode": "offline", "answer": "[{\"word\":\"repaired\",\"start\":1.89,\"end\":2.55}]"}
{"utterance_id": "84-121123-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.15, 7.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":7.15,\"end\":7.36}]"}
{"utterance_id": "3853-163249-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.91, 3.34, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":2.91,\"end\":3.34}]"}
{"utterance_id": "251-137823-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0007.flac", "task_type": "lexical_detection", "query": "stirred", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"stirred\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.27, "stirred"]], "response_mode": "offline", "answer": "[{\"word\":\"stirred\",\"start\":1.77,\"end\":2.27}]"}
{"utterance_id": "422-122949-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0002.flac", "task_type": "lexical_detection", "query": "truthful", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"truthful\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.55, 1.14, "truthful"]], "response_mode": "offline", "answer": "[{\"word\":\"truthful\",\"start\":0.55,\"end\":1.14}]"}
{"utterance_id": "5338-24640-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0009.flac", "task_type": "lexical_detection", "query": "western", "prompt": "Once the audio has finished, identify every instance of the word \"western\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.08, 6.46, "western"]], "response_mode": "offline", "answer": "[{\"word\":\"western\",\"start\":6.08,\"end\":6.46}]"}
{"utterance_id": "2035-147961-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0005.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After listening to the whole recording, output every instance of an organization name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.39, "armies"]], "response_mode": "offline", "answer": "[{\"word\":\"armies\",\"start\":1.87,\"end\":2.39}]"}
{"utterance_id": "8842-302201-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0006.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the entire clip, then report the count sequence for a body part as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.5, "one"], [5.93, 6.23, "two"], [8.1, 8.42, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.21,\"end\":1.5},{\"word\":\"two\",\"start\":5.93,\"end\":6.23},{\"word\":\"three\",\"start\":8.1,\"end\":8.42}]"}
{"utterance_id": "6295-244435-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0001.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.68, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":2.37,\"end\":2.68}]"}
{"utterance_id": "3000-15664-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0037.flac", "task_type": "lexical_detection", "query": "exception", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"exception\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.93, 7.8, "exception"]], "response_mode": "offline", "answer": "[{\"word\":\"exception\",\"start\":6.93,\"end\":7.8}]"}
{"utterance_id": "6345-93302-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0005.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After the audio ends, extract all occurrences of a date. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.65, "may"], [2.91, 3.81, "january"]], "response_mode": "offline", "answer": "[{\"word\":\"may\",\"start\":1.15,\"end\":1.65},{\"word\":\"january\",\"start\":2.91,\"end\":3.81}]"}
{"utterance_id": "5338-24615-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0011.flac", "task_type": "lexical_detection", "query": "court", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"court\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.08, 4.46, "court"]], "response_mode": "offline", "answer": "[{\"word\":\"court\",\"start\":4.08,\"end\":4.46}]"}
{"utterance_id": "3536-23268-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.9, 7.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":6.9,\"end\":7.18}]"}
{"utterance_id": "1919-142785-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0034.flac", "task_type": "lexical_detection", "query": "simmer", "prompt": "Once the audio has finished, identify every instance of the word \"simmer\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.3, 0.67, "simmer"]], "response_mode": "offline", "answer": "[{\"word\":\"simmer\",\"start\":0.3,\"end\":0.67}]"}
{"utterance_id": "6241-61946-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.95, 5.4, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":4.95,\"end\":5.4}]"}
{"utterance_id": "2412-153948-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[8.41, 8.62, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":8.41,\"end\":8.62}]"}
{"utterance_id": "6313-66125-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0019.flac", "task_type": "lexical_detection", "query": "sounding", "prompt": "Process the full audio offline. Find every instance of the word \"sounding\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 2.72, "sounding"]], "response_mode": "offline", "answer": "[{\"word\":\"sounding\",\"start\":2.46,\"end\":2.72}]"}
{"utterance_id": "84-121550-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.83, 7.35, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":6.83,\"end\":7.35}]"}
{"utterance_id": "6295-64301-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0020.flac", "task_type": "lexical_detection", "query": "shape", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"shape\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.66, "shape"]], "response_mode": "offline", "answer": "[{\"word\":\"shape\",\"start\":2.08,\"end\":2.66}]"}
{"utterance_id": "174-50561-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0013.flac", "task_type": "lexical_detection", "query": "bough", "prompt": "Once the audio has finished, identify every instance of the word \"bough\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.05, 2.49, "bough"]], "response_mode": "offline", "answer": "[{\"word\":\"bough\",\"start\":2.05,\"end\":2.49}]"}
{"utterance_id": "6313-76958-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0027.flac", "task_type": "lexical_detection", "query": "foreman", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"foreman\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.92, 3.32, "foreman"]], "response_mode": "offline", "answer": "[{\"word\":\"foreman\",\"start\":2.92,\"end\":3.32}]"}
{"utterance_id": "6345-93302-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0028.flac", "task_type": "lexical_detection", "query": "silly", "prompt": "When the utterance is complete, report all instances of the word \"silly\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.17, "silly"]], "response_mode": "offline", "answer": "[{\"word\":\"silly\",\"start\":0.89,\"end\":1.17}]"}
{"utterance_id": "6241-66616-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0014.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.04, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.6,\"end\":1.04}]"}
{"utterance_id": "2035-152373-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0012.flac", "task_type": "lexical_detection", "query": "reduce", "prompt": "After listening to the whole recording, output every instance of the word \"reduce\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.48, "reduce"]], "response_mode": "offline", "answer": "[{\"word\":\"reduce\",\"start\":4.04,\"end\":4.48}]"}
{"utterance_id": "3853-163249-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0044.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.76, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.48,\"end\":0.76}]"}
{"utterance_id": "7976-110124-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0012.flac", "task_type": "lexical_detection", "query": "sprinkled", "prompt": "When the utterance is complete, report all instances of the word \"sprinkled\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.61, 6.27, "sprinkled"]], "response_mode": "offline", "answer": "[{\"word\":\"sprinkled\",\"start\":5.61,\"end\":6.27}]"}
{"utterance_id": "251-118436-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.49, "shoulders"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulders\",\"start\":1.85,\"end\":2.49}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "When the utterance is complete, report how a person's name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.54, 8.66, "one"], [11.69, 12.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.54,\"end\":8.66},{\"word\":\"two\",\"start\":11.69,\"end\":12.3}]"}
{"utterance_id": "3853-163249-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 2.96, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.65,\"end\":2.96}]"}
{"utterance_id": "2086-149220-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0003.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "When the utterance is complete, report how a plant name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.36, "one"], [2.44, 4.26, "two"], [10.4, 11.13, "three"], [15.01, 15.7, "four"], [18.32, 18.71, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.71,\"end\":1.36},{\"word\":\"two\",\"start\":2.44,\"end\":4.26},{\"word\":\"three\",\"start\":10.4,\"end\":11.13},{\"word\":\"four\",\"start\":15.01,\"end\":15.7},{\"word\":\"five\",\"start\":18.32,\"end\":18.71}]"}
{"utterance_id": "3853-163249-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0040.flac", "task_type": "lexical_detection", "query": "enjoy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"enjoy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.92, "enjoy"]], "response_mode": "offline", "answer": "[{\"word\":\"enjoy\",\"start\":1.46,\"end\":1.92}]"}
{"utterance_id": "6295-64301-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0010.flac", "task_type": "lexical_detection", "query": "father's", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"father's\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.06, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"father's\",\"start\":5.59,\"end\":6.06}]"}
{"utterance_id": "2035-147960-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0007.flac", "task_type": "lexical_detection", "query": "brave", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"brave\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.1, "brave"]], "response_mode": "offline", "answer": "[{\"word\":\"brave\",\"start\":1.64,\"end\":2.1}]"}
{"utterance_id": "422-122949-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0017.flac", "task_type": "lexical_detection", "query": "problem", "prompt": "Once the audio has finished, identify every instance of the word \"problem\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.95, 1.4, "problem"]], "response_mode": "offline", "answer": "[{\"word\":\"problem\",\"start\":0.95,\"end\":1.4}]"}
{"utterance_id": "777-126732-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.63, 9.14, "chest"]], "response_mode": "offline", "answer": "[{\"word\":\"chest\",\"start\":8.63,\"end\":9.14}]"}
{"utterance_id": "6313-66129-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0017.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[3.47, 3.79, "one"], [5.97, 6.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.47,\"end\":3.79},{\"word\":\"two\",\"start\":5.97,\"end\":6.37}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen carefully to the entire clip, then report every instance of a state or province name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[27.18, 27.97, "anatolia"]], "response_mode": "offline", "answer": "[{\"word\":\"anatolia\",\"start\":27.18,\"end\":27.97}]"}
{"utterance_id": "7976-105575-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0018.flac", "task_type": "lexical_detection", "query": "smiling", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"smiling\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.49, "smiling"]], "response_mode": "offline", "answer": "[{\"word\":\"smiling\",\"start\":1.9,\"end\":2.49}]"}
{"utterance_id": "652-130737-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 1.08, "hochheimer"], [3.14, 3.54, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"hochheimer\",\"start\":0.31,\"end\":1.08},{\"word\":\"wine\",\"start\":3.14,\"end\":3.54}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.51, 9.04, "one"], [20.78, 21.25, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.51,\"end\":9.04},{\"word\":\"two\",\"start\":20.78,\"end\":21.25}]"}
{"utterance_id": "1462-170142-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0024.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.83, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.25,\"end\":1.83}]"}
{"utterance_id": "5694-64025-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0016.flac", "task_type": "lexical_detection", "query": "frightened", "prompt": "Once the audio has finished, identify every instance of the word \"frightened\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.25, 7.75, "frightened"]], "response_mode": "offline", "answer": "[{\"word\":\"frightened\",\"start\":7.25,\"end\":7.75}]"}
{"utterance_id": "6313-76958-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0030.flac", "task_type": "lexical_detection", "query": "asked", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"asked\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.55, "asked"]], "response_mode": "offline", "answer": "[{\"word\":\"asked\",\"start\":1.11,\"end\":1.55}]"}
{"utterance_id": "251-136532-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0023.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.26, 1.87, "society"]], "response_mode": "offline", "answer": "[{\"word\":\"society\",\"start\":1.26,\"end\":1.87}]"}
{"utterance_id": "5694-64025-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0010.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.64, 5.2, "horses"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":4.64,\"end\":5.2}]"}
{"utterance_id": "8297-275154-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0004.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.04, 3.73, "sydenham"]], "response_mode": "offline", "answer": "[{\"word\":\"sydenham\",\"start\":3.04,\"end\":3.73}]"}
{"utterance_id": "174-50561-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.06, 9.48, "thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"thousand\",\"start\":9.06,\"end\":9.48}]"}
{"utterance_id": "6241-61946-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0009.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After the audio ends, extract all occurrences of a disease name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.01, "cramp"]], "response_mode": "offline", "answer": "[{\"word\":\"cramp\",\"start\":1.61,\"end\":2.01}]"}
{"utterance_id": "3170-137482-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0042.flac", "task_type": "lexical_detection", "query": "real", "prompt": "When the utterance is complete, report all instances of the word \"real\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.38, "real"]], "response_mode": "offline", "answer": "[{\"word\":\"real\",\"start\":4.04,\"end\":4.38}]"}
{"utterance_id": "2277-149897-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0032.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.43, "watchman"]], "response_mode": "offline", "answer": "[{\"word\":\"watchman\",\"start\":4.99,\"end\":5.43}]"}
{"utterance_id": "174-50561-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0009.flac", "task_type": "lexical_detection", "query": "wandering", "prompt": "Process the full audio offline. Find every instance of the word \"wandering\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 1.09, "wandering"]], "response_mode": "offline", "answer": "[{\"word\":\"wandering\",\"start\":0.65,\"end\":1.09}]"}
{"utterance_id": "1673-143397-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0017.flac", "task_type": "lexical_detection", "query": "awful", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"awful\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.87, 1.22, "awful"]], "response_mode": "offline", "answer": "[{\"word\":\"awful\",\"start\":0.87,\"end\":1.22}]"}
{"utterance_id": "84-121550-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0023.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.1, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":0.8,\"end\":1.1}]"}
{"utterance_id": "7850-281318-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0011.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "After the audio ends, tally every instance of an animal name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.63, 1.19, "one"], [6.74, 7.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.63,\"end\":1.19},{\"word\":\"two\",\"start\":6.74,\"end\":7.36}]"}
{"utterance_id": "3081-166546-0088", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0088.flac", "task_type": "lexical_detection", "query": "blow", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"blow\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.11, 1.43, "blow"]], "response_mode": "offline", "answer": "[{\"word\":\"blow\",\"start\":1.11,\"end\":1.43}]"}
{"utterance_id": "2428-83699-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0018.flac", "task_type": "lexical_detection", "query": "mile", "prompt": "After the audio ends, extract all occurrences of the word \"mile\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.56, 4.06, "mile"]], "response_mode": "offline", "answer": "[{\"word\":\"mile\",\"start\":3.56,\"end\":4.06}]"}
{"utterance_id": "8842-304647-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0009.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.53, 0.84, "one"], [3.58, 3.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.53,\"end\":0.84},{\"word\":\"two\",\"start\":3.58,\"end\":3.91}]"}
{"utterance_id": "3081-166546-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0017.flac", "task_type": "lexical_detection", "query": "mercurial", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"mercurial\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.43, 2.36, "mercurial"]], "response_mode": "offline", "answer": "[{\"word\":\"mercurial\",\"start\":1.43,\"end\":2.36}]"}
{"utterance_id": "3536-23268-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.53, 1.14, "dorriforth"]], "response_mode": "offline", "answer": "[{\"word\":\"dorriforth\",\"start\":0.53,\"end\":1.14}]"}
{"utterance_id": "251-118436-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.32, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":2.72,\"end\":3.32}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Process the full audio offline. Find every instance of a money or currency term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.5, 13.92, "rent"]], "response_mode": "offline", "answer": "[{\"word\":\"rent\",\"start\":13.5,\"end\":13.92}]"}
{"utterance_id": "777-126732-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0055.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.29, 0.63, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":0.29,\"end\":0.63}]"}
{"utterance_id": "2902-9008-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.22, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":0.86,\"end\":1.22}]"}
{"utterance_id": "1462-170142-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0002.flac", "task_type": "lexical_detection", "query": "card", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"card\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.86, 3.21, "card"]], "response_mode": "offline", "answer": "[{\"word\":\"card\",\"start\":2.86,\"end\":3.21}]"}
{"utterance_id": "1988-24833-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 1.91, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":1.61,\"end\":1.91}]"}
{"utterance_id": "6295-64301-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.04, 3.71, "joseph"]], "response_mode": "offline", "answer": "[{\"word\":\"joseph\",\"start\":3.04,\"end\":3.71}]"}
{"utterance_id": "5694-64025-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.17, 1.47, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.17,\"end\":1.47}]"}
{"utterance_id": "3536-23268-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.27, 2.64, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.27,\"end\":2.64}]"}
{"utterance_id": "1272-135031-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0012.flac", "task_type": "lexical_detection", "query": "brother", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"brother\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.43, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.11,\"end\":1.43}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[15.41, 16.02, "sunset"]], "response_mode": "offline", "answer": "[{\"word\":\"sunset\",\"start\":15.41,\"end\":16.02}]"}
{"utterance_id": "2803-154328-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0022.flac", "task_type": "lexical_detection", "query": "rolled", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"rolled\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.79, "rolled"]], "response_mode": "offline", "answer": "[{\"word\":\"rolled\",\"start\":1.51,\"end\":1.79}]"}
{"utterance_id": "2078-142845-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0042.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.28, "victoria"]], "response_mode": "offline", "answer": "[{\"word\":\"victoria\",\"start\":0.52,\"end\":1.28}]"}
{"utterance_id": "1272-141231-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0018.flac", "task_type": "lexical_detection", "query": "inexhaustible", "prompt": "When the utterance is complete, report all instances of the word \"inexhaustible\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 4.34, "inexhaustible"]], "response_mode": "offline", "answer": "[{\"word\":\"inexhaustible\",\"start\":3.5,\"end\":4.34}]"}
{"utterance_id": "1673-143396-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0013.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.73, 1.25, "one"], [3.32, 3.87, "two"], [11.97, 12.34, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.73,\"end\":1.25},{\"word\":\"two\",\"start\":3.32,\"end\":3.87},{\"word\":\"three\",\"start\":11.97,\"end\":12.34}]"}
{"utterance_id": "3081-166546-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0026.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After the audio ends, extract all occurrences of a time of day. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.22, 8.59, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":8.22,\"end\":8.59}]"}
{"utterance_id": "3536-23268-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0021.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 1.98, "tea"], [13.16, 13.55, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"tea\",\"start\":1.64,\"end\":1.98},{\"word\":\"dinner\",\"start\":13.16,\"end\":13.55}]"}
{"utterance_id": "5895-34629-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0028.flac", "task_type": "lexical_detection", "query": "observe", "prompt": "After listening to the whole recording, output every instance of the word \"observe\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.43, 4.8, "observe"]], "response_mode": "offline", "answer": "[{\"word\":\"observe\",\"start\":4.43,\"end\":4.8}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.52, "one"], [6.67, 7.05, "two"], [8.68, 8.93, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.15,\"end\":3.52},{\"word\":\"two\",\"start\":6.67,\"end\":7.05},{\"word\":\"three\",\"start\":8.68,\"end\":8.93}]"}
{"utterance_id": "2803-161169-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0008.flac", "task_type": "lexical_detection", "query": "split", "prompt": "When the utterance is complete, report all instances of the word \"split\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 5.56, "split"]], "response_mode": "offline", "answer": "[{\"word\":\"split\",\"start\":5.14,\"end\":5.56}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Process the full audio offline. Find every instance of an ordinal number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 5.44, "first"], [7.74, 8.08, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.14,\"end\":5.44},{\"word\":\"second\",\"start\":7.74,\"end\":8.08}]"}
{"utterance_id": "3170-137482-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.5, 0.71, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.5,\"end\":0.71}]"}
{"utterance_id": "1673-143397-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0002.flac", "task_type": "lexical_detection", "query": "magistrate", "prompt": "After listening to the whole recording, output every instance of the word \"magistrate\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.96, "magistrate"]], "response_mode": "offline", "answer": "[{\"word\":\"magistrate\",\"start\":3.31,\"end\":3.96}]"}
{"utterance_id": "1462-170145-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0015.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Do not answer until the audio ends. Then list each instance of a country name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.82, 1.28, "canada"]], "response_mode": "offline", "answer": "[{\"word\":\"canada\",\"start\":0.82,\"end\":1.28}]"}
{"utterance_id": "1919-142785-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.72, 2.89, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.72,\"end\":2.89}]"}
{"utterance_id": "8842-302203-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0008.flac", "task_type": "lexical_counting", "query": "part", "prompt": "Once the audio has finished, list each occurrence of the word \"part\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.59, 0.81, "one"], [3.2, 3.41, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.59,\"end\":0.81},{\"word\":\"two\",\"start\":3.2,\"end\":3.41}]"}
{"utterance_id": "5338-284437-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0027.flac", "task_type": "lexical_detection", "query": "misfortune", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"misfortune\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 1.06, "misfortune"]], "response_mode": "offline", "answer": "[{\"word\":\"misfortune\",\"start\":0.48,\"end\":1.06}]"}
{"utterance_id": "3576-138058-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0027.flac", "task_type": "lexical_detection", "query": "abode", "prompt": "Once the audio has finished, identify every instance of the word \"abode\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.47, 8.92, "abode"]], "response_mode": "offline", "answer": "[{\"word\":\"abode\",\"start\":8.47,\"end\":8.92}]"}
{"utterance_id": "1462-170142-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0023.flac", "task_type": "lexical_detection", "query": "worn", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"worn\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.0, "worn"]], "response_mode": "offline", "answer": "[{\"word\":\"worn\",\"start\":1.76,\"end\":2.0}]"}
{"utterance_id": "3576-138058-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.49, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":0.51,\"end\":1.49}]"}
{"utterance_id": "174-84280-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.34, 7.74, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":7.34,\"end\":7.74}]"}
{"utterance_id": "1988-24833-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0002.flac", "task_type": "lexical_counting", "query": "doesn't", "prompt": "After listening to the whole recording, output timed count words for the word \"doesn't\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.49, 0.77, "one"], [3.09, 3.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.49,\"end\":0.77},{\"word\":\"two\",\"start\":3.09,\"end\":3.38}]"}
{"utterance_id": "3081-166546-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0006.flac", "task_type": "lexical_detection", "query": "wide", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"wide\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.25, 1.51, "wide"]], "response_mode": "offline", "answer": "[{\"word\":\"wide\",\"start\":1.25,\"end\":1.51}]"}
{"utterance_id": "2035-147961-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0030.flac", "task_type": "lexical_detection", "query": "knocked", "prompt": "Once the audio has finished, identify every instance of the word \"knocked\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.56, 0.85, "knocked"]], "response_mode": "offline", "answer": "[{\"word\":\"knocked\",\"start\":0.56,\"end\":0.85}]"}
{"utterance_id": "2428-83705-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0018.flac", "task_type": "lexical_detection", "query": "pounds", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pounds\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.91, 5.34, "pounds"]], "response_mode": "offline", "answer": "[{\"word\":\"pounds\",\"start\":4.91,\"end\":5.34}]"}
{"utterance_id": "5895-34615-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0015.flac", "task_type": "lexical_detection", "query": "joyous", "prompt": "Once the audio has finished, identify every instance of the word \"joyous\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.53, 1.04, "joyous"]], "response_mode": "offline", "answer": "[{\"word\":\"joyous\",\"start\":0.53,\"end\":1.04}]"}
{"utterance_id": "1993-147149-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.49, 2.92, "wilsons"]], "response_mode": "offline", "answer": "[{\"word\":\"wilsons\",\"start\":2.49,\"end\":2.92}]"}
{"utterance_id": "3170-137482-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0030.flac", "task_type": "lexical_detection", "query": "occur", "prompt": "Process the full audio offline. Find every instance of the word \"occur\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 2.75, "occur"]], "response_mode": "offline", "answer": "[{\"word\":\"occur\",\"start\":2.34,\"end\":2.75}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 5.51, "twenty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty\",\"start\":5.14,\"end\":5.51}]"}
{"utterance_id": "3576-138058-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 2.16, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":1.5,\"end\":2.16}]"}
{"utterance_id": "7850-281318-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0021.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After listening to the whole recording, output every instance of a plant name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.98, "elm tree"]], "response_mode": "offline", "answer": "[{\"word\":\"elm tree\",\"start\":3.29,\"end\":3.98}]"}
{"utterance_id": "8297-275154-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0010.flac", "task_type": "lexical_detection", "query": "gloom", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"gloom\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.94, 6.34, "gloom"]], "response_mode": "offline", "answer": "[{\"word\":\"gloom\",\"start\":5.94,\"end\":6.34}]"}
{"utterance_id": "2428-83705-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0021.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.36, 2.89, "pounds"]], "response_mode": "offline", "answer": "[{\"word\":\"pounds\",\"start\":2.36,\"end\":2.89}]"}
{"utterance_id": "2035-147960-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0012.flac", "task_type": "lexical_detection", "query": "dragging", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"dragging\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.64, 3.14, "dragging"]], "response_mode": "offline", "answer": "[{\"word\":\"dragging\",\"start\":2.64,\"end\":3.14}]"}
{"utterance_id": "5895-34622-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "When the utterance is complete, report all instances of a color name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.17, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":1.87,\"end\":2.17}]"}
{"utterance_id": "2078-142845-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.23, 1.77, "seventeen twenty four"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen twenty four\",\"start\":0.23,\"end\":1.77}]"}
{"utterance_id": "1988-24833-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0022.flac", "task_type": "lexical_detection", "query": "shouldn't", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"shouldn't\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.45, "shouldn't"]], "response_mode": "offline", "answer": "[{\"word\":\"shouldn't\",\"start\":1.21,\"end\":1.45}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "lexical_detection", "query": "bounty", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bounty\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.63, 3.96, "bounty"]], "response_mode": "offline", "answer": "[{\"word\":\"bounty\",\"start\":3.63,\"end\":3.96}]"}
{"utterance_id": "652-130737-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0005.flac", "task_type": "lexical_detection", "query": "flavor", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"flavor\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.81, 2.33, "flavor"]], "response_mode": "offline", "answer": "[{\"word\":\"flavor\",\"start\":1.81,\"end\":2.33}]"}
{"utterance_id": "3853-163249-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0042.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Once the audio has finished, list each occurrence of a family relation as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.77, 3.12, "one"], [10.15, 10.58, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.77,\"end\":3.12},{\"word\":\"two\",\"start\":10.15,\"end\":10.58}]"}
{"utterance_id": "2412-153954-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0018.flac", "task_type": "lexical_counting", "query": "looked", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"looked\" with word/start/end. If none, reply [].", "labels": [[1.68, 1.89, "one"], [6.81, 7.04, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.68,\"end\":1.89},{\"word\":\"two\",\"start\":6.81,\"end\":7.04}]"}
{"utterance_id": "1462-170138-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0016.flac", "task_type": "lexical_detection", "query": "keen", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"keen\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.3, 1.56, "keen"]], "response_mode": "offline", "answer": "[{\"word\":\"keen\",\"start\":1.3,\"end\":1.56}]"}
{"utterance_id": "5895-34622-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0009.flac", "task_type": "lexical_detection", "query": "executed", "prompt": "After the audio ends, extract all occurrences of the word \"executed\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.45, 8.07, "executed"]], "response_mode": "offline", "answer": "[{\"word\":\"executed\",\"start\":7.45,\"end\":8.07}]"}
{"utterance_id": "2902-9008-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0007.flac", "task_type": "lexical_detection", "query": "authority", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"authority\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.4, "authority"]], "response_mode": "offline", "answer": "[{\"word\":\"authority\",\"start\":1.93,\"end\":2.4}]"}
{"utterance_id": "84-121123-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.72, "morrel"]], "response_mode": "offline", "answer": "[{\"word\":\"morrel\",\"start\":0.38,\"end\":0.72}]"}
{"utterance_id": "5694-64029-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0007.flac", "task_type": "lexical_detection", "query": "shoot", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"shoot\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.35, 4.75, "shoot"]], "response_mode": "offline", "answer": "[{\"word\":\"shoot\",\"start\":4.35,\"end\":4.75}]"}
{"utterance_id": "2277-149896-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0004.flac", "task_type": "lexical_detection", "query": "papers", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"papers\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.65, 1.02, "papers"]], "response_mode": "offline", "answer": "[{\"word\":\"papers\",\"start\":0.65,\"end\":1.02}]"}
{"utterance_id": "3000-15664-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0025.flac", "task_type": "lexical_detection", "query": "traveling", "prompt": "When the utterance is complete, report all instances of the word \"traveling\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.63, 1.16, "traveling"]], "response_mode": "offline", "answer": "[{\"word\":\"traveling\",\"start\":0.63,\"end\":1.16}]"}
{"utterance_id": "5338-24640-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0003.flac", "task_type": "lexical_detection", "query": "waverley's", "prompt": "After the audio ends, extract all occurrences of the word \"waverley's\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.42, 4.01, "waverley's"]], "response_mode": "offline", "answer": "[{\"word\":\"waverley's\",\"start\":3.42,\"end\":4.01}]"}
{"utterance_id": "1673-143397-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[9.08, 9.63, "nestorius"]], "response_mode": "offline", "answer": "[{\"word\":\"nestorius\",\"start\":9.08,\"end\":9.63}]"}
{"utterance_id": "3576-138058-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0032.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the full audio. Count every occurrence of a number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.58, "one"], [4.12, 4.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.38,\"end\":0.58},{\"word\":\"two\",\"start\":4.12,\"end\":4.45}]"}
{"utterance_id": "1919-142785-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0008.flac", "task_type": "lexical_detection", "query": "pint", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pint\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.21, 2.57, "pint"]], "response_mode": "offline", "answer": "[{\"word\":\"pint\",\"start\":2.21,\"end\":2.57}]"}
{"utterance_id": "2803-154320-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.6, 4.06, "duncan god"]], "response_mode": "offline", "answer": "[{\"word\":\"duncan god\",\"start\":2.6,\"end\":4.06}]"}
{"utterance_id": "6319-275224-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0016.flac", "task_type": "lexical_detection", "query": "awaited", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"awaited\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.71, 2.15, "awaited"]], "response_mode": "offline", "answer": "[{\"word\":\"awaited\",\"start\":1.71,\"end\":2.15}]"}
{"utterance_id": "1673-143396-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0020.flac", "task_type": "lexical_detection", "query": "abbot", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"abbot\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.65, 2.0, "abbot"]], "response_mode": "offline", "answer": "[{\"word\":\"abbot\",\"start\":1.65,\"end\":2.0}]"}
{"utterance_id": "652-130737-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0004.flac", "task_type": "lexical_detection", "query": "agreeable", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"agreeable\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 4.06, "agreeable"]], "response_mode": "offline", "answer": "[{\"word\":\"agreeable\",\"start\":3.59,\"end\":4.06}]"}
{"utterance_id": "7850-111771-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0007.flac", "task_type": "lexical_detection", "query": "confederate", "prompt": "When the utterance is complete, report all instances of the word \"confederate\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.99, 6.63, "confederate"]], "response_mode": "offline", "answer": "[{\"word\":\"confederate\",\"start\":5.99,\"end\":6.63}]"}
{"utterance_id": "6319-64726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.05, 10.59, "grandmother"]], "response_mode": "offline", "answer": "[{\"word\":\"grandmother\",\"start\":10.05,\"end\":10.59}]"}
{"utterance_id": "1462-170142-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0029.flac", "task_type": "lexical_detection", "query": "simply", "prompt": "Once the audio has finished, identify every instance of the word \"simply\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.94, 3.48, "simply"]], "response_mode": "offline", "answer": "[{\"word\":\"simply\",\"start\":2.94,\"end\":3.48}]"}
{"utterance_id": "1272-135031-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.61, 7.13, "dragon"]], "response_mode": "offline", "answer": "[{\"word\":\"dragon\",\"start\":6.61,\"end\":7.13}]"}
{"utterance_id": "3081-166546-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0031.flac", "task_type": "lexical_detection", "query": "glancing", "prompt": "Once the audio has finished, identify every instance of the word \"glancing\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.41, 3.92, "glancing"]], "response_mode": "offline", "answer": "[{\"word\":\"glancing\",\"start\":3.41,\"end\":3.92}]"}
{"utterance_id": "3853-163249-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0012.flac", "task_type": "lexical_counting", "query": "how", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"how\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[5.37, 5.52, "one"], [7.43, 7.59, "two"], [10.39, 10.58, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.37,\"end\":5.52},{\"word\":\"two\",\"start\":7.43,\"end\":7.59},{\"word\":\"three\",\"start\":10.39,\"end\":10.58}]"}
{"utterance_id": "5895-34629-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.26, "ursus"], [3.72, 4.27, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":1.84,\"end\":2.26},{\"word\":\"gwynplaine\",\"start\":3.72,\"end\":4.27}]"}
{"utterance_id": "2277-149897-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0029.flac", "task_type": "lexical_detection", "query": "manage", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"manage\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.52, 0.85, "manage"]], "response_mode": "offline", "answer": "[{\"word\":\"manage\",\"start\":0.52,\"end\":0.85}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.54, "chef"], [4.28, 4.59, "chef"]], "response_mode": "offline", "answer": "[{\"word\":\"chef\",\"start\":1.25,\"end\":1.54},{\"word\":\"chef\",\"start\":4.28,\"end\":4.59}]"}
{"utterance_id": "3081-166546-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0007.flac", "task_type": "lexical_detection", "query": "person", "prompt": "After listening to the whole recording, output every instance of the word \"person\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.5, "person"]], "response_mode": "offline", "answer": "[{\"word\":\"person\",\"start\":3.1,\"end\":3.5}]"}
{"utterance_id": "3170-137482-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0045.flac", "task_type": "lexical_detection", "query": "whoever", "prompt": "After listening to the whole recording, output every instance of the word \"whoever\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.04, "whoever"]], "response_mode": "offline", "answer": "[{\"word\":\"whoever\",\"start\":0.53,\"end\":1.04}]"}
{"utterance_id": "2803-154320-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0004.flac", "task_type": "lexical_detection", "query": "indifferent", "prompt": "After listening to the whole recording, output every instance of the word \"indifferent\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.38, 10.98, "indifferent"]], "response_mode": "offline", "answer": "[{\"word\":\"indifferent\",\"start\":10.38,\"end\":10.98}]"}
{"utterance_id": "1272-141231-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0011.flac", "task_type": "lexical_detection", "query": "clearly", "prompt": "Process the full audio offline. Find every instance of the word \"clearly\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.17, 3.71, "clearly"]], "response_mode": "offline", "answer": "[{\"word\":\"clearly\",\"start\":3.17,\"end\":3.71}]"}
{"utterance_id": "2412-153947-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.89, 5.19, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":4.89,\"end\":5.19}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.28, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":1.83,\"end\":2.28}]"}
{"utterance_id": "2035-152373-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.76, 1.02, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.76,\"end\":1.02}]"}
{"utterance_id": "84-121123-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0006.flac", "task_type": "lexical_detection", "query": "chamber", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"chamber\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.08, 2.46, "chamber"]], "response_mode": "offline", "answer": "[{\"word\":\"chamber\",\"start\":2.08,\"end\":2.46}]"}
{"utterance_id": "6319-275224-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[7.58, 7.98, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.58,\"end\":7.98}]"}
{"utterance_id": "5536-43363-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0009.flac", "task_type": "lexical_counting", "query": "days", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"days\" (word, start, end). If none, reply [].", "labels": [[5.58, 6.09, "one"], [14.61, 15.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.58,\"end\":6.09},{\"word\":\"two\",\"start\":14.61,\"end\":15.16}]"}
{"utterance_id": "6313-76958-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0021.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[6.37, 6.81, "ponies"], [10.93, 11.24, "cattle"]], "response_mode": "offline", "answer": "[{\"word\":\"ponies\",\"start\":6.37,\"end\":6.81},{\"word\":\"cattle\",\"start\":10.93,\"end\":11.24}]"}
{"utterance_id": "5694-64025-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0008.flac", "task_type": "lexical_detection", "query": "troops", "prompt": "Once the audio has finished, identify every instance of the word \"troops\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.47, 2.05, "troops"]], "response_mode": "offline", "answer": "[{\"word\":\"troops\",\"start\":1.47,\"end\":2.05}]"}
{"utterance_id": "2086-149220-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.35, 8.81, "hepzibah's"], [9.84, 10.38, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"hepzibah's\",\"start\":8.35,\"end\":8.81},{\"word\":\"phoebe\",\"start\":9.84,\"end\":10.38}]"}
{"utterance_id": "1462-170142-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0034.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.98, 2.4, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":1.98,\"end\":2.4}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "lexical_detection", "query": "mule's", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"mule's\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.39, 3.78, "mule's"]], "response_mode": "offline", "answer": "[{\"word\":\"mule's\",\"start\":3.39,\"end\":3.78}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[21.95, 22.42, "cedars"], [25.99, 26.54, "chestnut"]], "response_mode": "offline", "answer": "[{\"word\":\"cedars\",\"start\":21.95,\"end\":22.42},{\"word\":\"chestnut\",\"start\":25.99,\"end\":26.54}]"}
{"utterance_id": "8297-275156-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.53, 13.19, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":12.53,\"end\":13.19}]"}
{"utterance_id": "777-126732-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.0, "skin"]], "response_mode": "offline", "answer": "[{\"word\":\"skin\",\"start\":0.66,\"end\":1.0}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.39, 0.64, "second"], [2.56, 2.96, "first"], [8.36, 8.78, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.39,\"end\":0.64},{\"word\":\"first\",\"start\":2.56,\"end\":2.96},{\"word\":\"second\",\"start\":8.36,\"end\":8.78}]"}
{"utterance_id": "2412-153954-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0006.flac", "task_type": "lexical_detection", "query": "parted", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"parted\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.0, 4.35, "parted"]], "response_mode": "offline", "answer": "[{\"word\":\"parted\",\"start\":4.0,\"end\":4.35}]"}
{"utterance_id": "777-126732-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0047.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.49, "mister vladimir"]], "response_mode": "offline", "answer": "[{\"word\":\"mister vladimir\",\"start\":2.74,\"end\":3.49}]"}
{"utterance_id": "2078-142845-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0008.flac", "task_type": "lexical_counting", "query": "loaves", "prompt": "Process the full audio offline. For every occurrence of the word \"loaves\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.8, 10.4, "one"], [15.25, 15.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.8,\"end\":10.4},{\"word\":\"two\",\"start\":15.25,\"end\":15.89}]"}
{"utterance_id": "6295-244435-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0031.flac", "task_type": "lexical_detection", "query": "unlike", "prompt": "Process the full audio offline. Find every instance of the word \"unlike\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.48, "unlike"]], "response_mode": "offline", "answer": "[{\"word\":\"unlike\",\"start\":1.01,\"end\":1.48}]"}
{"utterance_id": "2277-149896-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.43, 0.71, "many"]], "response_mode": "offline", "answer": "[{\"word\":\"many\",\"start\":0.43,\"end\":0.71}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "lexical_counting", "query": "form", "prompt": "Process the full audio offline. For every occurrence of the word \"form\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.45, 2.73, "one"], [4.85, 5.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.45,\"end\":2.73},{\"word\":\"two\",\"start\":4.85,\"end\":5.27}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "lexical_counting", "query": "minister", "prompt": "Once the audio has finished, list each occurrence of the word \"minister\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[5.31, 5.87, "one"], [6.96, 7.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.31,\"end\":5.87},{\"word\":\"two\",\"start\":6.96,\"end\":7.38}]"}
{"utterance_id": "6295-244435-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0039.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.12, 5.39, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":5.12,\"end\":5.39}]"}
{"utterance_id": "2086-149220-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.56, 6.16, "chanticleer"]], "response_mode": "offline", "answer": "[{\"word\":\"chanticleer\",\"start\":5.56,\"end\":6.16}]"}
{"utterance_id": "8297-275155-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0032.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.65, 6.02, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":5.65,\"end\":6.02}]"}
{"utterance_id": "3576-138058-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0024.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.52, 4.96, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":4.52,\"end\":4.96}]"}
{"utterance_id": "1993-147149-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0026.flac", "task_type": "lexical_detection", "query": "thank", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"thank\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.31, "thank"]], "response_mode": "offline", "answer": "[{\"word\":\"thank\",\"start\":0.99,\"end\":1.31}]"}
{"utterance_id": "251-118436-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0007.flac", "task_type": "lexical_detection", "query": "heavens", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"heavens\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.49, "heavens"]], "response_mode": "offline", "answer": "[{\"word\":\"heavens\",\"start\":1.02,\"end\":1.49}]"}
{"utterance_id": "5694-64038-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0024.flac", "task_type": "lexical_detection", "query": "attack", "prompt": "After listening to the whole recording, output every instance of the word \"attack\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.96, "attack"]], "response_mode": "offline", "answer": "[{\"word\":\"attack\",\"start\":1.37,\"end\":1.96}]"}
{"utterance_id": "2803-154320-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0005.flac", "task_type": "lexical_detection", "query": "wandered", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"wandered\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.89, "wandered"]], "response_mode": "offline", "answer": "[{\"word\":\"wandered\",\"start\":1.33,\"end\":1.89}]"}
{"utterance_id": "8297-275155-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0031.flac", "task_type": "lexical_detection", "query": "arrived", "prompt": "After the audio ends, extract all occurrences of the word \"arrived\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.08, "arrived"]], "response_mode": "offline", "answer": "[{\"word\":\"arrived\",\"start\":0.53,\"end\":1.08}]"}
{"utterance_id": "7976-110124-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0004.flac", "task_type": "lexical_detection", "query": "unfortunate", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"unfortunate\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 3.03, "unfortunate"]], "response_mode": "offline", "answer": "[{\"word\":\"unfortunate\",\"start\":2.18,\"end\":3.03}]"}
{"utterance_id": "1988-147956-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.19, 4.78, "teeth"]], "response_mode": "offline", "answer": "[{\"word\":\"teeth\",\"start\":4.19,\"end\":4.78}]"}
{"utterance_id": "1988-24833-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.09, 3.38, "pop"]], "response_mode": "offline", "answer": "[{\"word\":\"pop\",\"start\":3.09,\"end\":3.38}]"}
{"utterance_id": "422-122949-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0020.flac", "task_type": "lexical_counting", "query": "nation", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"nation\" (word, start, end). If none, reply [].", "labels": [[2.26, 2.74, "one"], [22.28, 22.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.26,\"end\":2.74},{\"word\":\"two\",\"start\":22.28,\"end\":22.89}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.57, "officer"], [4.01, 4.55, "colonel"]], "response_mode": "offline", "answer": "[{\"word\":\"officer\",\"start\":2.09,\"end\":2.57},{\"word\":\"colonel\",\"start\":4.01,\"end\":4.55}]"}
{"utterance_id": "5694-64029-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.5, "murfreesboro"]], "response_mode": "offline", "answer": "[{\"word\":\"murfreesboro\",\"start\":1.64,\"end\":2.5}]"}
{"utterance_id": "6241-66616-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0016.flac", "task_type": "lexical_detection", "query": "necessity", "prompt": "After listening to the whole recording, output every instance of the word \"necessity\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.11, "necessity"]], "response_mode": "offline", "answer": "[{\"word\":\"necessity\",\"start\":0.5,\"end\":1.11}]"}
{"utterance_id": "6313-66125-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0027.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.99, 9.37, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":8.99,\"end\":9.37}]"}
{"utterance_id": "2086-149220-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0027.flac", "task_type": "lexical_detection", "query": "daguerreotype", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"daguerreotype\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.93, "daguerreotype"]], "response_mode": "offline", "answer": "[{\"word\":\"daguerreotype\",\"start\":1.27,\"end\":1.93}]"}
{"utterance_id": "7976-110523-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0016.flac", "task_type": "lexical_detection", "query": "prayed", "prompt": "After listening to the whole recording, output every instance of the word \"prayed\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.0, "prayed"]], "response_mode": "offline", "answer": "[{\"word\":\"prayed\",\"start\":2.54,\"end\":3.0}]"}
{"utterance_id": "2412-153954-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0013.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.53, "sheep"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":1.2,\"end\":1.53}]"}
{"utterance_id": "1462-170138-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0019.flac", "task_type": "lexical_detection", "query": "she's", "prompt": "Once the audio has finished, identify every instance of the word \"she's\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.1, 1.37, "she's"]], "response_mode": "offline", "answer": "[{\"word\":\"she's\",\"start\":1.1,\"end\":1.37}]"}
{"utterance_id": "1919-142785-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0053.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.14, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.82,\"end\":1.14}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "semantic_detection", "query": "money", "prompt": "You will hear a complete utterance. When it ends, list every instance of a money or currency term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.53, 4.88, "dollars"]], "response_mode": "offline", "answer": "[{\"word\":\"dollars\",\"start\":4.53,\"end\":4.88}]"}
{"utterance_id": "2428-83705-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0024.flac", "task_type": "lexical_detection", "query": "girls", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"girls\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.88, 4.34, "girls"]], "response_mode": "offline", "answer": "[{\"word\":\"girls\",\"start\":3.88,\"end\":4.34}]"}
{"utterance_id": "1462-170145-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0021.flac", "task_type": "lexical_detection", "query": "angrily", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"angrily\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.71, 2.31, "angrily"]], "response_mode": "offline", "answer": "[{\"word\":\"angrily\",\"start\":1.71,\"end\":2.31}]"}
{"utterance_id": "3576-138058-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.12, 1.6, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":1.12,\"end\":1.6}]"}
{"utterance_id": "84-121550-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0007.flac", "task_type": "lexical_detection", "query": "default", "prompt": "After listening to the whole recording, output every instance of the word \"default\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.31, "default"], [3.9, 4.5, "default"]], "response_mode": "offline", "answer": "[{\"word\":\"default\",\"start\":0.77,\"end\":1.31},{\"word\":\"default\",\"start\":3.9,\"end\":4.5}]"}
{"utterance_id": "8297-275154-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.45, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":1.01,\"end\":1.45}]"}
{"utterance_id": "2412-153947-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0005.flac", "task_type": "lexical_counting", "query": "book", "prompt": "Listen to the full audio. Count every occurrence of the word \"book\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.95, 2.19, "one"], [10.05, 10.23, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.95,\"end\":2.19},{\"word\":\"two\",\"start\":10.05,\"end\":10.23}]"}
{"utterance_id": "1993-147964-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0008.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Once the audio has finished, identify every instance of a plant name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.07, 3.41, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":3.07,\"end\":3.41}]"}
{"utterance_id": "2086-149220-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.84, "bees"]], "response_mode": "offline", "answer": "[{\"word\":\"bees\",\"start\":0.5,\"end\":0.84}]"}
{"utterance_id": "2035-147961-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0010.flac", "task_type": "lexical_detection", "query": "seemed", "prompt": "Once the audio has finished, identify every instance of the word \"seemed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.27, 0.67, "seemed"]], "response_mode": "offline", "answer": "[{\"word\":\"seemed\",\"start\":0.27,\"end\":0.67}]"}
{"utterance_id": "84-121123-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0017.flac", "task_type": "lexical_counting", "query": "father", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"father\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.29, 0.71, "one"], [2.46, 2.9, "two"], [8.6, 9.07, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.29,\"end\":0.71},{\"word\":\"two\",\"start\":2.46,\"end\":2.9},{\"word\":\"three\",\"start\":8.6,\"end\":9.07}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.94, "master"]], "response_mode": "offline", "answer": "[{\"word\":\"master\",\"start\":0.51,\"end\":0.94}]"}
{"utterance_id": "3752-4943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.24, 0.83, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":0.24,\"end\":0.83}]"}
{"utterance_id": "3081-166546-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0077.flac", "task_type": "lexical_detection", "query": "moment's", "prompt": "After listening to the whole recording, output every instance of the word \"moment's\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.77, 5.19, "moment's"]], "response_mode": "offline", "answer": "[{\"word\":\"moment's\",\"start\":4.77,\"end\":5.19}]"}
{"utterance_id": "2902-9006-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0013.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "Listen to the full audio. Count every occurrence of an organization name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 1.09, "one"], [13.83, 14.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.49,\"end\":1.09},{\"word\":\"two\",\"start\":13.83,\"end\":14.18}]"}
{"utterance_id": "6313-66129-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0024.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[6.72, 7.0, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":6.72,\"end\":7.0}]"}
{"utterance_id": "3081-166546-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0036.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.57, "gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"gryce\",\"start\":2.05,\"end\":2.57}]"}
{"utterance_id": "6345-64257-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0018.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.91, 2.19, "one"], [4.1, 4.63, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.91,\"end\":2.19},{\"word\":\"two\",\"start\":4.1,\"end\":4.63}]"}
{"utterance_id": "3576-138058-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0032.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.58, "one"], [4.12, 4.45, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.38,\"end\":0.58},{\"word\":\"three\",\"start\":4.12,\"end\":4.45}]"}
{"utterance_id": "3081-166546-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0045.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 1.39, "sweetwater"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":0.54,\"end\":1.39}]"}
{"utterance_id": "251-118436-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0023.flac", "task_type": "lexical_detection", "query": "send", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"send\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.2, 0.69, "send"]], "response_mode": "offline", "answer": "[{\"word\":\"send\",\"start\":0.2,\"end\":0.69}]"}
{"utterance_id": "2078-142845-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0046.flac", "task_type": "semantic_detection", "query": "country", "prompt": "You will hear a complete utterance. When it ends, list every instance of a country name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.68, "england"]], "response_mode": "offline", "answer": "[{\"word\":\"england\",\"start\":2.08,\"end\":2.68}]"}
{"utterance_id": "3081-166546-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0068.flac", "task_type": "lexical_detection", "query": "they'll", "prompt": "When the utterance is complete, report all instances of the word \"they'll\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.62, "they'll"]], "response_mode": "offline", "answer": "[{\"word\":\"they'll\",\"start\":0.45,\"end\":0.62}]"}
{"utterance_id": "2428-83699-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0031.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.79, 2.46, "driver"]], "response_mode": "offline", "answer": "[{\"word\":\"driver\",\"start\":1.79,\"end\":2.46}]"}
{"utterance_id": "1462-170138-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0020.flac", "task_type": "lexical_detection", "query": "possibly", "prompt": "Process the full audio offline. Find every instance of the word \"possibly\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 1.87, "possibly"]], "response_mode": "offline", "answer": "[{\"word\":\"possibly\",\"start\":1.32,\"end\":1.87}]"}
{"utterance_id": "1462-170145-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0003.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the full audio. After it finishes, find every instance of a date. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.84, 8.44, "tuesday"]], "response_mode": "offline", "answer": "[{\"word\":\"tuesday\",\"start\":7.84,\"end\":8.44}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the entire clip, then report the count sequence for a food name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.33, "one"], [7.34, 8.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.86,\"end\":2.33},{\"word\":\"two\",\"start\":7.34,\"end\":8.13}]"}
{"utterance_id": "174-50561-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Once the audio has finished, identify every instance of a plant name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.69, 2.49, "apple bough"], [8.97, 9.32, "apple"]], "response_mode": "offline", "answer": "[{\"word\":\"apple bough\",\"start\":1.69,\"end\":2.49},{\"word\":\"apple\",\"start\":8.97,\"end\":9.32}]"}
{"utterance_id": "1272-128104-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0002.flac", "task_type": "lexical_detection", "query": "beef", "prompt": "After listening to the whole recording, output every instance of the word \"beef\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.54, 5.89, "beef"]], "response_mode": "offline", "answer": "[{\"word\":\"beef\",\"start\":5.54,\"end\":5.89}]"}
{"utterance_id": "2803-154328-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.72, 1.26, "last"], [5.15, 5.76, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.72,\"end\":1.26},{\"word\":\"last\",\"start\":5.15,\"end\":5.76}]"}
{"utterance_id": "1988-147956-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0029.flac", "task_type": "lexical_detection", "query": "embarrassed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"embarrassed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.84, "embarrassed"]], "response_mode": "offline", "answer": "[{\"word\":\"embarrassed\",\"start\":1.15,\"end\":1.84}]"}
{"utterance_id": "5895-34615-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0010.flac", "task_type": "lexical_detection", "query": "augmented", "prompt": "Process the full audio offline. Find every instance of the word \"augmented\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.42, "augmented"]], "response_mode": "offline", "answer": "[{\"word\":\"augmented\",\"start\":2.9,\"end\":3.42}]"}
{"utterance_id": "5338-284437-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.41, "tourmaline"]], "response_mode": "offline", "answer": "[{\"word\":\"tourmaline\",\"start\":1.87,\"end\":2.41}]"}
{"utterance_id": "2078-142845-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0039.flac", "task_type": "lexical_detection", "query": "varied", "prompt": "Process the full audio offline. Find every instance of the word \"varied\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.98, "varied"]], "response_mode": "offline", "answer": "[{\"word\":\"varied\",\"start\":1.46,\"end\":1.98}]"}
{"utterance_id": "3536-23268-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0004.flac", "task_type": "lexical_detection", "query": "milner", "prompt": "Process the full audio offline. Find every instance of the word \"milner\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.32, "milner"]], "response_mode": "offline", "answer": "[{\"word\":\"milner\",\"start\":0.99,\"end\":1.32}]"}
{"utterance_id": "3000-15664-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0023.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.88, 3.33, "sheep"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":2.88,\"end\":3.33}]"}
{"utterance_id": "1919-142785-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0017.flac", "task_type": "semantic_detection", "query": "country", "prompt": "When the utterance is complete, report all instances of a country name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.08, 3.57, "asia"]], "response_mode": "offline", "answer": "[{\"word\":\"asia\",\"start\":3.08,\"end\":3.57}]"}
{"utterance_id": "1919-142785-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0018.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.91, "orange"], [3.12, 3.51, "citron"]], "response_mode": "offline", "answer": "[{\"word\":\"orange\",\"start\":1.39,\"end\":1.91},{\"word\":\"citron\",\"start\":3.12,\"end\":3.51}]"}
{"utterance_id": "6345-93302-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0027.flac", "task_type": "lexical_detection", "query": "paying", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"paying\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 2.73, "paying"]], "response_mode": "offline", "answer": "[{\"word\":\"paying\",\"start\":2.46,\"end\":2.73}]"}
{"utterance_id": "3752-4944-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0041.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Once the audio has finished, identify every instance of a plant name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.25, 1.58, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":1.25,\"end\":1.58}]"}
{"utterance_id": "2078-142845-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0023.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.52, "rolls"]], "response_mode": "offline", "answer": "[{\"word\":\"rolls\",\"start\":0.83,\"end\":1.52}]"}
{"utterance_id": "5694-64038-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0025.flac", "task_type": "lexical_detection", "query": "blazed", "prompt": "When the utterance is complete, report all instances of the word \"blazed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.32, "blazed"]], "response_mode": "offline", "answer": "[{\"word\":\"blazed\",\"start\":1.91,\"end\":2.32}]"}
{"utterance_id": "2277-149897-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0027.flac", "task_type": "lexical_detection", "query": "complications", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"complications\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.45, 3.29, "complications"]], "response_mode": "offline", "answer": "[{\"word\":\"complications\",\"start\":2.45,\"end\":3.29}]"}
{"utterance_id": "3853-163249-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0047.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.67, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.3,\"end\":0.67}]"}
{"utterance_id": "2277-149897-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.41, "manager"]], "response_mode": "offline", "answer": "[{\"word\":\"manager\",\"start\":0.82,\"end\":1.41}]"}
{"utterance_id": "174-50561-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.2, "thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"thousand\",\"start\":3.81,\"end\":4.2}]"}
{"utterance_id": "1919-142785-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0036.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.15, 7.67, "soups"], [11.62, 12.49, "mackerel sauce"]], "response_mode": "offline", "answer": "[{\"word\":\"soups\",\"start\":7.15,\"end\":7.67},{\"word\":\"mackerel sauce\",\"start\":11.62,\"end\":12.49}]"}
{"utterance_id": "5895-34622-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.99, 3.41, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"mouth\",\"start\":2.99,\"end\":3.41}]"}
{"utterance_id": "2078-142845-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0014.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen carefully to the entire clip, then report every instance of a disease name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.49, 7.03, "mildew"]], "response_mode": "offline", "answer": "[{\"word\":\"mildew\",\"start\":6.49,\"end\":7.03}]"}
{"utterance_id": "777-126732-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.6, 7.06, "throat"]], "response_mode": "offline", "answer": "[{\"word\":\"throat\",\"start\":6.6,\"end\":7.06}]"}
{"utterance_id": "174-50561-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.02, 8.35, "singer"]], "response_mode": "offline", "answer": "[{\"word\":\"singer\",\"start\":8.02,\"end\":8.35}]"}
{"utterance_id": "5338-284437-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0031.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.31, 6.67, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":6.31,\"end\":6.67}]"}
{"utterance_id": "3576-138058-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.55, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":2.09,\"end\":2.55}]"}
{"utterance_id": "174-50561-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0008.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.64, 15.01, "rose"]], "response_mode": "offline", "answer": "[{\"word\":\"rose\",\"start\":14.64,\"end\":15.01}]"}
{"utterance_id": "5536-43358-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0000.flac", "task_type": "lexical_detection", "query": "philosopher", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"philosopher\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.63, "philosopher"]], "response_mode": "offline", "answer": "[{\"word\":\"philosopher\",\"start\":0.77,\"end\":1.63}]"}
{"utterance_id": "6295-64301-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0023.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.43, 4.84, "dog"]], "response_mode": "offline", "answer": "[{\"word\":\"dog\",\"start\":4.43,\"end\":4.84}]"}
{"utterance_id": "6295-244435-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0008.flac", "task_type": "lexical_detection", "query": "carolina", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"carolina\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.01, 6.74, "carolina"]], "response_mode": "offline", "answer": "[{\"word\":\"carolina\",\"start\":6.01,\"end\":6.74}]"}
{"utterance_id": "3081-166546-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0035.flac", "task_type": "lexical_detection", "query": "least", "prompt": "Once the audio has finished, identify every instance of the word \"least\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.54, 1.04, "least"]], "response_mode": "offline", "answer": "[{\"word\":\"least\",\"start\":0.54,\"end\":1.04}]"}
{"utterance_id": "251-136532-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0007.flac", "task_type": "lexical_detection", "query": "locked", "prompt": "Once the audio has finished, identify every instance of the word \"locked\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.6, 0.96, "locked"]], "response_mode": "offline", "answer": "[{\"word\":\"locked\",\"start\":0.6,\"end\":0.96}]"}
{"utterance_id": "84-121550-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0016.flac", "task_type": "lexical_detection", "query": "rome", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"rome\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.8, "rome"]], "response_mode": "offline", "answer": "[{\"word\":\"rome\",\"start\":1.29,\"end\":1.8}]"}
{"utterance_id": "3000-15664-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0036.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.36, 6.69, "one"], [11.16, 11.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.36,\"end\":6.69},{\"word\":\"two\",\"start\":11.16,\"end\":11.38}]"}
{"utterance_id": "3536-23268-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0005.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After listening to the whole recording, output every instance of a time of day with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.61, 3.95, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":3.61,\"end\":3.95}]"}
{"utterance_id": "3536-23268-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0023.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[11.94, 12.32, "tea"]], "response_mode": "offline", "answer": "[{\"word\":\"tea\",\"start\":11.94,\"end\":12.32}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.04, 11.56, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":11.04,\"end\":11.56}]"}
{"utterance_id": "174-50561-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0019.flac", "task_type": "lexical_detection", "query": "treatment", "prompt": "After the audio ends, extract all occurrences of the word \"treatment\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.98, "treatment"]], "response_mode": "offline", "answer": "[{\"word\":\"treatment\",\"start\":1.42,\"end\":1.98}]"}
{"utterance_id": "2803-161169-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[4.69, 5.27, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":4.69,\"end\":5.27}]"}
{"utterance_id": "1919-142785-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0019.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.57, 0.87, "lemon"]], "response_mode": "offline", "answer": "[{\"word\":\"lemon\",\"start\":0.57,\"end\":0.87}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.26, 2.45, "two"], [4.71, 5.17, "one"], [7.48, 7.78, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.26,\"end\":2.45},{\"word\":\"one\",\"start\":4.71,\"end\":5.17},{\"word\":\"one\",\"start\":7.48,\"end\":7.78}]"}
{"utterance_id": "3081-166546-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0022.flac", "task_type": "lexical_detection", "query": "grown", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"grown\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.86, 6.21, "grown"]], "response_mode": "offline", "answer": "[{\"word\":\"grown\",\"start\":5.86,\"end\":6.21}]"}
{"utterance_id": "3170-137482-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0036.flac", "task_type": "lexical_detection", "query": "obscure", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"obscure\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.95, "obscure"]], "response_mode": "offline", "answer": "[{\"word\":\"obscure\",\"start\":1.33,\"end\":1.95}]"}
{"utterance_id": "3536-8226-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0002.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen carefully to the entire clip, then report every instance of a country name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.42, 3.87, "england"]], "response_mode": "offline", "answer": "[{\"word\":\"england\",\"start\":3.42,\"end\":3.87}]"}
{"utterance_id": "652-129742-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0008.flac", "task_type": "lexical_detection", "query": "mayonnaise", "prompt": "When the utterance is complete, report all instances of the word \"mayonnaise\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.22, "mayonnaise"]], "response_mode": "offline", "answer": "[{\"word\":\"mayonnaise\",\"start\":0.59,\"end\":1.22}]"}
{"utterance_id": "2803-161169-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0005.flac", "task_type": "lexical_detection", "query": "seeming", "prompt": "Once the audio has finished, identify every instance of the word \"seeming\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[13.28, 13.74, "seeming"]], "response_mode": "offline", "answer": "[{\"word\":\"seeming\",\"start\":13.28,\"end\":13.74}]"}
{"utterance_id": "3536-8226-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0006.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Process the full audio offline. Find every instance of a country name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.8, 8.31, "england"]], "response_mode": "offline", "answer": "[{\"word\":\"england\",\"start\":7.8,\"end\":8.31}]"}
{"utterance_id": "3081-166546-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0077.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.25, "miss clarke"]], "response_mode": "offline", "answer": "[{\"word\":\"miss clarke\",\"start\":0.51,\"end\":1.25}]"}
{"utterance_id": "2412-153954-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0017.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.06, 1.45, "pale"]], "response_mode": "offline", "answer": "[{\"word\":\"pale\",\"start\":1.06,\"end\":1.45}]"}
{"utterance_id": "2277-149897-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.37, 1.54, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.37,\"end\":1.54}]"}
{"utterance_id": "1673-143396-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[10.97, 11.44, "adam"]], "response_mode": "offline", "answer": "[{\"word\":\"adam\",\"start\":10.97,\"end\":11.44}]"}
{"utterance_id": "3853-163249-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.56, "bennet"]], "response_mode": "offline", "answer": "[{\"word\":\"bennet\",\"start\":0.28,\"end\":0.56}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "lexical_detection", "query": "mouth", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"mouth\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.07, 7.47, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"mouth\",\"start\":7.07,\"end\":7.47}]"}
{"utterance_id": "2803-154320-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.52, 2.39, "will halley"]], "response_mode": "offline", "answer": "[{\"word\":\"will halley\",\"start\":1.52,\"end\":2.39}]"}
{"utterance_id": "6319-275224-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0009.flac", "task_type": "lexical_detection", "query": "flower", "prompt": "Once the audio has finished, identify every instance of the word \"flower\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.72, 2.21, "flower"]], "response_mode": "offline", "answer": "[{\"word\":\"flower\",\"start\":1.72,\"end\":2.21}]"}
{"utterance_id": "7976-110523-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0008.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.38, 7.73, "former"]], "response_mode": "offline", "answer": "[{\"word\":\"former\",\"start\":7.38,\"end\":7.73}]"}
{"utterance_id": "1462-170142-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0034.flac", "task_type": "lexical_detection", "query": "break", "prompt": "When the utterance is complete, report all instances of the word \"break\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 1.98, "break"]], "response_mode": "offline", "answer": "[{\"word\":\"break\",\"start\":1.63,\"end\":1.98}]"}
{"utterance_id": "777-126732-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.24, "michaelis"]], "response_mode": "offline", "answer": "[{\"word\":\"michaelis\",\"start\":0.53,\"end\":1.24}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[8.48, 8.76, "one"], [14.42, 14.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.48,\"end\":8.76},{\"word\":\"two\",\"start\":14.42,\"end\":14.8}]"}
{"utterance_id": "5694-64025-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the full audio. After it finishes, find every instance of a disease name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.03, 5.48, "wounds"]], "response_mode": "offline", "answer": "[{\"word\":\"wounds\",\"start\":5.03,\"end\":5.48}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.21, 0.51, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":0.21,\"end\":0.51}]"}
{"utterance_id": "1919-142785-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0022.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.04, 5.33, "eggs"]], "response_mode": "offline", "answer": "[{\"word\":\"eggs\",\"start\":5.04,\"end\":5.33}]"}
{"utterance_id": "6295-244435-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0034.flac", "task_type": "lexical_detection", "query": "never", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"never\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.86, "never"]], "response_mode": "offline", "answer": "[{\"word\":\"never\",\"start\":0.51,\"end\":0.86}]"}
{"utterance_id": "251-136532-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0005.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.46, "terra"]], "response_mode": "offline", "answer": "[{\"word\":\"terra\",\"start\":2.01,\"end\":2.46}]"}
{"utterance_id": "7850-73752-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0019.flac", "task_type": "lexical_detection", "query": "temple", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"temple\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.89, "temple"]], "response_mode": "offline", "answer": "[{\"word\":\"temple\",\"start\":2.35,\"end\":2.89}]"}
{"utterance_id": "8842-302196-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.33, 4.75, "rossetti"]], "response_mode": "offline", "answer": "[{\"word\":\"rossetti\",\"start\":4.33,\"end\":4.75}]"}
{"utterance_id": "3853-163249-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0032.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.07, 1.45, "hepsey"]], "response_mode": "offline", "answer": "[{\"word\":\"hepsey\",\"start\":1.07,\"end\":1.45}]"}
{"utterance_id": "2086-149220-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.89, "hepzibah"], [5.9, 6.39, "pyncheon"]], "response_mode": "offline", "answer": "[{\"word\":\"hepzibah\",\"start\":2.3,\"end\":2.89},{\"word\":\"pyncheon\",\"start\":5.9,\"end\":6.39}]"}
{"utterance_id": "3000-15664-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0036.flac", "task_type": "lexical_counting", "query": "hear", "prompt": "When the utterance is complete, report how the word \"hear\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.71, "one"], [9.96, 10.25, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.44,\"end\":2.71},{\"word\":\"two\",\"start\":9.96,\"end\":10.25}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "lexical_detection", "query": "arranging", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"arranging\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.85, 5.3, "arranging"]], "response_mode": "offline", "answer": "[{\"word\":\"arranging\",\"start\":4.85,\"end\":5.3}]"}
{"utterance_id": "2277-149897-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0037.flac", "task_type": "lexical_detection", "query": "see", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"see\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.08, 1.24, "see"]], "response_mode": "offline", "answer": "[{\"word\":\"see\",\"start\":1.08,\"end\":1.24}]"}
{"utterance_id": "1919-142785-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0012.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.37, "milk"]], "response_mode": "offline", "answer": "[{\"word\":\"milk\",\"start\":2.04,\"end\":2.37}]"}
{"utterance_id": "3853-163249-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0055.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.3, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":1.83,\"end\":2.3}]"}
{"utterance_id": "8297-275155-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0022.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a family relation (word, start, end). If none, reply [].", "labels": [[2.01, 2.41, "one"], [5.43, 5.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.01,\"end\":2.41},{\"word\":\"two\",\"start\":5.43,\"end\":5.91}]"}
{"utterance_id": "2277-149874-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0017.flac", "task_type": "lexical_detection", "query": "destiny", "prompt": "Process the full audio offline. Find every instance of the word \"destiny\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.67, "destiny"]], "response_mode": "offline", "answer": "[{\"word\":\"destiny\",\"start\":1.2,\"end\":1.67}]"}
{"utterance_id": "777-126732-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0032.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a body part: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.84, 1.15, "one"], [4.34, 4.64, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.84,\"end\":1.15},{\"word\":\"two\",\"start\":4.34,\"end\":4.64}]"}
{"utterance_id": "6295-244435-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0009.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.0, 6.72, "confederacy"]], "response_mode": "offline", "answer": "[{\"word\":\"confederacy\",\"start\":6.0,\"end\":6.72}]"}
{"utterance_id": "2086-149220-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0043.flac", "task_type": "lexical_detection", "query": "instrument", "prompt": "Process the full audio offline. Find every instance of the word \"instrument\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.38, "instrument"]], "response_mode": "offline", "answer": "[{\"word\":\"instrument\",\"start\":0.81,\"end\":1.38}]"}
{"utterance_id": "3752-4943-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0004.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.03, 6.49, "bishop"]], "response_mode": "offline", "answer": "[{\"word\":\"bishop\",\"start\":6.03,\"end\":6.49}]"}
{"utterance_id": "652-130737-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0013.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.5, 1.15, "claret"]], "response_mode": "offline", "answer": "[{\"word\":\"claret\",\"start\":0.5,\"end\":1.15}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "When the utterance is complete, report all instances of a disease name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.27, "cancer"], [4.37, 4.81, "cancer"]], "response_mode": "offline", "answer": "[{\"word\":\"cancer\",\"start\":0.86,\"end\":1.27},{\"word\":\"cancer\",\"start\":4.37,\"end\":4.81}]"}
{"utterance_id": "251-118436-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0016.flac", "task_type": "lexical_detection", "query": "flame", "prompt": "Once the audio has finished, identify every instance of the word \"flame\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.55, 7.17, "flame"]], "response_mode": "offline", "answer": "[{\"word\":\"flame\",\"start\":6.55,\"end\":7.17}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.83, 5.16, "gold"]], "response_mode": "offline", "answer": "[{\"word\":\"gold\",\"start\":4.83,\"end\":5.16}]"}
{"utterance_id": "3170-137482-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0012.flac", "task_type": "lexical_detection", "query": "boat", "prompt": "Once the audio has finished, identify every instance of the word \"boat\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.35, 2.76, "boat"]], "response_mode": "offline", "answer": "[{\"word\":\"boat\",\"start\":2.35,\"end\":2.76}]"}
{"utterance_id": "5536-43359-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0015.flac", "task_type": "lexical_detection", "query": "camp", "prompt": "After the audio ends, extract all occurrences of the word \"camp\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.95, 4.44, "camp"]], "response_mode": "offline", "answer": "[{\"word\":\"camp\",\"start\":3.95,\"end\":4.44}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 4.32, "montesinos"], [5.42, 6.3, "dulcinea"], [12.96, 13.79, "sage merlin"], [21.05, 21.85, "dulcinea"]], "response_mode": "offline", "answer": "[{\"word\":\"montesinos\",\"start\":3.29,\"end\":4.32},{\"word\":\"dulcinea\",\"start\":5.42,\"end\":6.3},{\"word\":\"sage merlin\",\"start\":12.96,\"end\":13.79},{\"word\":\"dulcinea\",\"start\":21.05,\"end\":21.85}]"}
{"utterance_id": "8842-302196-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0001.flac", "task_type": "lexical_detection", "query": "original", "prompt": "Once the audio has finished, identify every instance of the word \"original\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.18, 1.54, "original"]], "response_mode": "offline", "answer": "[{\"word\":\"original\",\"start\":1.18,\"end\":1.54}]"}
{"utterance_id": "7850-286674-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.57, "lip"], [7.02, 7.28, "arm"], [8.76, 9.33, "pincers"]], "response_mode": "offline", "answer": "[{\"word\":\"lip\",\"start\":1.4,\"end\":1.57},{\"word\":\"arm\",\"start\":7.02,\"end\":7.28},{\"word\":\"pincers\",\"start\":8.76,\"end\":9.33}]"}
{"utterance_id": "5694-64025-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.51, 6.84, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":6.51,\"end\":6.84}]"}
{"utterance_id": "3853-163249-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.24, "letty"]], "response_mode": "offline", "answer": "[{\"word\":\"letty\",\"start\":0.96,\"end\":1.24}]"}
{"utterance_id": "6319-275224-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.93, 5.53, "gardener's"]], "response_mode": "offline", "answer": "[{\"word\":\"gardener's\",\"start\":4.93,\"end\":5.53}]"}
{"utterance_id": "2803-161169-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0015.flac", "task_type": "lexical_detection", "query": "makers", "prompt": "Once the audio has finished, identify every instance of the word \"makers\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.42, 2.92, "makers"]], "response_mode": "offline", "answer": "[{\"word\":\"makers\",\"start\":2.42,\"end\":2.92}]"}
{"utterance_id": "2428-83705-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 6.35, "snellings"]], "response_mode": "offline", "answer": "[{\"word\":\"snellings\",\"start\":5.55,\"end\":6.35}]"}
{"utterance_id": "5694-64038-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0006.flac", "task_type": "lexical_detection", "query": "says", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"says\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.56, "says"]], "response_mode": "offline", "answer": "[{\"word\":\"says\",\"start\":1.22,\"end\":1.56}]"}
{"utterance_id": "2035-147960-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 1.88, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":1.66,\"end\":1.88}]"}
{"utterance_id": "2035-147961-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0021.flac", "task_type": "lexical_detection", "query": "control", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"control\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 4.02, "control"]], "response_mode": "offline", "answer": "[{\"word\":\"control\",\"start\":3.48,\"end\":4.02}]"}
{"utterance_id": "1272-141231-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.77, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":0.49,\"end\":0.77}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "lexical_detection", "query": "goodness", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"goodness\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.67, 2.13, "goodness"]], "response_mode": "offline", "answer": "[{\"word\":\"goodness\",\"start\":1.67,\"end\":2.13}]"}
{"utterance_id": "6345-93306-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0008.flac", "task_type": "lexical_detection", "query": "drink", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"drink\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.43, 0.69, "drink"]], "response_mode": "offline", "answer": "[{\"word\":\"drink\",\"start\":0.43,\"end\":0.69}]"}
{"utterance_id": "7976-110523-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.4, 0.75, "grethel"], [3.82, 4.32, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"grethel\",\"start\":0.4,\"end\":0.75},{\"word\":\"hansel\",\"start\":3.82,\"end\":4.32}]"}
{"utterance_id": "2902-9008-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0004.flac", "task_type": "lexical_detection", "query": "elect", "prompt": "After the audio ends, extract all occurrences of the word \"elect\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.76, 5.29, "elect"]], "response_mode": "offline", "answer": "[{\"word\":\"elect\",\"start\":4.76,\"end\":5.29}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.09, 3.49, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.09,\"end\":3.49}]"}
{"utterance_id": "652-130726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0004.flac", "task_type": "lexical_detection", "query": "greatly", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"greatly\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.97, 11.4, "greatly"]], "response_mode": "offline", "answer": "[{\"word\":\"greatly\",\"start\":10.97,\"end\":11.4}]"}
{"utterance_id": "3170-137482-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.99, "terro"]], "response_mode": "offline", "answer": "[{\"word\":\"terro\",\"start\":2.43,\"end\":2.99}]"}
{"utterance_id": "1993-147149-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0012.flac", "task_type": "lexical_detection", "query": "cries", "prompt": "After the audio ends, extract all occurrences of the word \"cries\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.99, "cries"]], "response_mode": "offline", "answer": "[{\"word\":\"cries\",\"start\":0.47,\"end\":0.99}]"}
{"utterance_id": "2078-142845-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0005.flac", "task_type": "lexical_counting", "query": "milk", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"milk\" with word/start/end. If none, reply [].", "labels": [[23.7, 23.98, "one"], [28.67, 29.09, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":23.7,\"end\":23.98},{\"word\":\"two\",\"start\":28.67,\"end\":29.09}]"}
{"utterance_id": "2902-9008-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.65, "eyes"], [8.66, 9.11, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.44,\"end\":1.65},{\"word\":\"lips\",\"start\":8.66,\"end\":9.11}]"}
{"utterance_id": "6313-76958-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.81, "cowboy"]], "response_mode": "offline", "answer": "[{\"word\":\"cowboy\",\"start\":0.39,\"end\":0.81}]"}
{"utterance_id": "8297-275155-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.7, 10.08, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":9.7,\"end\":10.08}]"}
{"utterance_id": "3000-15664-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0032.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.87, 6.46, "rushes"]], "response_mode": "offline", "answer": "[{\"word\":\"rushes\",\"start\":5.87,\"end\":6.46}]"}
{"utterance_id": "3752-4943-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0023.flac", "task_type": "lexical_detection", "query": "steadily", "prompt": "After listening to the whole recording, output every instance of the word \"steadily\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 1.98, "steadily"]], "response_mode": "offline", "answer": "[{\"word\":\"steadily\",\"start\":1.45,\"end\":1.98}]"}
{"utterance_id": "7976-110523-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0021.flac", "task_type": "lexical_detection", "query": "neck", "prompt": "After the audio ends, extract all occurrences of the word \"neck\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.37, 4.76, "neck"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":4.37,\"end\":4.76}]"}
{"utterance_id": "5536-43363-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.8, 3.35, "parents"]], "response_mode": "offline", "answer": "[{\"word\":\"parents\",\"start\":2.8,\"end\":3.35}]"}
{"utterance_id": "84-121550-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0031.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.2, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":1.8,\"end\":2.2}]"}
{"utterance_id": "5338-24615-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0007.flac", "task_type": "lexical_detection", "query": "inclosure", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"inclosure\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.07, 9.81, "inclosure"]], "response_mode": "offline", "answer": "[{\"word\":\"inclosure\",\"start\":9.07,\"end\":9.81}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.98, 2.35, "wabi"], [5.14, 5.87, "minnetaki"]], "response_mode": "offline", "answer": "[{\"word\":\"wabi\",\"start\":1.98,\"end\":2.35},{\"word\":\"minnetaki\",\"start\":5.14,\"end\":5.87}]"}
{"utterance_id": "251-118436-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0003.flac", "task_type": "lexical_counting", "query": "guard", "prompt": "After the audio ends, tally every instance of the word \"guard\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[8.39, 8.71, "one"], [9.84, 10.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.39,\"end\":8.71},{\"word\":\"two\",\"start\":9.84,\"end\":10.18}]"}
{"utterance_id": "5895-34629-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0022.flac", "task_type": "lexical_detection", "query": "applause", "prompt": "After the audio ends, extract all occurrences of the word \"applause\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.74, "applause"]], "response_mode": "offline", "answer": "[{\"word\":\"applause\",\"start\":2.25,\"end\":2.74}]"}
{"utterance_id": "251-118436-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0011.flac", "task_type": "lexical_detection", "query": "portions", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"portions\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.21, 1.72, "portions"]], "response_mode": "offline", "answer": "[{\"word\":\"portions\",\"start\":1.21,\"end\":1.72}]"}
{"utterance_id": "7850-286674-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0016.flac", "task_type": "lexical_detection", "query": "shell", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"shell\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.55, 6.86, "shell"]], "response_mode": "offline", "answer": "[{\"word\":\"shell\",\"start\":6.55,\"end\":6.86}]"}
{"utterance_id": "1919-142785-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0022.flac", "task_type": "lexical_detection", "query": "eggs", "prompt": "When the utterance is complete, report all instances of the word \"eggs\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.04, 5.33, "eggs"]], "response_mode": "offline", "answer": "[{\"word\":\"eggs\",\"start\":5.04,\"end\":5.33}]"}
{"utterance_id": "3000-15664-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[17.69, 18.22, "bee"]], "response_mode": "offline", "answer": "[{\"word\":\"bee\",\"start\":17.69,\"end\":18.22}]"}
{"utterance_id": "2035-152373-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.76, 4.62, "fifteen thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen thousand\",\"start\":3.76,\"end\":4.62}]"}
{"utterance_id": "3081-166546-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0048.flac", "task_type": "lexical_detection", "query": "events", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"events\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.55, 4.29, "events"]], "response_mode": "offline", "answer": "[{\"word\":\"events\",\"start\":3.55,\"end\":4.29}]"}
{"utterance_id": "2277-149874-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0004.flac", "task_type": "lexical_detection", "query": "build", "prompt": "When the utterance is complete, report all instances of the word \"build\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 1.91, "build"]], "response_mode": "offline", "answer": "[{\"word\":\"build\",\"start\":1.67,\"end\":1.91}]"}
{"utterance_id": "2086-149214-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0002.flac", "task_type": "lexical_detection", "query": "author", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"author\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 0.93, "author"]], "response_mode": "offline", "answer": "[{\"word\":\"author\",\"start\":0.62,\"end\":0.93}]"}
{"utterance_id": "2078-142845-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0037.flac", "task_type": "lexical_detection", "query": "centre", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"centre\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.91, 7.42, "centre"]], "response_mode": "offline", "answer": "[{\"word\":\"centre\",\"start\":6.91,\"end\":7.42}]"}
{"utterance_id": "84-121123-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0017.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "After the audio ends, tally every instance of a family relation. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.29, 0.71, "one"], [2.46, 2.9, "two"], [8.6, 9.07, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.29,\"end\":0.71},{\"word\":\"two\",\"start\":2.46,\"end\":2.9},{\"word\":\"three\",\"start\":8.6,\"end\":9.07}]"}
{"utterance_id": "2428-83699-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0035.flac", "task_type": "lexical_detection", "query": "inquired", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"inquired\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.7, 1.32, "inquired"]], "response_mode": "offline", "answer": "[{\"word\":\"inquired\",\"start\":0.7,\"end\":1.32}]"}
{"utterance_id": "7976-105575-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0018.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.45, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":1.03,\"end\":1.45}]"}
{"utterance_id": "422-122949-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0032.flac", "task_type": "lexical_detection", "query": "teeth", "prompt": "After the audio ends, extract all occurrences of the word \"teeth\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.6, 4.08, "teeth"]], "response_mode": "offline", "answer": "[{\"word\":\"teeth\",\"start\":3.6,\"end\":4.08}]"}
{"utterance_id": "1462-170142-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0024.flac", "task_type": "lexical_detection", "query": "first", "prompt": "Process the full audio offline. Find every instance of the word \"first\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.83, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.25,\"end\":1.83}]"}
{"utterance_id": "1919-142785-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0032.flac", "task_type": "semantic_counting", "query": "food", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a food name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.82, 1.22, "one"], [3.26, 3.75, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.82,\"end\":1.22},{\"word\":\"two\",\"start\":3.26,\"end\":3.75}]"}
{"utterance_id": "6345-64257-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0019.flac", "task_type": "lexical_detection", "query": "careless", "prompt": "After the audio ends, extract all occurrences of the word \"careless\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.65, 7.37, "careless"]], "response_mode": "offline", "answer": "[{\"word\":\"careless\",\"start\":6.65,\"end\":7.37}]"}
{"utterance_id": "2412-153948-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.64, 2.84, "grey"]], "response_mode": "offline", "answer": "[{\"word\":\"grey\",\"start\":2.64,\"end\":2.84}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.66, 11.91, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":11.66,\"end\":11.91}]"}
{"utterance_id": "2428-83705-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0026.flac", "task_type": "lexical_detection", "query": "doing", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"doing\" with fields word, start, end (seconds). If none, reply [].", "labels": [[8.97, 9.28, "doing"]], "response_mode": "offline", "answer": "[{\"word\":\"doing\",\"start\":8.97,\"end\":9.28}]"}
{"utterance_id": "1919-142785-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0056.flac", "task_type": "lexical_detection", "query": "crumbs", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"crumbs\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.95, "crumbs"]], "response_mode": "offline", "answer": "[{\"word\":\"crumbs\",\"start\":1.35,\"end\":1.95}]"}
{"utterance_id": "8297-275154-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0011.flac", "task_type": "lexical_detection", "query": "glass", "prompt": "Process the full audio offline. Find every instance of the word \"glass\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 2.13, "glass"]], "response_mode": "offline", "answer": "[{\"word\":\"glass\",\"start\":1.65,\"end\":2.13}]"}
{"utterance_id": "2086-149220-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[18.81, 19.33, "limbs"]], "response_mode": "offline", "answer": "[{\"word\":\"limbs\",\"start\":18.81,\"end\":19.33}]"}
{"utterance_id": "6345-93302-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0016.flac", "task_type": "lexical_detection", "query": "deeply", "prompt": "When the utterance is complete, report all instances of the word \"deeply\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.92, 6.39, "deeply"]], "response_mode": "offline", "answer": "[{\"word\":\"deeply\",\"start\":5.92,\"end\":6.39}]"}
{"utterance_id": "84-121550-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.37, 3.64, "lids"]], "response_mode": "offline", "answer": "[{\"word\":\"lids\",\"start\":3.37,\"end\":3.64}]"}
{"utterance_id": "5338-284437-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0012.flac", "task_type": "lexical_detection", "query": "evident", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"evident\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.12, "evident"]], "response_mode": "offline", "answer": "[{\"word\":\"evident\",\"start\":1.74,\"end\":2.12}]"}
{"utterance_id": "2902-9006-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.21, 4.41, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.21,\"end\":4.41}]"}
{"utterance_id": "1272-135031-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0002.flac", "task_type": "lexical_detection", "query": "stepped", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"stepped\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.53, 6.84, "stepped"]], "response_mode": "offline", "answer": "[{\"word\":\"stepped\",\"start\":6.53,\"end\":6.84}]"}
{"utterance_id": "5895-34615-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0007.flac", "task_type": "lexical_detection", "query": "laughter", "prompt": "After listening to the whole recording, output every instance of the word \"laughter\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 2.97, "laughter"]], "response_mode": "offline", "answer": "[{\"word\":\"laughter\",\"start\":2.52,\"end\":2.97}]"}
{"utterance_id": "3752-4943-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0022.flac", "task_type": "lexical_detection", "query": "giant", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"giant\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.15, 2.65, "giant"]], "response_mode": "offline", "answer": "[{\"word\":\"giant\",\"start\":2.15,\"end\":2.65}]"}
{"utterance_id": "5895-34622-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0008.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.65, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":2.08,\"end\":2.65}]"}
{"utterance_id": "3752-4943-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 2.93, "watchman"]], "response_mode": "offline", "answer": "[{\"word\":\"watchman\",\"start\":2.32,\"end\":2.93}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.77, 5.2, "soul"], [7.41, 7.77, "soul"], [10.67, 11.09, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"soul\",\"start\":4.77,\"end\":5.2},{\"word\":\"soul\",\"start\":7.41,\"end\":7.77},{\"word\":\"body\",\"start\":10.67,\"end\":11.09}]"}
{"utterance_id": "251-136532-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0014.flac", "task_type": "lexical_detection", "query": "charcoal", "prompt": "Once the audio has finished, identify every instance of the word \"charcoal\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.3, 2.25, "charcoal"]], "response_mode": "offline", "answer": "[{\"word\":\"charcoal\",\"start\":1.3,\"end\":2.25}]"}
{"utterance_id": "2428-83699-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0006.flac", "task_type": "lexical_detection", "query": "madge", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"madge\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.99, 5.5, "madge"]], "response_mode": "offline", "answer": "[{\"word\":\"madge\",\"start\":4.99,\"end\":5.5}]"}
{"utterance_id": "7976-110124-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0007.flac", "task_type": "lexical_detection", "query": "old", "prompt": "After the audio ends, extract all occurrences of the word \"old\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.1, "old"]], "response_mode": "offline", "answer": "[{\"word\":\"old\",\"start\":0.78,\"end\":1.1}]"}
{"utterance_id": "3576-138058-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0026.flac", "task_type": "lexical_detection", "query": "fate", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"fate\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.98, 3.45, "fate"]], "response_mode": "offline", "answer": "[{\"word\":\"fate\",\"start\":2.98,\"end\":3.45}]"}
{"utterance_id": "1919-142785-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0009.flac", "task_type": "lexical_detection", "query": "seasonable", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"seasonable\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.18, "seasonable"]], "response_mode": "offline", "answer": "[{\"word\":\"seasonable\",\"start\":0.52,\"end\":1.18}]"}
{"utterance_id": "1993-147964-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0005.flac", "task_type": "lexical_detection", "query": "mounted", "prompt": "Once the audio has finished, identify every instance of the word \"mounted\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.78, 1.18, "mounted"]], "response_mode": "offline", "answer": "[{\"word\":\"mounted\",\"start\":0.78,\"end\":1.18}]"}
{"utterance_id": "84-121550-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0015.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[1.86, 2.31, "one"], [4.39, 4.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.86,\"end\":2.31},{\"word\":\"two\",\"start\":4.39,\"end\":4.67}]"}
{"utterance_id": "6345-93306-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.9, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":0.67,\"end\":0.9}]"}
{"utterance_id": "6319-57405-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0010.flac", "task_type": "lexical_detection", "query": "pyrrha", "prompt": "When the utterance is complete, report all instances of the word \"pyrrha\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.48, 2.9, "pyrrha"]], "response_mode": "offline", "answer": "[{\"word\":\"pyrrha\",\"start\":2.48,\"end\":2.9}]"}
{"utterance_id": "5895-34629-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 2.08, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":1.39,\"end\":2.08}]"}
{"utterance_id": "7850-73752-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0015.flac", "task_type": "lexical_detection", "query": "fixed", "prompt": "Process the full audio offline. Find every instance of the word \"fixed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.96, 4.4, "fixed"]], "response_mode": "offline", "answer": "[{\"word\":\"fixed\",\"start\":3.96,\"end\":4.4}]"}
{"utterance_id": "2428-83699-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0022.flac", "task_type": "lexical_detection", "query": "meant", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"meant\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.87, "meant"]], "response_mode": "offline", "answer": "[{\"word\":\"meant\",\"start\":1.37,\"end\":1.87}]"}
{"utterance_id": "2277-149897-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.08, 9.34, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.08,\"end\":9.34}]"}
{"utterance_id": "2035-147961-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0033.flac", "task_type": "lexical_detection", "query": "run", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"run\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.61, 0.79, "run"]], "response_mode": "offline", "answer": "[{\"word\":\"run\",\"start\":0.61,\"end\":0.79}]"}
{"utterance_id": "1673-143396-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0014.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.81, 7.2, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":6.81,\"end\":7.2}]"}
{"utterance_id": "652-130726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0001.flac", "task_type": "semantic_counting", "query": "city", "prompt": "After the audio ends, tally every instance of a city name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.25, 2.68, "one"], [3.71, 4.22, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.25,\"end\":2.68},{\"word\":\"two\",\"start\":3.71,\"end\":4.22}]"}
{"utterance_id": "5536-43359-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0003.flac", "task_type": "lexical_counting", "query": "mother", "prompt": "After the audio ends, tally every instance of the word \"mother\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.13, 1.48, "one"], [3.05, 3.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.13,\"end\":1.48},{\"word\":\"two\",\"start\":3.05,\"end\":3.46}]"}
{"utterance_id": "2078-142845-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0009.flac", "task_type": "lexical_detection", "query": "millet", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"millet\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.44, 3.07, "millet"]], "response_mode": "offline", "answer": "[{\"word\":\"millet\",\"start\":2.44,\"end\":3.07}]"}
{"utterance_id": "652-129742-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0007.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.64, 4.02, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":3.64,\"end\":4.02}]"}
{"utterance_id": "251-137823-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0022.flac", "task_type": "lexical_detection", "query": "concerned", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"concerned\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.38, 5.95, "concerned"]], "response_mode": "offline", "answer": "[{\"word\":\"concerned\",\"start\":5.38,\"end\":5.95}]"}
{"utterance_id": "2035-147960-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.54, "shoulder"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulder\",\"start\":1.93,\"end\":2.54}]"}
{"utterance_id": "1272-128104-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.53, 3.35, "mister ruskin"]], "response_mode": "offline", "answer": "[{\"word\":\"mister ruskin\",\"start\":2.53,\"end\":3.35}]"}
{"utterance_id": "3081-166546-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0045.flac", "task_type": "lexical_detection", "query": "favourite", "prompt": "Once the audio has finished, identify every instance of the word \"favourite\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.37, 8.96, "favourite"]], "response_mode": "offline", "answer": "[{\"word\":\"favourite\",\"start\":8.37,\"end\":8.96}]"}
{"utterance_id": "2428-83705-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0040.flac", "task_type": "lexical_detection", "query": "honeymoon", "prompt": "After the audio ends, extract all occurrences of the word \"honeymoon\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.15, "honeymoon"]], "response_mode": "offline", "answer": "[{\"word\":\"honeymoon\",\"start\":1.67,\"end\":2.15}]"}
{"utterance_id": "5536-43359-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0005.flac", "task_type": "lexical_detection", "query": "traditions", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"traditions\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.04, 5.68, "traditions"]], "response_mode": "offline", "answer": "[{\"word\":\"traditions\",\"start\":5.04,\"end\":5.68}]"}
{"utterance_id": "6345-93302-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0018.flac", "task_type": "lexical_detection", "query": "idiot", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"idiot\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.78, 5.09, "idiot"]], "response_mode": "offline", "answer": "[{\"word\":\"idiot\",\"start\":4.78,\"end\":5.09}]"}
{"utterance_id": "8842-302201-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0005.flac", "task_type": "lexical_detection", "query": "courage", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"courage\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.53, 5.92, "courage"]], "response_mode": "offline", "answer": "[{\"word\":\"courage\",\"start\":5.53,\"end\":5.92}]"}
{"utterance_id": "6345-64257-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0009.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Once the audio has finished, identify every instance of a date. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.25, 10.78, "morrow"]], "response_mode": "offline", "answer": "[{\"word\":\"morrow\",\"start\":10.25,\"end\":10.78}]"}
{"utterance_id": "5338-24640-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0007.flac", "task_type": "lexical_detection", "query": "innocence", "prompt": "Process the full audio offline. Find every instance of the word \"innocence\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.34, "innocence"]], "response_mode": "offline", "answer": "[{\"word\":\"innocence\",\"start\":1.61,\"end\":2.34}]"}
{"utterance_id": "5895-34622-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0012.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.39, 1.64, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.39,\"end\":1.64}]"}
{"utterance_id": "1673-143397-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 1.87, "memnon"]], "response_mode": "offline", "answer": "[{\"word\":\"memnon\",\"start\":1.32,\"end\":1.87}]"}
{"utterance_id": "2803-161169-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0004.flac", "task_type": "lexical_detection", "query": "entrance", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"entrance\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 1.16, "entrance"]], "response_mode": "offline", "answer": "[{\"word\":\"entrance\",\"start\":0.56,\"end\":1.16}]"}
{"utterance_id": "2803-161169-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0017.flac", "task_type": "lexical_detection", "query": "rough", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"rough\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.1, "rough"]], "response_mode": "offline", "answer": "[{\"word\":\"rough\",\"start\":2.9,\"end\":3.1}]"}
{"utterance_id": "3576-138058-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0029.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.85, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":0.36,\"end\":0.85}]"}
{"utterance_id": "2078-142845-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0019.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.9, 2.28, "maize"]], "response_mode": "offline", "answer": "[{\"word\":\"maize\",\"start\":1.9,\"end\":2.28}]"}
{"utterance_id": "652-130726-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0030.flac", "task_type": "lexical_detection", "query": "season", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"season\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.23, 0.61, "season"]], "response_mode": "offline", "answer": "[{\"word\":\"season\",\"start\":0.23,\"end\":0.61}]"}
{"utterance_id": "5694-64029-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0022.flac", "task_type": "lexical_detection", "query": "storm", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"storm\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.01, 1.45, "storm"]], "response_mode": "offline", "answer": "[{\"word\":\"storm\",\"start\":1.01,\"end\":1.45}]"}
{"utterance_id": "5536-43359-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0013.flac", "task_type": "lexical_detection", "query": "purest", "prompt": "Process the full audio offline. Find every instance of the word \"purest\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.22, 4.71, "purest"]], "response_mode": "offline", "answer": "[{\"word\":\"purest\",\"start\":4.22,\"end\":4.71}]"}
{"utterance_id": "174-50561-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0007.flac", "task_type": "lexical_detection", "query": "leagues", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"leagues\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.2, 4.59, "leagues"]], "response_mode": "offline", "answer": "[{\"word\":\"leagues\",\"start\":4.2,\"end\":4.59}]"}
{"utterance_id": "5694-64025-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0022.flac", "task_type": "lexical_detection", "query": "rope", "prompt": "Once the audio has finished, identify every instance of the word \"rope\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.49, 0.78, "rope"], [5.73, 6.18, "rope"]], "response_mode": "offline", "answer": "[{\"word\":\"rope\",\"start\":0.49,\"end\":0.78},{\"word\":\"rope\",\"start\":5.73,\"end\":6.18}]"}
{"utterance_id": "174-168635-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0003.flac", "task_type": "lexical_detection", "query": "pangs", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"pangs\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.1, "pangs"]], "response_mode": "offline", "answer": "[{\"word\":\"pangs\",\"start\":1.6,\"end\":2.1}]"}
{"utterance_id": "2902-9006-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0001.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "Listen to the full audio. Count every occurrence of an organization name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.14, 4.64, "one"], [19.29, 19.76, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.14,\"end\":4.64},{\"word\":\"two\",\"start\":19.29,\"end\":19.76}]"}
{"utterance_id": "1462-170138-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0014.flac", "task_type": "lexical_detection", "query": "nodded", "prompt": "When the utterance is complete, report all instances of the word \"nodded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 0.97, "nodded"]], "response_mode": "offline", "answer": "[{\"word\":\"nodded\",\"start\":0.58,\"end\":0.97}]"}
{"utterance_id": "1673-143397-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.36, 13.98, "nestorius"]], "response_mode": "offline", "answer": "[{\"word\":\"nestorius\",\"start\":13.36,\"end\":13.98}]"}
{"utterance_id": "174-168635-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0004.flac", "task_type": "lexical_detection", "query": "flowed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"flowed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.14, 9.53, "flowed"]], "response_mode": "offline", "answer": "[{\"word\":\"flowed\",\"start\":9.14,\"end\":9.53}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.03, 4.53, "fifteen"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen\",\"start\":4.03,\"end\":4.53}]"}
{"utterance_id": "251-136532-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 3.93, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.58,\"end\":3.93}]"}
{"utterance_id": "2428-83705-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0036.flac", "task_type": "lexical_detection", "query": "someone", "prompt": "Process the full audio offline. Find every instance of the word \"someone\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.93, "someone"]], "response_mode": "offline", "answer": "[{\"word\":\"someone\",\"start\":0.53,\"end\":0.93}]"}
{"utterance_id": "1462-170142-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0006.flac", "task_type": "lexical_detection", "query": "cheek", "prompt": "When the utterance is complete, report all instances of the word \"cheek\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.92, "cheek"]], "response_mode": "offline", "answer": "[{\"word\":\"cheek\",\"start\":1.62,\"end\":1.92}]"}
{"utterance_id": "8842-304647-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0002.flac", "task_type": "lexical_detection", "query": "reigned", "prompt": "Once the audio has finished, identify every instance of the word \"reigned\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.93, 8.23, "reigned"]], "response_mode": "offline", "answer": "[{\"word\":\"reigned\",\"start\":7.93,\"end\":8.23}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_counting", "query": "city", "prompt": "After listening to the whole recording, output timed count words for a city name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[4.65, 5.21, "one"], [28.33, 29.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.65,\"end\":5.21},{\"word\":\"two\",\"start\":28.33,\"end\":29.06}]"}
{"utterance_id": "777-126732-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0046.flac", "task_type": "lexical_detection", "query": "case", "prompt": "Process the full audio offline. Find every instance of the word \"case\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 0.95, "case"]], "response_mode": "offline", "answer": "[{\"word\":\"case\",\"start\":0.7,\"end\":0.95}]"}
{"utterance_id": "6345-64257-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0004.flac", "task_type": "lexical_detection", "query": "nearly", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"nearly\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.57, 1.9, "nearly"]], "response_mode": "offline", "answer": "[{\"word\":\"nearly\",\"start\":1.57,\"end\":1.9}]"}
{"utterance_id": "1462-170138-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0026.flac", "task_type": "lexical_detection", "query": "glad", "prompt": "When the utterance is complete, report all instances of the word \"glad\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 0.88, "glad"]], "response_mode": "offline", "answer": "[{\"word\":\"glad\",\"start\":0.59,\"end\":0.88}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "lexical_counting", "query": "wants", "prompt": "After listening to the whole recording, output timed count words for the word \"wants\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[7.73, 8.31, "one"], [16.44, 17.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.73,\"end\":8.31},{\"word\":\"two\",\"start\":16.44,\"end\":17.06}]"}
{"utterance_id": "8842-304647-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0002.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the full audio. Count every occurrence of a body part. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.93, "one"], [11.67, 12.04, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.4,\"end\":0.93},{\"word\":\"two\",\"start\":11.67,\"end\":12.04}]"}
{"utterance_id": "7850-73752-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.91, 11.45, "ferdinand"], [12.97, 13.68, "henrietta"]], "response_mode": "offline", "answer": "[{\"word\":\"ferdinand\",\"start\":10.91,\"end\":11.45},{\"word\":\"henrietta\",\"start\":12.97,\"end\":13.68}]"}
{"utterance_id": "2803-154328-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0006.flac", "task_type": "lexical_detection", "query": "till", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"till\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.42, 2.64, "till"]], "response_mode": "offline", "answer": "[{\"word\":\"till\",\"start\":2.42,\"end\":2.64}]"}
{"utterance_id": "1993-147149-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0029.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.42, 11.17, "dressmaker"]], "response_mode": "offline", "answer": "[{\"word\":\"dressmaker\",\"start\":10.42,\"end\":11.17}]"}
{"utterance_id": "6313-66129-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.34, "tad's"]], "response_mode": "offline", "answer": "[{\"word\":\"tad's\",\"start\":1.0,\"end\":1.34}]"}
{"utterance_id": "5536-43363-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0014.flac", "task_type": "lexical_detection", "query": "saved", "prompt": "When the utterance is complete, report all instances of the word \"saved\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 2.94, "saved"]], "response_mode": "offline", "answer": "[{\"word\":\"saved\",\"start\":2.61,\"end\":2.94}]"}
{"utterance_id": "5694-64038-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0024.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After listening to the whole recording, output every instance of an organization name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.82, "regiment"], [2.82, 3.29, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":0.42,\"end\":0.82},{\"word\":\"regiment\",\"start\":2.82,\"end\":3.29}]"}
{"utterance_id": "2428-83705-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.36, "five hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"five hundred\",\"start\":1.79,\"end\":2.36}]"}
{"utterance_id": "2086-149220-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0039.flac", "task_type": "lexical_detection", "query": "altogether", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"altogether\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.55, "altogether"]], "response_mode": "offline", "answer": "[{\"word\":\"altogether\",\"start\":1.06,\"end\":1.55}]"}
{"utterance_id": "6313-66129-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.2, 0.55, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":0.2,\"end\":0.55}]"}
{"utterance_id": "1272-141231-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0024.flac", "task_type": "lexical_detection", "query": "impossible", "prompt": "After the audio ends, extract all occurrences of the word \"impossible\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 2.2, "impossible"]], "response_mode": "offline", "answer": "[{\"word\":\"impossible\",\"start\":1.45,\"end\":2.2}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a profession (word, start, end). If none, reply [].", "labels": [[2.9, 3.54, "one"], [10.12, 10.75, "two"], [17.88, 18.38, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.9,\"end\":3.54},{\"word\":\"two\",\"start\":10.12,\"end\":10.75},{\"word\":\"three\",\"start\":17.88,\"end\":18.38}]"}
{"utterance_id": "1988-147956-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0011.flac", "task_type": "lexical_detection", "query": "cheeks", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"cheeks\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.44, 2.89, "cheeks"]], "response_mode": "offline", "answer": "[{\"word\":\"cheeks\",\"start\":2.44,\"end\":2.89}]"}
{"utterance_id": "1919-142785-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0046.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.23, "basil"]], "response_mode": "offline", "answer": "[{\"word\":\"basil\",\"start\":1.69,\"end\":2.23}]"}
{"utterance_id": "2277-149896-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0014.flac", "task_type": "lexical_detection", "query": "headed", "prompt": "Process the full audio offline. Find every instance of the word \"headed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 2.9, "headed"]], "response_mode": "offline", "answer": "[{\"word\":\"headed\",\"start\":2.61,\"end\":2.9}]"}
{"utterance_id": "3752-4943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0007.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.5, "pale"], [5.8, 6.16, "blue"]], "response_mode": "offline", "answer": "[{\"word\":\"pale\",\"start\":2.01,\"end\":2.5},{\"word\":\"blue\",\"start\":5.8,\"end\":6.16}]"}
{"utterance_id": "6313-66129-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0029.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.4, 6.79, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":6.4,\"end\":6.79}]"}
{"utterance_id": "1272-135031-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0023.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a person's name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.5, 0.97, "one"], [3.7, 4.09, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.5,\"end\":0.97},{\"word\":\"two\",\"start\":3.7,\"end\":4.09}]"}
{"utterance_id": "1462-170138-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0027.flac", "task_type": "lexical_detection", "query": "poverty", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"poverty\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.75, "poverty"]], "response_mode": "offline", "answer": "[{\"word\":\"poverty\",\"start\":1.2,\"end\":1.75}]"}
{"utterance_id": "7976-110124-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0001.flac", "task_type": "lexical_detection", "query": "distant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"distant\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.47, 4.92, "distant"]], "response_mode": "offline", "answer": "[{\"word\":\"distant\",\"start\":4.47,\"end\":4.92}]"}
{"utterance_id": "6295-64301-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0005.flac", "task_type": "lexical_detection", "query": "blessed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"blessed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.86, "blessed"], [3.8, 4.17, "blessed"]], "response_mode": "offline", "answer": "[{\"word\":\"blessed\",\"start\":0.4,\"end\":0.86},{\"word\":\"blessed\",\"start\":3.8,\"end\":4.17}]"}
{"utterance_id": "6313-76958-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0004.flac", "task_type": "lexical_detection", "query": "admitted", "prompt": "After listening to the whole recording, output every instance of the word \"admitted\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.2, "admitted"]], "response_mode": "offline", "answer": "[{\"word\":\"admitted\",\"start\":1.78,\"end\":2.2}]"}
{"utterance_id": "2078-142845-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0036.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.65, 3.4, "seventeen twenty nine"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen twenty nine\",\"start\":1.65,\"end\":3.4}]"}
{"utterance_id": "5338-284437-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0020.flac", "task_type": "lexical_detection", "query": "agent", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"agent\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.1, "agent"]], "response_mode": "offline", "answer": "[{\"word\":\"agent\",\"start\":1.61,\"end\":2.1}]"}
{"utterance_id": "2428-83699-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0025.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After listening to the whole recording, output every instance of a date with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 3.08, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":2.4,\"end\":3.08}]"}
{"utterance_id": "652-130737-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0000.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.14, 9.59, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":9.14,\"end\":9.59}]"}
{"utterance_id": "2803-154328-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.94, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":0.36,\"end\":0.94}]"}
{"utterance_id": "422-122949-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0005.flac", "task_type": "lexical_counting", "query": "situation", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"situation\" (word, start, end). If none, reply [].", "labels": [[3.42, 4.02, "one"], [8.61, 9.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.42,\"end\":4.02},{\"word\":\"two\",\"start\":8.61,\"end\":9.47}]"}
{"utterance_id": "1988-24833-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0021.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.95, 1.2, "one"], [2.47, 2.72, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.95,\"end\":1.2},{\"word\":\"two\",\"start\":2.47,\"end\":2.72}]"}
{"utterance_id": "1988-147956-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0010.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 2.02, "conductor"]], "response_mode": "offline", "answer": "[{\"word\":\"conductor\",\"start\":1.52,\"end\":2.02}]"}
{"utterance_id": "174-84280-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.97, 6.56, "justin"]], "response_mode": "offline", "answer": "[{\"word\":\"justin\",\"start\":5.97,\"end\":6.56}]"}
{"utterance_id": "2803-154320-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0014.flac", "task_type": "lexical_detection", "query": "forbid", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"forbid\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.06, 5.04, "forbid"]], "response_mode": "offline", "answer": "[{\"word\":\"forbid\",\"start\":4.06,\"end\":5.04}]"}
{"utterance_id": "1673-143396-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.26, 2.16, "athanasius"], [3.42, 3.97, "julian"]], "response_mode": "offline", "answer": "[{\"word\":\"athanasius\",\"start\":1.26,\"end\":2.16},{\"word\":\"julian\",\"start\":3.42,\"end\":3.97}]"}
{"utterance_id": "2078-142845-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0003.flac", "task_type": "lexical_detection", "query": "eighteen", "prompt": "Once the audio has finished, identify every instance of the word \"eighteen\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.2, 2.0, "eighteen"]], "response_mode": "offline", "answer": "[{\"word\":\"eighteen\",\"start\":1.2,\"end\":2.0}]"}
{"utterance_id": "8842-302201-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0011.flac", "task_type": "lexical_detection", "query": "purpose", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"purpose\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.89, 2.28, "purpose"]], "response_mode": "offline", "answer": "[{\"word\":\"purpose\",\"start\":1.89,\"end\":2.28}]"}
{"utterance_id": "1919-142785-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0051.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.9, 1.2, "udder"]], "response_mode": "offline", "answer": "[{\"word\":\"udder\",\"start\":0.9,\"end\":1.2}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.13, 3.63, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":3.13,\"end\":3.63}]"}
{"utterance_id": "2277-149896-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0019.flac", "task_type": "lexical_detection", "query": "compromised", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"compromised\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.54, 1.16, "compromised"]], "response_mode": "offline", "answer": "[{\"word\":\"compromised\",\"start\":0.54,\"end\":1.16}]"}
{"utterance_id": "8842-304647-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.46, 2.98, "magdalen"]], "response_mode": "offline", "answer": "[{\"word\":\"magdalen\",\"start\":2.46,\"end\":2.98}]"}
{"utterance_id": "3853-163249-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0036.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.65, "fat"]], "response_mode": "offline", "answer": "[{\"word\":\"fat\",\"start\":1.39,\"end\":1.65}]"}
{"utterance_id": "652-129742-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.34, "two"], [4.07, 4.27, "two"], [6.35, 6.62, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.06,\"end\":2.34},{\"word\":\"two\",\"start\":4.07,\"end\":4.27},{\"word\":\"half\",\"start\":6.35,\"end\":6.62}]"}
{"utterance_id": "3576-138058-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0001.flac", "task_type": "lexical_detection", "query": "hunger", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hunger\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 3.9, "hunger"]], "response_mode": "offline", "answer": "[{\"word\":\"hunger\",\"start\":3.48,\"end\":3.9}]"}
{"utterance_id": "251-137823-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[0.94, 1.43, "eyelids"]], "response_mode": "offline", "answer": "[{\"word\":\"eyelids\",\"start\":0.94,\"end\":1.43}]"}
{"utterance_id": "3536-8226-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.36, 5.96, "trevelyan"]], "response_mode": "offline", "answer": "[{\"word\":\"trevelyan\",\"start\":5.36,\"end\":5.96}]"}
{"utterance_id": "1993-147149-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0022.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[11.35, 11.69, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":11.35,\"end\":11.69}]"}
{"utterance_id": "2902-9006-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0001.flac", "task_type": "lexical_counting", "query": "sins", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"sins\" (word, start, end). If none, reply [].", "labels": [[3.6, 3.96, "one"], [9.55, 9.87, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.6,\"end\":3.96},{\"word\":\"two\",\"start\":9.55,\"end\":9.87}]"}
{"utterance_id": "8842-302201-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0004.flac", "task_type": "lexical_counting", "query": "talk", "prompt": "After the audio ends, tally every instance of the word \"talk\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.61, 1.86, "one"], [5.9, 6.25, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.61,\"end\":1.86},{\"word\":\"two\",\"start\":5.9,\"end\":6.25}]"}
{"utterance_id": "5338-284437-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0006.flac", "task_type": "lexical_detection", "query": "ignorant", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"ignorant\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.89, "ignorant"]], "response_mode": "offline", "answer": "[{\"word\":\"ignorant\",\"start\":1.37,\"end\":1.89}]"}
{"utterance_id": "7850-286674-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0003.flac", "task_type": "lexical_counting", "query": "ate", "prompt": "Process the full audio offline. For every occurrence of the word \"ate\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.13, "one"], [4.71, 4.94, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.9,\"end\":1.13},{\"word\":\"two\",\"start\":4.71,\"end\":4.94}]"}
{"utterance_id": "251-137823-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0014.flac", "task_type": "lexical_detection", "query": "frightful", "prompt": "After listening to the whole recording, output every instance of the word \"frightful\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.34, 3.91, "frightful"]], "response_mode": "offline", "answer": "[{\"word\":\"frightful\",\"start\":3.34,\"end\":3.91}]"}
{"utterance_id": "1988-24833-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.02, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.79,\"end\":4.02}]"}
{"utterance_id": "3576-138058-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0016.flac", "task_type": "lexical_detection", "query": "girdle", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"girdle\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.63, 10.95, "girdle"]], "response_mode": "offline", "answer": "[{\"word\":\"girdle\",\"start\":10.63,\"end\":10.95}]"}
{"utterance_id": "5895-34629-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0028.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.84, 2.26, "one"], [3.72, 4.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.84,\"end\":2.26},{\"word\":\"two\",\"start\":3.72,\"end\":4.27}]"}
{"utterance_id": "1272-141231-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.35, 1.79, "brion s"]], "response_mode": "offline", "answer": "[{\"word\":\"brion s\",\"start\":1.35,\"end\":1.79}]"}
{"utterance_id": "3752-4944-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0050.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.13, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":1.7,\"end\":2.13}]"}
{"utterance_id": "6345-93306-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0009.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.13, 1.4, "cake"]], "response_mode": "offline", "answer": "[{\"word\":\"cake\",\"start\":1.13,\"end\":1.4}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "lexical_detection", "query": "touching", "prompt": "Once the audio has finished, identify every instance of the word \"touching\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.84, 5.15, "touching"]], "response_mode": "offline", "answer": "[{\"word\":\"touching\",\"start\":4.84,\"end\":5.15}]"}
{"utterance_id": "422-122949-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0001.flac", "task_type": "lexical_counting", "query": "ruled", "prompt": "Listen to the full audio. Count every occurrence of the word \"ruled\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.18, 7.65, "one"], [8.78, 9.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.18,\"end\":7.65},{\"word\":\"two\",\"start\":8.78,\"end\":9.2}]"}
{"utterance_id": "8842-302196-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.83, 3.17, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.83,\"end\":3.17}]"}
{"utterance_id": "251-137823-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0021.flac", "task_type": "lexical_detection", "query": "swift", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"swift\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.3, "swift"], [4.72, 5.13, "swift"]], "response_mode": "offline", "answer": "[{\"word\":\"swift\",\"start\":0.85,\"end\":1.3},{\"word\":\"swift\",\"start\":4.72,\"end\":5.13}]"}
{"utterance_id": "5895-34622-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0005.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a color name with word/start/end. If none, reply [].", "labels": [[0.55, 0.82, "one"], [6.23, 6.56, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.55,\"end\":0.82},{\"word\":\"two\",\"start\":6.23,\"end\":6.56}]"}
{"utterance_id": "3853-163249-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0002.flac", "task_type": "lexical_detection", "query": "deaf", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"deaf\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.77, "deaf"]], "response_mode": "offline", "answer": "[{\"word\":\"deaf\",\"start\":1.51,\"end\":1.77}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[1.41, 1.95, "one"], [4.96, 5.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.41,\"end\":1.95},{\"word\":\"two\",\"start\":4.96,\"end\":5.53}]"}
{"utterance_id": "652-130737-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0011.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.62, "wines"]], "response_mode": "offline", "answer": "[{\"word\":\"wines\",\"start\":3.1,\"end\":3.62}]"}
{"utterance_id": "2086-149220-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0001.flac", "task_type": "lexical_detection", "query": "propped", "prompt": "After listening to the whole recording, output every instance of the word \"propped\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.97, 3.32, "propped"]], "response_mode": "offline", "answer": "[{\"word\":\"propped\",\"start\":2.97,\"end\":3.32}]"}
{"utterance_id": "3081-166546-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0061.flac", "task_type": "lexical_detection", "query": "happily", "prompt": "Once the audio has finished, identify every instance of the word \"happily\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.7, 3.22, "happily"]], "response_mode": "offline", "answer": "[{\"word\":\"happily\",\"start\":2.7,\"end\":3.22}]"}
{"utterance_id": "652-129742-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.73, 4.08, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"half\",\"start\":3.73,\"end\":4.08}]"}
{"utterance_id": "3000-15664-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0001.flac", "task_type": "lexical_detection", "query": "desolation", "prompt": "Process the full audio offline. Find every instance of the word \"desolation\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 2.37, "desolation"]], "response_mode": "offline", "answer": "[{\"word\":\"desolation\",\"start\":1.48,\"end\":2.37}]"}
{"utterance_id": "1462-170142-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0030.flac", "task_type": "lexical_detection", "query": "wrong", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"wrong\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.6, "wrong"]], "response_mode": "offline", "answer": "[{\"word\":\"wrong\",\"start\":2.19,\"end\":2.6}]"}
{"utterance_id": "1993-147965-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.31, 0.89, "grandfather"]], "response_mode": "offline", "answer": "[{\"word\":\"grandfather\",\"start\":0.31,\"end\":0.89}]"}
{"utterance_id": "2803-154328-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0008.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "When the utterance is complete, report how a person's name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.13, "one"], [3.7, 4.26, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.74,\"end\":2.13},{\"word\":\"two\",\"start\":3.7,\"end\":4.26}]"}
{"utterance_id": "3170-137482-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0008.flac", "task_type": "lexical_detection", "query": "closed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"closed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.41, "closed"]], "response_mode": "offline", "answer": "[{\"word\":\"closed\",\"start\":2.0,\"end\":2.41}]"}
{"utterance_id": "6241-61943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0024.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Do not answer until the audio ends. Then list each instance of a weather term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.33, 2.81, "winds"]], "response_mode": "offline", "answer": "[{\"word\":\"winds\",\"start\":2.33,\"end\":2.81}]"}
{"utterance_id": "2412-153954-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0012.flac", "task_type": "lexical_detection", "query": "walnuts", "prompt": "After listening to the whole recording, output every instance of the word \"walnuts\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.53, "walnuts"]], "response_mode": "offline", "answer": "[{\"word\":\"walnuts\",\"start\":4.04,\"end\":4.53}]"}
{"utterance_id": "5694-64038-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0000.flac", "task_type": "lexical_detection", "query": "tennessee", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"tennessee\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 2.16, "tennessee"]], "response_mode": "offline", "answer": "[{\"word\":\"tennessee\",\"start\":1.4,\"end\":2.16}]"}
{"utterance_id": "3752-4944-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0049.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.12, 2.58, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":2.12,\"end\":2.58}]"}
{"utterance_id": "2078-142845-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0016.flac", "task_type": "lexical_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"plant\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.41, 3.14, "plant"]], "response_mode": "offline", "answer": "[{\"word\":\"plant\",\"start\":2.41,\"end\":3.14}]"}
{"utterance_id": "3081-166546-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0003.flac", "task_type": "lexical_detection", "query": "aunt", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"aunt\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.42, 3.92, "aunt"]], "response_mode": "offline", "answer": "[{\"word\":\"aunt\",\"start\":3.42,\"end\":3.92}]"}
{"utterance_id": "5338-284437-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0029.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.36, 3.89, "sailor"]], "response_mode": "offline", "answer": "[{\"word\":\"sailor\",\"start\":3.36,\"end\":3.89}]"}
{"utterance_id": "5895-34615-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0007.flac", "task_type": "lexical_counting", "query": "laugh", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"laugh\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.45, 1.76, "one"], [4.68, 4.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.45,\"end\":1.76},{\"word\":\"two\",\"start\":4.68,\"end\":4.92}]"}
{"utterance_id": "84-121550-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.17, 0.66, "three"], [4.25, 4.73, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.17,\"end\":0.66},{\"word\":\"one\",\"start\":4.25,\"end\":4.73}]"}
{"utterance_id": "7850-286674-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.44, 0.93, "both"]], "response_mode": "offline", "answer": "[{\"word\":\"both\",\"start\":0.44,\"end\":0.93}]"}
{"utterance_id": "6241-66616-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.96, "wabi"]], "response_mode": "offline", "answer": "[{\"word\":\"wabi\",\"start\":0.54,\"end\":0.96}]"}
{"utterance_id": "2086-149220-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0031.flac", "task_type": "lexical_detection", "query": "imperious", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"imperious\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.98, 4.81, "imperious"]], "response_mode": "offline", "answer": "[{\"word\":\"imperious\",\"start\":3.98,\"end\":4.81}]"}
{"utterance_id": "1462-170142-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0007.flac", "task_type": "lexical_detection", "query": "dreamed", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"dreamed\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 0.94, "dreamed"]], "response_mode": "offline", "answer": "[{\"word\":\"dreamed\",\"start\":0.61,\"end\":0.94}]"}
{"utterance_id": "2412-153948-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0008.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.92, 4.34, "sheep"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":3.92,\"end\":4.34}]"}
{"utterance_id": "3081-166546-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0041.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.23, "gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"gryce\",\"start\":1.6,\"end\":2.23}]"}
{"utterance_id": "2078-142845-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0024.flac", "task_type": "lexical_detection", "query": "twenty", "prompt": "Process the full audio offline. Find every instance of the word \"twenty\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.28, "twenty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty\",\"start\":0.93,\"end\":1.28}]"}
{"utterance_id": "1988-24833-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0023.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.24, 0.45, "one"], [1.85, 2.26, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.24,\"end\":0.45},{\"word\":\"two\",\"start\":1.85,\"end\":2.26}]"}
{"utterance_id": "2035-147960-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0008.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.45, 3.78, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":3.45,\"end\":3.78}]"}
{"utterance_id": "251-118436-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0018.flac", "task_type": "lexical_detection", "query": "brings", "prompt": "After listening to the whole recording, output every instance of the word \"brings\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 1.72, "brings"]], "response_mode": "offline", "answer": "[{\"word\":\"brings\",\"start\":1.38,\"end\":1.72}]"}
{"utterance_id": "5338-24615-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0003.flac", "task_type": "lexical_detection", "query": "remedy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"remedy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.54, "remedy"]], "response_mode": "offline", "answer": "[{\"word\":\"remedy\",\"start\":1.06,\"end\":1.54}]"}
{"utterance_id": "251-137823-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0019.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.81, 6.43, "harkness"]], "response_mode": "offline", "answer": "[{\"word\":\"harkness\",\"start\":5.81,\"end\":6.43}]"}
{"utterance_id": "8297-275154-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0022.flac", "task_type": "lexical_detection", "query": "can't", "prompt": "Once the audio has finished, identify every instance of the word \"can't\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.73, 1.23, "can't"]], "response_mode": "offline", "answer": "[{\"word\":\"can't\",\"start\":0.73,\"end\":1.23}]"}
{"utterance_id": "6241-61943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0018.flac", "task_type": "lexical_detection", "query": "difficulty", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"difficulty\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.42, 5.95, "difficulty"]], "response_mode": "offline", "answer": "[{\"word\":\"difficulty\",\"start\":5.42,\"end\":5.95}]"}
{"utterance_id": "1919-142785-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0035.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 3.13, "mackerel"]], "response_mode": "offline", "answer": "[{\"word\":\"mackerel\",\"start\":2.71,\"end\":3.13}]"}
{"utterance_id": "1919-142785-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0059.flac", "task_type": "lexical_detection", "query": "crisp", "prompt": "When the utterance is complete, report all instances of the word \"crisp\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.33, "crisp"]], "response_mode": "offline", "answer": "[{\"word\":\"crisp\",\"start\":0.84,\"end\":1.33}]"}
{"utterance_id": "3752-4944-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[4.08, 4.43, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":4.08,\"end\":4.43}]"}
{"utterance_id": "1272-141231-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0002.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the full audio. Count every occurrence of a body part. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.09, 1.5, "one"], [5.09, 5.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.09,\"end\":1.5},{\"word\":\"two\",\"start\":5.09,\"end\":5.66}]"}
{"utterance_id": "6295-64301-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0002.flac", "task_type": "lexical_detection", "query": "strode", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"strode\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 5.0, "strode"]], "response_mode": "offline", "answer": "[{\"word\":\"strode\",\"start\":4.44,\"end\":5.0}]"}
{"utterance_id": "6295-64301-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.35, 4.74, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":4.35,\"end\":4.74}]"}
{"utterance_id": "6313-76958-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0031.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.02, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.69,\"end\":2.02}]"}
{"utterance_id": "3853-163249-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0044.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.76, 1.23, "one"], [2.57, 3.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.76,\"end\":1.23},{\"word\":\"two\",\"start\":2.57,\"end\":3.12}]"}
{"utterance_id": "3752-4943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0018.flac", "task_type": "lexical_detection", "query": "kirkland", "prompt": "After listening to the whole recording, output every instance of the word \"kirkland\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 4.97, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":4.47,\"end\":4.97}]"}
{"utterance_id": "652-129742-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0003.flac", "task_type": "lexical_counting", "query": "cabbage", "prompt": "Once the audio has finished, list each occurrence of the word \"cabbage\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.51, 0.91, "one"], [5.28, 5.75, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.91},{\"word\":\"two\",\"start\":5.28,\"end\":5.75}]"}
{"utterance_id": "3081-166546-0072", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0072.flac", "task_type": "lexical_detection", "query": "instance", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"instance\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.24, 2.95, "instance"]], "response_mode": "offline", "answer": "[{\"word\":\"instance\",\"start\":2.24,\"end\":2.95}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.79, 10.17, "sheep"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":9.79,\"end\":10.17}]"}
{"utterance_id": "6313-66125-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0002.flac", "task_type": "lexical_detection", "query": "bushes", "prompt": "When the utterance is complete, report all instances of the word \"bushes\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.42, 2.91, "bushes"]], "response_mode": "offline", "answer": "[{\"word\":\"bushes\",\"start\":2.42,\"end\":2.91}]"}
{"utterance_id": "5536-43359-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0004.flac", "task_type": "lexical_detection", "query": "takes", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"takes\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.84, "takes"]], "response_mode": "offline", "answer": "[{\"word\":\"takes\",\"start\":1.54,\"end\":1.84}]"}
{"utterance_id": "1919-142785-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0020.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.65, 1.2, "juice"], [8.57, 8.9, "oil"]], "response_mode": "offline", "answer": "[{\"word\":\"juice\",\"start\":0.65,\"end\":1.2},{\"word\":\"oil\",\"start\":8.57,\"end\":8.9}]"}
{"utterance_id": "2803-154320-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0011.flac", "task_type": "lexical_detection", "query": "glenarvan", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"glenarvan\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.07, 1.96, "glenarvan"]], "response_mode": "offline", "answer": "[{\"word\":\"glenarvan\",\"start\":1.07,\"end\":1.96}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.87, 6.23, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.87,\"end\":6.23}]"}
{"utterance_id": "6319-57405-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0006.flac", "task_type": "lexical_detection", "query": "pyrrha", "prompt": "Once the audio has finished, identify every instance of the word \"pyrrha\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.4, 1.87, "pyrrha"]], "response_mode": "offline", "answer": "[{\"word\":\"pyrrha\",\"start\":1.4,\"end\":1.87}]"}
{"utterance_id": "1988-148538-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0013.flac", "task_type": "lexical_detection", "query": "unmoved", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"unmoved\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.41, "unmoved"]], "response_mode": "offline", "answer": "[{\"word\":\"unmoved\",\"start\":0.82,\"end\":1.41}]"}
{"utterance_id": "6295-244435-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0023.flac", "task_type": "lexical_detection", "query": "share", "prompt": "After listening to the whole recording, output every instance of the word \"share\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.57, 2.89, "share"]], "response_mode": "offline", "answer": "[{\"word\":\"share\",\"start\":2.57,\"end\":2.89}]"}
{"utterance_id": "3752-4944-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0028.flac", "task_type": "lexical_detection", "query": "another", "prompt": "Process the full audio offline. Find every instance of the word \"another\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.25, 0.61, "another"]], "response_mode": "offline", "answer": "[{\"word\":\"another\",\"start\":0.25,\"end\":0.61}]"}
{"utterance_id": "8297-275156-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0006.flac", "task_type": "lexical_detection", "query": "interview", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"interview\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.72, 6.25, "interview"]], "response_mode": "offline", "answer": "[{\"word\":\"interview\",\"start\":5.72,\"end\":6.25}]"}
{"utterance_id": "1272-135031-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.19, "shaggy"]], "response_mode": "offline", "answer": "[{\"word\":\"shaggy\",\"start\":0.71,\"end\":1.19}]"}
{"utterance_id": "1919-142785-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0005.flac", "task_type": "lexical_counting", "query": "cucumbers", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"cucumbers\" (word, start, end). If none, reply [].", "labels": [[2.46, 3.11, "one"], [11.83, 12.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.46,\"end\":3.11},{\"word\":\"two\",\"start\":11.83,\"end\":12.38}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "lexical_counting", "query": "apple", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"apple\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[4.45, 4.83, "one"], [7.58, 7.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.45,\"end\":4.83},{\"word\":\"two\",\"start\":7.58,\"end\":7.92}]"}
{"utterance_id": "5895-34615-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0011.flac", "task_type": "lexical_detection", "query": "everlasting", "prompt": "After the audio ends, extract all occurrences of the word \"everlasting\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.76, "everlasting"]], "response_mode": "offline", "answer": "[{\"word\":\"everlasting\",\"start\":0.8,\"end\":1.76}]"}
{"utterance_id": "6345-93306-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.21, 3.46, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":3.21,\"end\":3.46}]"}
{"utterance_id": "3081-166546-0082", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0082.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.06, 2.52, "one"], [3.68, 4.29, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.52},{\"word\":\"one\",\"start\":3.68,\"end\":4.29}]"}
{"utterance_id": "84-121550-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0006.flac", "task_type": "lexical_counting", "query": "good", "prompt": "Once the audio has finished, list each occurrence of the word \"good\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.75, 1.03, "one"], [4.5, 4.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.75,\"end\":1.03},{\"word\":\"two\",\"start\":4.5,\"end\":4.92}]"}
{"utterance_id": "5338-24615-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0000.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Process the full audio offline. Find every instance of a time of day and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.34, "noon"]], "response_mode": "offline", "answer": "[{\"word\":\"noon\",\"start\":1.03,\"end\":1.34}]"}
{"utterance_id": "2803-154328-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 1.02, "john"], [2.41, 2.95, "mary"], [3.98, 4.84, "lady helena"]], "response_mode": "offline", "answer": "[{\"word\":\"john\",\"start\":0.56,\"end\":1.02},{\"word\":\"mary\",\"start\":2.41,\"end\":2.95},{\"word\":\"lady helena\",\"start\":3.98,\"end\":4.84}]"}
{"utterance_id": "2412-153954-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0024.flac", "task_type": "lexical_detection", "query": "presently", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"presently\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.47, "presently"]], "response_mode": "offline", "answer": "[{\"word\":\"presently\",\"start\":0.99,\"end\":1.47}]"}
{"utterance_id": "251-136532-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0016.flac", "task_type": "lexical_counting", "query": "martians", "prompt": "When the utterance is complete, report how the word \"martians\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 3.01, "one"], [4.25, 4.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.29,\"end\":3.01},{\"word\":\"two\",\"start\":4.25,\"end\":4.74}]"}
{"utterance_id": "652-130726-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0006.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen carefully to the entire clip, then report every instance of an organization name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.42, 6.68, "cliff"]], "response_mode": "offline", "answer": "[{\"word\":\"cliff\",\"start\":6.42,\"end\":6.68}]"}
{"utterance_id": "5536-43358-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0017.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.41, 10.22, "philosopher"]], "response_mode": "offline", "answer": "[{\"word\":\"philosopher\",\"start\":9.41,\"end\":10.22}]"}
{"utterance_id": "1272-135031-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.43, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.11,\"end\":1.43}]"}
{"utterance_id": "5694-64025-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0005.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After listening to the whole recording, output every instance of a time of day with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.56, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.0,\"end\":1.56}]"}
{"utterance_id": "6313-66125-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 1.87, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":1.49,\"end\":1.87}]"}
{"utterance_id": "3536-23268-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0007.flac", "task_type": "lexical_detection", "query": "uttered", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"uttered\" with fields word, start, end (seconds). If none, reply [].", "labels": [[8.65, 9.02, "uttered"]], "response_mode": "offline", "answer": "[{\"word\":\"uttered\",\"start\":8.65,\"end\":9.02}]"}
{"utterance_id": "8842-302203-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.37, 1.8, "joan"], [7.72, 9.48, "primavera spring"]], "response_mode": "offline", "answer": "[{\"word\":\"joan\",\"start\":1.37,\"end\":1.8},{\"word\":\"primavera spring\",\"start\":7.72,\"end\":9.48}]"}
{"utterance_id": "5694-64025-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.85, 7.67, "one half"]], "response_mode": "offline", "answer": "[{\"word\":\"one half\",\"start\":6.85,\"end\":7.67}]"}
{"utterance_id": "3853-163249-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0001.flac", "task_type": "lexical_detection", "query": "dream", "prompt": "After listening to the whole recording, output every instance of the word \"dream\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.84, 6.11, "dream"]], "response_mode": "offline", "answer": "[{\"word\":\"dream\",\"start\":5.84,\"end\":6.11}]"}
{"utterance_id": "652-130726-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0005.flac", "task_type": "lexical_detection", "query": "garish", "prompt": "Once the audio has finished, identify every instance of the word \"garish\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.54, 0.97, "garish"]], "response_mode": "offline", "answer": "[{\"word\":\"garish\",\"start\":0.54,\"end\":0.97}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Once the audio has finished, list each occurrence of a profession as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[4.02, 4.92, "one"], [6.47, 6.99, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.02,\"end\":4.92},{\"word\":\"two\",\"start\":6.47,\"end\":6.99}]"}
{"utterance_id": "1673-143396-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.96, 10.25, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":9.96,\"end\":10.25}]"}
{"utterance_id": "5895-34629-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0029.flac", "task_type": "lexical_detection", "query": "visitor", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"visitor\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.16, 2.76, "visitor"]], "response_mode": "offline", "answer": "[{\"word\":\"visitor\",\"start\":2.16,\"end\":2.76}]"}
{"utterance_id": "84-121550-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.09, 4.71, "urania"]], "response_mode": "offline", "answer": "[{\"word\":\"urania\",\"start\":4.09,\"end\":4.71}]"}
{"utterance_id": "1988-148538-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0010.flac", "task_type": "lexical_counting", "query": "freedom", "prompt": "When the utterance is complete, report how the word \"freedom\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.8, "one"], [4.67, 5.11, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.23,\"end\":1.8},{\"word\":\"two\",\"start\":4.67,\"end\":5.11}]"}
{"utterance_id": "6313-76958-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0013.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.78, "horses"], [9.34, 9.8, "animals"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":0.38,\"end\":0.78},{\"word\":\"animals\",\"start\":9.34,\"end\":9.8}]"}
{"utterance_id": "6345-93302-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0029.flac", "task_type": "lexical_detection", "query": "explanation", "prompt": "After listening to the whole recording, output every instance of the word \"explanation\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.31, 9.16, "explanation"]], "response_mode": "offline", "answer": "[{\"word\":\"explanation\",\"start\":8.31,\"end\":9.16}]"}
{"utterance_id": "251-118436-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.74, 6.34, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":5.74,\"end\":6.34}]"}
{"utterance_id": "6295-244435-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0016.flac", "task_type": "lexical_counting", "query": "understand", "prompt": "Listen to the entire clip, then report the count sequence for the word \"understand\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.38, 2.02, "one"], [6.88, 7.42, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.38,\"end\":2.02},{\"word\":\"two\",\"start\":6.88,\"end\":7.42}]"}
{"utterance_id": "8842-304647-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0000.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "You will hear a complete utterance. When it ends, list every instance of a weather term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.84, 5.35, "lightnings"]], "response_mode": "offline", "answer": "[{\"word\":\"lightnings\",\"start\":4.84,\"end\":5.35}]"}
{"utterance_id": "8297-275156-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0001.flac", "task_type": "lexical_detection", "query": "catherine", "prompt": "After the audio ends, extract all occurrences of the word \"catherine\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 4.13, "catherine"]], "response_mode": "offline", "answer": "[{\"word\":\"catherine\",\"start\":3.5,\"end\":4.13}]"}
{"utterance_id": "6319-275224-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0014.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Once the audio has finished, list each occurrence of a plant name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.64, 1.37, "one"], [3.3, 4.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.64,\"end\":1.37},{\"word\":\"two\",\"start\":3.3,\"end\":4.12}]"}
{"utterance_id": "6313-66129-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.19, 4.47, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":4.19,\"end\":4.47}]"}
{"utterance_id": "1673-143396-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.14, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":0.75,\"end\":1.14}]"}
{"utterance_id": "1462-170138-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0022.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[1.19, 1.54, "one"], [3.28, 3.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.19,\"end\":1.54},{\"word\":\"two\",\"start\":3.28,\"end\":3.88}]"}
{"utterance_id": "6319-275224-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0003.flac", "task_type": "lexical_detection", "query": "roots", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"roots\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.57, 11.12, "roots"]], "response_mode": "offline", "answer": "[{\"word\":\"roots\",\"start\":10.57,\"end\":11.12}]"}
{"utterance_id": "6313-76958-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0024.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "After listening to the whole recording, output timed count words for a profession (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.91, 1.26, "one"], [2.38, 2.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.91,\"end\":1.26},{\"word\":\"two\",\"start\":2.38,\"end\":2.83}]"}
{"utterance_id": "1988-147956-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0023.flac", "task_type": "lexical_detection", "query": "questioned", "prompt": "Once the audio has finished, identify every instance of the word \"questioned\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.53, 3.0, "questioned"]], "response_mode": "offline", "answer": "[{\"word\":\"questioned\",\"start\":2.53,\"end\":3.0}]"}
{"utterance_id": "6295-64301-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0011.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.75, "one"], [5.09, 5.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.33,\"end\":0.75},{\"word\":\"two\",\"start\":5.09,\"end\":5.59}]"}
{"utterance_id": "5694-64038-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0010.flac", "task_type": "lexical_detection", "query": "comrade", "prompt": "When the utterance is complete, report all instances of the word \"comrade\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.57, 8.04, "comrade"]], "response_mode": "offline", "answer": "[{\"word\":\"comrade\",\"start\":7.57,\"end\":8.04}]"}
{"utterance_id": "7976-110523-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0017.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.86, 5.34, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":4.86,\"end\":5.34}]"}
{"utterance_id": "8842-302201-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 0.91, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.6,\"end\":0.91}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.48, 8.76, "hand"], [14.42, 14.8, "senses"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":8.48,\"end\":8.76},{\"word\":\"senses\",\"start\":14.42,\"end\":14.8}]"}
{"utterance_id": "3536-23268-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0017.flac", "task_type": "lexical_detection", "query": "evinced", "prompt": "When the utterance is complete, report all instances of the word \"evinced\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.1, 6.51, "evinced"]], "response_mode": "offline", "answer": "[{\"word\":\"evinced\",\"start\":6.1,\"end\":6.51}]"}
{"utterance_id": "84-121123-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0008.flac", "task_type": "lexical_detection", "query": "rave", "prompt": "Process the full audio offline. Find every instance of the word \"rave\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.04, "rave"], [6.42, 6.91, "rave"]], "response_mode": "offline", "answer": "[{\"word\":\"rave\",\"start\":0.66,\"end\":1.04},{\"word\":\"rave\",\"start\":6.42,\"end\":6.91}]"}
{"utterance_id": "2277-149896-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0010.flac", "task_type": "lexical_detection", "query": "drouet", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"drouet\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 2.06, "drouet"]], "response_mode": "offline", "answer": "[{\"word\":\"drouet\",\"start\":1.51,\"end\":2.06}]"}
{"utterance_id": "2035-147961-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0012.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a disease name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.05, 3.49, "croup"]], "response_mode": "offline", "answer": "[{\"word\":\"croup\",\"start\":3.05,\"end\":3.49}]"}
{"utterance_id": "3170-137482-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0013.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.99, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":1.18,\"end\":1.99}]"}
{"utterance_id": "251-137823-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0004.flac", "task_type": "lexical_detection", "query": "chain", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"chain\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.6, 4.96, "chain"]], "response_mode": "offline", "answer": "[{\"word\":\"chain\",\"start\":4.6,\"end\":4.96}]"}
{"utterance_id": "6241-61946-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0001.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.75, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.3,\"end\":2.75}]"}
{"utterance_id": "3752-4943-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0008.flac", "task_type": "lexical_detection", "query": "bars", "prompt": "After the audio ends, extract all occurrences of the word \"bars\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.44, "bars"]], "response_mode": "offline", "answer": "[{\"word\":\"bars\",\"start\":0.88,\"end\":1.44}]"}
{"utterance_id": "5536-43363-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0011.flac", "task_type": "lexical_detection", "query": "carried", "prompt": "When the utterance is complete, report all instances of the word \"carried\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.09, "carried"]], "response_mode": "offline", "answer": "[{\"word\":\"carried\",\"start\":0.75,\"end\":1.09}]"}
{"utterance_id": "6319-64726-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0011.flac", "task_type": "lexical_detection", "query": "faces", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"faces\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.29, "faces"]], "response_mode": "offline", "answer": "[{\"word\":\"faces\",\"start\":0.78,\"end\":1.29}]"}
{"utterance_id": "652-130726-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0034.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.56, "abalone"]], "response_mode": "offline", "answer": "[{\"word\":\"abalone\",\"start\":2.07,\"end\":2.56}]"}
{"utterance_id": "3170-137482-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0026.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.86, 3.25, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":2.86,\"end\":3.25}]"}
{"utterance_id": "5694-64029-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.26, 4.63, "cheatham"]], "response_mode": "offline", "answer": "[{\"word\":\"cheatham\",\"start\":4.26,\"end\":4.63}]"}
{"utterance_id": "422-122949-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0034.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.36, 8.7, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":8.36,\"end\":8.7}]"}
{"utterance_id": "5895-34622-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.77, "poet"]], "response_mode": "offline", "answer": "[{\"word\":\"poet\",\"start\":1.31,\"end\":1.77}]"}
{"utterance_id": "1462-170142-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.6, 0.97, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.6,\"end\":0.97}]"}
{"utterance_id": "3081-166546-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0037.flac", "task_type": "lexical_detection", "query": "shot", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"shot\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.09, 2.34, "shot"]], "response_mode": "offline", "answer": "[{\"word\":\"shot\",\"start\":2.09,\"end\":2.34}]"}
{"utterance_id": "5895-34629-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0006.flac", "task_type": "lexical_detection", "query": "yard", "prompt": "Once the audio has finished, identify every instance of the word \"yard\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.85, 5.29, "yard"]], "response_mode": "offline", "answer": "[{\"word\":\"yard\",\"start\":4.85,\"end\":5.29}]"}
{"utterance_id": "84-121123-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.43, "morrel"]], "response_mode": "offline", "answer": "[{\"word\":\"morrel\",\"start\":2.01,\"end\":2.43}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "When the utterance is complete, report all instances of a weather term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.16, "storm"], [5.09, 5.48, "wind"], [7.58, 7.91, "rain"]], "response_mode": "offline", "answer": "[{\"word\":\"storm\",\"start\":1.73,\"end\":2.16},{\"word\":\"wind\",\"start\":5.09,\"end\":5.48},{\"word\":\"rain\",\"start\":7.58,\"end\":7.91}]"}
{"utterance_id": "2035-147961-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0001.flac", "task_type": "lexical_detection", "query": "peter", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"peter\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.81, "peter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":0.47,\"end\":0.81}]"}
{"utterance_id": "5694-64025-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.71, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":2.33,\"end\":2.71}]"}
{"utterance_id": "3536-8226-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0014.flac", "task_type": "lexical_detection", "query": "child", "prompt": "After listening to the whole recording, output every instance of the word \"child\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.13, 1.59, "child"]], "response_mode": "offline", "answer": "[{\"word\":\"child\",\"start\":1.13,\"end\":1.59}]"}
{"utterance_id": "84-121123-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.17, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":0.84,\"end\":1.17}]"}
{"utterance_id": "2078-142845-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0031.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.62, 11.99, "bread"], [15.56, 16.2, "toast"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":11.62,\"end\":11.99},{\"word\":\"toast\",\"start\":15.56,\"end\":16.2}]"}
{"utterance_id": "8842-302196-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0009.flac", "task_type": "lexical_detection", "query": "itself", "prompt": "After listening to the whole recording, output every instance of the word \"itself\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.13, 6.74, "itself"]], "response_mode": "offline", "answer": "[{\"word\":\"itself\",\"start\":6.13,\"end\":6.74}]"}
{"utterance_id": "1673-143396-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.91, 6.32, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":5.91,\"end\":6.32}]"}
{"utterance_id": "2428-83705-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0034.flac", "task_type": "lexical_detection", "query": "macpherson", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"macpherson\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.01, 2.73, "macpherson"]], "response_mode": "offline", "answer": "[{\"word\":\"macpherson\",\"start\":2.01,\"end\":2.73}]"}
{"utterance_id": "7976-105575-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0029.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.76, 7.28, "hogs"]], "response_mode": "offline", "answer": "[{\"word\":\"hogs\",\"start\":6.76,\"end\":7.28}]"}
{"utterance_id": "5536-43363-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0009.flac", "task_type": "lexical_detection", "query": "occult", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"occult\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.52, 3.98, "occult"]], "response_mode": "offline", "answer": "[{\"word\":\"occult\",\"start\":3.52,\"end\":3.98}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "lexical_counting", "query": "stomach", "prompt": "When the utterance is complete, report how the word \"stomach\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.95, "one"], [4.96, 5.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.41,\"end\":1.95},{\"word\":\"two\",\"start\":4.96,\"end\":5.53}]"}
{"utterance_id": "2902-9008-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0011.flac", "task_type": "lexical_detection", "query": "fruit", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"fruit\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.13, "fruit"]], "response_mode": "offline", "answer": "[{\"word\":\"fruit\",\"start\":0.73,\"end\":1.13}]"}
{"utterance_id": "2902-9008-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0016.flac", "task_type": "lexical_detection", "query": "enter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"enter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 2.87, "enter"]], "response_mode": "offline", "answer": "[{\"word\":\"enter\",\"start\":2.54,\"end\":2.87}]"}
{"utterance_id": "6295-244435-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0033.flac", "task_type": "lexical_detection", "query": "sturdy", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sturdy\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.19, 5.62, "sturdy"]], "response_mode": "offline", "answer": "[{\"word\":\"sturdy\",\"start\":5.19,\"end\":5.62}]"}
{"utterance_id": "1919-142785-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0040.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.16, "leaves"], [4.61, 5.0, "herb"]], "response_mode": "offline", "answer": "[{\"word\":\"leaves\",\"start\":2.81,\"end\":3.16},{\"word\":\"herb\",\"start\":4.61,\"end\":5.0}]"}
{"utterance_id": "1993-147149-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0029.flac", "task_type": "lexical_detection", "query": "scolded", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"scolded\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.82, 2.31, "scolded"]], "response_mode": "offline", "answer": "[{\"word\":\"scolded\",\"start\":1.82,\"end\":2.31}]"}
{"utterance_id": "5694-64038-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0004.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.75, 7.25, "lice"]], "response_mode": "offline", "answer": "[{\"word\":\"lice\",\"start\":6.75,\"end\":7.25}]"}
{"utterance_id": "652-130726-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0015.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.5, 7.91, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":7.5,\"end\":7.91}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the full audio. After it finishes, find every instance of a country name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.61, 13.0, "italy"], [18.75, 19.02, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"italy\",\"start\":12.61,\"end\":13.0},{\"word\":\"europe\",\"start\":18.75,\"end\":19.02}]"}
{"utterance_id": "1462-170138-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.38, 10.9, "donkey"]], "response_mode": "offline", "answer": "[{\"word\":\"donkey\",\"start\":10.38,\"end\":10.9}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.25, 7.65, "priest"], [10.61, 11.07, "praefect"]], "response_mode": "offline", "answer": "[{\"word\":\"priest\",\"start\":7.25,\"end\":7.65},{\"word\":\"praefect\",\"start\":10.61,\"end\":11.07}]"}
{"utterance_id": "6295-244435-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0007.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.46, "garrison"]], "response_mode": "offline", "answer": "[{\"word\":\"garrison\",\"start\":2.93,\"end\":3.46}]"}
{"utterance_id": "6295-64301-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0005.flac", "task_type": "lexical_counting", "query": "blessed", "prompt": "Process the full audio offline. For every occurrence of the word \"blessed\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.86, "one"], [3.8, 4.17, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.4,\"end\":0.86},{\"word\":\"two\",\"start\":3.8,\"end\":4.17}]"}
{"utterance_id": "1988-148538-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0008.flac", "task_type": "lexical_detection", "query": "impatient", "prompt": "After listening to the whole recording, output every instance of the word \"impatient\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.12, 3.71, "impatient"]], "response_mode": "offline", "answer": "[{\"word\":\"impatient\",\"start\":3.12,\"end\":3.71}]"}
{"utterance_id": "7976-110124-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 2.14, "merchant"]], "response_mode": "offline", "answer": "[{\"word\":\"merchant\",\"start\":1.66,\"end\":2.14}]"}
{"utterance_id": "84-121123-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.53, "pores"]], "response_mode": "offline", "answer": "[{\"word\":\"pores\",\"start\":1.89,\"end\":2.53}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "semantic_counting", "query": "food", "prompt": "When the utterance is complete, report how a food name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.44, "one"], [5.37, 5.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.79,\"end\":2.44},{\"word\":\"two\",\"start\":5.37,\"end\":5.67}]"}
{"utterance_id": "3752-4944-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.12, 6.58, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cat\",\"start\":6.12,\"end\":6.58}]"}
{"utterance_id": "3170-137482-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0023.flac", "task_type": "lexical_counting", "query": "ordered", "prompt": "Process the full audio offline. For every occurrence of the word \"ordered\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.45, 2.84, "one"], [6.94, 7.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.45,\"end\":2.84},{\"word\":\"two\",\"start\":6.94,\"end\":7.3}]"}
{"utterance_id": "3081-166546-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0026.flac", "task_type": "lexical_detection", "query": "forgot", "prompt": "After the audio ends, extract all occurrences of the word \"forgot\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.89, 4.4, "forgot"]], "response_mode": "offline", "answer": "[{\"word\":\"forgot\",\"start\":3.89,\"end\":4.4}]"}
{"utterance_id": "2035-147961-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0039.flac", "task_type": "lexical_detection", "query": "log", "prompt": "After the audio ends, extract all occurrences of the word \"log\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 2.88, "log"]], "response_mode": "offline", "answer": "[{\"word\":\"log\",\"start\":2.62,\"end\":2.88}]"}
{"utterance_id": "2803-154320-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0000.flac", "task_type": "lexical_detection", "query": "halley", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"halley\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.83, 2.39, "halley"]], "response_mode": "offline", "answer": "[{\"word\":\"halley\",\"start\":1.83,\"end\":2.39}]"}
{"utterance_id": "777-126732-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0056.flac", "task_type": "lexical_detection", "query": "weight", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"weight\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.4, 4.67, "weight"]], "response_mode": "offline", "answer": "[{\"word\":\"weight\",\"start\":4.4,\"end\":4.67}]"}
{"utterance_id": "5895-34622-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0005.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.55, 0.82, "green"], [6.23, 6.56, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":0.55,\"end\":0.82},{\"word\":\"green\",\"start\":6.23,\"end\":6.56}]"}
{"utterance_id": "84-121123-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0024.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.13, "doctor"]], "response_mode": "offline", "answer": "[{\"word\":\"doctor\",\"start\":0.74,\"end\":1.13}]"}
{"utterance_id": "6313-66129-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.93, 1.3, "stacy"], [6.75, 7.33, "tad butler"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":0.93,\"end\":1.3},{\"word\":\"tad butler\",\"start\":6.75,\"end\":7.33}]"}
{"utterance_id": "5694-64029-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0018.flac", "task_type": "lexical_detection", "query": "forward", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"forward\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.78, "forward"]], "response_mode": "offline", "answer": "[{\"word\":\"forward\",\"start\":1.39,\"end\":1.78}]"}
{"utterance_id": "3576-138058-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0032.flac", "task_type": "lexical_detection", "query": "worth", "prompt": "When the utterance is complete, report all instances of the word \"worth\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.74, 4.12, "worth"]], "response_mode": "offline", "answer": "[{\"word\":\"worth\",\"start\":3.74,\"end\":4.12}]"}
{"utterance_id": "1673-143397-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.25, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":1.86,\"end\":2.25}]"}
{"utterance_id": "2277-149896-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0017.flac", "task_type": "lexical_detection", "query": "lunch", "prompt": "After the audio ends, extract all occurrences of the word \"lunch\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.52, "lunch"]], "response_mode": "offline", "answer": "[{\"word\":\"lunch\",\"start\":2.16,\"end\":2.52}]"}
{"utterance_id": "6295-244435-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0035.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.13, "bill skelly"]], "response_mode": "offline", "answer": "[{\"word\":\"bill skelly\",\"start\":0.51,\"end\":1.13}]"}
{"utterance_id": "3170-137482-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0031.flac", "task_type": "lexical_detection", "query": "unwilling", "prompt": "When the utterance is complete, report all instances of the word \"unwilling\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.81, "unwilling"]], "response_mode": "offline", "answer": "[{\"word\":\"unwilling\",\"start\":0.36,\"end\":0.81}]"}
{"utterance_id": "2078-142845-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0047.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.31, 6.9, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":6.31,\"end\":6.9}]"}
{"utterance_id": "777-126732-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0026.flac", "task_type": "lexical_detection", "query": "himself", "prompt": "After listening to the whole recording, output every instance of the word \"himself\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.73, "himself"]], "response_mode": "offline", "answer": "[{\"word\":\"himself\",\"start\":1.16,\"end\":1.73}]"}
{"utterance_id": "2803-154328-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.87, "animal"]], "response_mode": "offline", "answer": "[{\"word\":\"animal\",\"start\":0.28,\"end\":0.87}]"}
{"utterance_id": "7850-111771-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0005.flac", "task_type": "lexical_detection", "query": "forces", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"forces\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.45, 3.14, "forces"]], "response_mode": "offline", "answer": "[{\"word\":\"forces\",\"start\":2.45,\"end\":3.14}]"}
{"utterance_id": "174-50561-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0010.flac", "task_type": "lexical_counting", "query": "played", "prompt": "Once the audio has finished, list each occurrence of the word \"played\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[3.35, 3.67, "one"], [12.62, 12.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.35,\"end\":3.67},{\"word\":\"two\",\"start\":12.62,\"end\":12.83}]"}
{"utterance_id": "1919-142785-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0038.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.72, 3.12, "eggs"]], "response_mode": "offline", "answer": "[{\"word\":\"eggs\",\"start\":2.72,\"end\":3.12}]"}
{"utterance_id": "6313-66129-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0008.flac", "task_type": "lexical_detection", "query": "honey", "prompt": "After the audio ends, extract all occurrences of the word \"honey\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.41, 4.73, "honey"]], "response_mode": "offline", "answer": "[{\"word\":\"honey\",\"start\":4.41,\"end\":4.73}]"}
{"utterance_id": "777-126732-0065", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0065.flac", "task_type": "lexical_detection", "query": "stevie", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"stevie\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.68, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":2.33,\"end\":2.68}]"}
{"utterance_id": "3752-4943-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0022.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.15, 2.65, "giant"]], "response_mode": "offline", "answer": "[{\"word\":\"giant\",\"start\":2.15,\"end\":2.65}]"}
{"utterance_id": "8297-275154-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0020.flac", "task_type": "lexical_detection", "query": "hand", "prompt": "After listening to the whole recording, output every instance of the word \"hand\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.9, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.4,\"end\":2.9}]"}
{"utterance_id": "84-121550-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0007.flac", "task_type": "lexical_counting", "query": "default", "prompt": "Process the full audio offline. For every occurrence of the word \"default\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.31, "one"], [3.9, 4.5, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.77,\"end\":1.31},{\"word\":\"two\",\"start\":3.9,\"end\":4.5}]"}
{"utterance_id": "3576-138058-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0030.flac", "task_type": "lexical_detection", "query": "cities", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cities\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.28, 6.04, "cities"]], "response_mode": "offline", "answer": "[{\"word\":\"cities\",\"start\":5.28,\"end\":6.04}]"}
{"utterance_id": "1673-143396-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0020.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.42, 8.71, "one"], [10.74, 11.0, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.42,\"end\":8.71},{\"word\":\"four\",\"start\":10.74,\"end\":11.0}]"}
{"utterance_id": "3536-23268-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0000.flac", "task_type": "lexical_detection", "query": "discouraged", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"discouraged\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.65, "discouraged"]], "response_mode": "offline", "answer": "[{\"word\":\"discouraged\",\"start\":2.0,\"end\":2.65}]"}
{"utterance_id": "2035-147960-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.07, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.9,\"end\":1.07}]"}
{"utterance_id": "6313-66125-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.72, "walt"], [3.89, 4.7, "mister thomas"]], "response_mode": "offline", "answer": "[{\"word\":\"walt\",\"start\":1.33,\"end\":1.72},{\"word\":\"mister thomas\",\"start\":3.89,\"end\":4.7}]"}
{"utterance_id": "2035-147961-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0014.flac", "task_type": "lexical_detection", "query": "hollow", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"hollow\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.94, 2.27, "hollow"]], "response_mode": "offline", "answer": "[{\"word\":\"hollow\",\"start\":1.94,\"end\":2.27}]"}
{"utterance_id": "2412-153954-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0008.flac", "task_type": "lexical_detection", "query": "handsome", "prompt": "Once the audio has finished, identify every instance of the word \"handsome\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.09, 1.48, "handsome"]], "response_mode": "offline", "answer": "[{\"word\":\"handsome\",\"start\":1.09,\"end\":1.48}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.96, "one"], [4.69, 5.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.45,\"end\":0.96},{\"word\":\"two\",\"start\":4.69,\"end\":5.37}]"}
{"utterance_id": "2412-153947-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0007.flac", "task_type": "lexical_detection", "query": "sincere", "prompt": "When the utterance is complete, report all instances of the word \"sincere\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.1, "sincere"]], "response_mode": "offline", "answer": "[{\"word\":\"sincere\",\"start\":2.65,\"end\":3.1}]"}
{"utterance_id": "6241-66616-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0001.flac", "task_type": "lexical_detection", "query": "wabigoon's", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wabigoon's\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.24, 4.84, "wabigoon's"]], "response_mode": "offline", "answer": "[{\"word\":\"wabigoon's\",\"start\":4.24,\"end\":4.84}]"}
{"utterance_id": "7976-110124-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.27, 6.7, "merchant's"]], "response_mode": "offline", "answer": "[{\"word\":\"merchant's\",\"start\":6.27,\"end\":6.7}]"}
{"utterance_id": "6345-93306-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0000.flac", "task_type": "lexical_detection", "query": "admiration", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"admiration\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.8, 15.4, "admiration"]], "response_mode": "offline", "answer": "[{\"word\":\"admiration\",\"start\":14.8,\"end\":15.4}]"}
{"utterance_id": "1462-170142-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.38, 6.83, "hilda's"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda's\",\"start\":6.38,\"end\":6.83}]"}
{"utterance_id": "2086-149220-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 4.88, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":4.47,\"end\":4.88}]"}
{"utterance_id": "6295-244435-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0011.flac", "task_type": "lexical_detection", "query": "single", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"single\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.3, "single"]], "response_mode": "offline", "answer": "[{\"word\":\"single\",\"start\":0.91,\"end\":1.3}]"}
{"utterance_id": "5694-64029-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0005.flac", "task_type": "lexical_detection", "query": "resting", "prompt": "When the utterance is complete, report all instances of the word \"resting\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.3, 3.79, "resting"]], "response_mode": "offline", "answer": "[{\"word\":\"resting\",\"start\":3.3,\"end\":3.79}]"}
{"utterance_id": "1993-147149-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0009.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Once the audio has finished, list each occurrence of a family relation as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.37, 1.72, "one"], [9.58, 9.93, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.37,\"end\":1.72},{\"word\":\"two\",\"start\":9.58,\"end\":9.93}]"}
{"utterance_id": "174-84280-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.43, 3.96, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":3.43,\"end\":3.96}]"}
{"utterance_id": "3081-166546-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0040.flac", "task_type": "lexical_detection", "query": "bolt", "prompt": "When the utterance is complete, report all instances of the word \"bolt\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.46, 3.78, "bolt"]], "response_mode": "offline", "answer": "[{\"word\":\"bolt\",\"start\":3.46,\"end\":3.78}]"}
{"utterance_id": "2086-149220-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0014.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.33, "crumbs"]], "response_mode": "offline", "answer": "[{\"word\":\"crumbs\",\"start\":3.83,\"end\":4.33}]"}
{"utterance_id": "251-136532-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0001.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.76, 6.45, "syrtis"]], "response_mode": "offline", "answer": "[{\"word\":\"syrtis\",\"start\":5.76,\"end\":6.45}]"}
{"utterance_id": "1462-170138-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.59, "sir harry"]], "response_mode": "offline", "answer": "[{\"word\":\"sir harry\",\"start\":0.82,\"end\":1.59}]"}
{"utterance_id": "2086-149220-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.52, 9.0, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":8.52,\"end\":9.0}]"}
{"utterance_id": "1272-128104-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.78, "fred"]], "response_mode": "offline", "answer": "[{\"word\":\"fred\",\"start\":2.37,\"end\":2.78}]"}
{"utterance_id": "5338-24615-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0008.flac", "task_type": "lexical_detection", "query": "built", "prompt": "When the utterance is complete, report all instances of the word \"built\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.12, "built"]], "response_mode": "offline", "answer": "[{\"word\":\"built\",\"start\":0.81,\"end\":1.12}]"}
{"utterance_id": "8842-302201-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0013.flac", "task_type": "lexical_detection", "query": "desired", "prompt": "Once the audio has finished, identify every instance of the word \"desired\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.74, 5.36, "desired"]], "response_mode": "offline", "answer": "[{\"word\":\"desired\",\"start\":4.74,\"end\":5.36}]"}
{"utterance_id": "251-136532-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0003.flac", "task_type": "lexical_detection", "query": "snapping", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"snapping\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.61, 8.08, "snapping"]], "response_mode": "offline", "answer": "[{\"word\":\"snapping\",\"start\":7.61,\"end\":8.08}]"}
{"utterance_id": "6313-66129-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.11, 4.92, "stacy brown"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy brown\",\"start\":4.11,\"end\":4.92}]"}
{"utterance_id": "251-136532-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 1.37, "tony lattimer"], [5.24, 5.79, "gloria"], [7.18, 7.72, "sid"]], "response_mode": "offline", "answer": "[{\"word\":\"tony lattimer\",\"start\":0.36,\"end\":1.37},{\"word\":\"gloria\",\"start\":5.24,\"end\":5.79},{\"word\":\"sid\",\"start\":7.18,\"end\":7.72}]"}
{"utterance_id": "1673-143396-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0008.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.47, 6.79, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":6.47,\"end\":6.79}]"}
{"utterance_id": "1272-141231-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0017.flac", "task_type": "lexical_detection", "query": "aching", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"aching\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.72, "aching"]], "response_mode": "offline", "answer": "[{\"word\":\"aching\",\"start\":2.33,\"end\":2.72}]"}
{"utterance_id": "6345-93302-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0010.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen carefully to the entire clip, then report every instance of a weather term as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.31, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":0.91,\"end\":1.31}]"}
{"utterance_id": "2086-149220-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0012.flac", "task_type": "lexical_detection", "query": "chicken", "prompt": "When the utterance is complete, report all instances of the word \"chicken\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.84, "chicken"]], "response_mode": "offline", "answer": "[{\"word\":\"chicken\",\"start\":0.51,\"end\":0.84}]"}
{"utterance_id": "8297-275154-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0006.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "When the utterance is complete, report how a person's name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.88, "one"], [7.61, 8.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.38,\"end\":0.88},{\"word\":\"two\",\"start\":7.61,\"end\":8.13}]"}
{"utterance_id": "3853-163249-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0054.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.5, 4.85, "blue"]], "response_mode": "offline", "answer": "[{\"word\":\"blue\",\"start\":4.5,\"end\":4.85}]"}
{"utterance_id": "2277-149897-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0025.flac", "task_type": "lexical_detection", "query": "married", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"married\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.19, 2.59, "married"]], "response_mode": "offline", "answer": "[{\"word\":\"married\",\"start\":2.19,\"end\":2.59}]"}
{"utterance_id": "2412-153948-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0012.flac", "task_type": "lexical_detection", "query": "farther", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"farther\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.25, "farther"]], "response_mode": "offline", "answer": "[{\"word\":\"farther\",\"start\":1.92,\"end\":2.25}]"}
{"utterance_id": "652-130726-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0003.flac", "task_type": "lexical_detection", "query": "places", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"places\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.95, 2.47, "places"]], "response_mode": "offline", "answer": "[{\"word\":\"places\",\"start\":1.95,\"end\":2.47}]"}
{"utterance_id": "6241-61943-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0004.flac", "task_type": "lexical_detection", "query": "walls", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"walls\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.71, 3.17, "walls"]], "response_mode": "offline", "answer": "[{\"word\":\"walls\",\"start\":2.71,\"end\":3.17}]"}
{"utterance_id": "1272-141231-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0010.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.08, "one"], [6.79, 7.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.83,\"end\":4.08},{\"word\":\"two\",\"start\":6.79,\"end\":7.47}]"}
{"utterance_id": "2086-149220-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.14, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.82,\"end\":2.14}]"}
{"utterance_id": "1673-143397-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0005.flac", "task_type": "lexical_detection", "query": "obstacle", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"obstacle\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.22, 4.71, "obstacle"], [7.92, 8.38, "obstacle"]], "response_mode": "offline", "answer": "[{\"word\":\"obstacle\",\"start\":4.22,\"end\":4.71},{\"word\":\"obstacle\",\"start\":7.92,\"end\":8.38}]"}
{"utterance_id": "6241-66616-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 0.98, "thousand"], [2.1, 2.54, "thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"thousand\",\"start\":0.48,\"end\":0.98},{\"word\":\"thousand\",\"start\":2.1,\"end\":2.54}]"}
{"utterance_id": "2035-147960-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 1.81, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.7,\"end\":1.81}]"}
{"utterance_id": "84-121550-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0020.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Once the audio has finished, identify every instance of a weather term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.05, 3.48, "thunder"]], "response_mode": "offline", "answer": "[{\"word\":\"thunder\",\"start\":3.05,\"end\":3.48}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "lexical_detection", "query": "earlier", "prompt": "After listening to the whole recording, output every instance of the word \"earlier\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.8, 3.23, "earlier"]], "response_mode": "offline", "answer": "[{\"word\":\"earlier\",\"start\":2.8,\"end\":3.23}]"}
{"utterance_id": "3536-23268-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.9, 9.48, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss milner\",\"start\":8.9,\"end\":9.48}]"}
{"utterance_id": "3081-166546-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0044.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.41, "detective"]], "response_mode": "offline", "answer": "[{\"word\":\"detective\",\"start\":1.75,\"end\":2.41}]"}
{"utterance_id": "2902-9008-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.88, 10.49, "pelagia"]], "response_mode": "offline", "answer": "[{\"word\":\"pelagia\",\"start\":9.88,\"end\":10.49}]"}
{"utterance_id": "1272-128104-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0007.flac", "task_type": "lexical_detection", "query": "painting", "prompt": "After listening to the whole recording, output every instance of the word \"painting\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.99, "painting"]], "response_mode": "offline", "answer": "[{\"word\":\"painting\",\"start\":0.5,\"end\":0.99}]"}
{"utterance_id": "6295-244435-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0029.flac", "task_type": "lexical_detection", "query": "native", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"native\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.79, 3.15, "native"]], "response_mode": "offline", "answer": "[{\"word\":\"native\",\"start\":2.79,\"end\":3.15}]"}
{"utterance_id": "3576-138058-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0006.flac", "task_type": "lexical_detection", "query": "don", "prompt": "After the audio ends, extract all occurrences of the word \"don\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.1, "don"]], "response_mode": "offline", "answer": "[{\"word\":\"don\",\"start\":2.81,\"end\":3.1}]"}
{"utterance_id": "7850-111771-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0004.flac", "task_type": "lexical_detection", "query": "mississippi", "prompt": "After listening to the whole recording, output every instance of the word \"mississippi\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.05, 5.6, "mississippi"]], "response_mode": "offline", "answer": "[{\"word\":\"mississippi\",\"start\":5.05,\"end\":5.6}]"}
{"utterance_id": "777-126732-0065", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0065.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.33, 2.68, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":2.33,\"end\":2.68}]"}
{"utterance_id": "6241-66616-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0011.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.12, 7.78, "port arthur"]], "response_mode": "offline", "answer": "[{\"word\":\"port arthur\",\"start\":7.12,\"end\":7.78}]"}
{"utterance_id": "2902-9006-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0004.flac", "task_type": "lexical_detection", "query": "emperors", "prompt": "When the utterance is complete, report all instances of the word \"emperors\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.4, "emperors"]], "response_mode": "offline", "answer": "[{\"word\":\"emperors\",\"start\":0.86,\"end\":1.4}]"}
{"utterance_id": "1919-142785-0063", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0063.flac", "task_type": "lexical_detection", "query": "sage", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sage\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.74, 2.34, "sage"]], "response_mode": "offline", "answer": "[{\"word\":\"sage\",\"start\":1.74,\"end\":2.34}]"}
{"utterance_id": "3853-163249-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0008.flac", "task_type": "lexical_detection", "query": "sometimes", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sometimes\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.45, "sometimes"]], "response_mode": "offline", "answer": "[{\"word\":\"sometimes\",\"start\":3.86,\"end\":4.45}]"}
{"utterance_id": "6345-93306-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0007.flac", "task_type": "lexical_detection", "query": "there's", "prompt": "When the utterance is complete, report all instances of the word \"there's\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 3.63, "there's"]], "response_mode": "offline", "answer": "[{\"word\":\"there's\",\"start\":3.5,\"end\":3.63}]"}
{"utterance_id": "1988-24833-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0027.flac", "task_type": "lexical_detection", "query": "check", "prompt": "After listening to the whole recording, output every instance of the word \"check\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.23, "check"]], "response_mode": "offline", "answer": "[{\"word\":\"check\",\"start\":0.97,\"end\":1.23}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "lexical_detection", "query": "ministers", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ministers\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.11, 4.61, "ministers"]], "response_mode": "offline", "answer": "[{\"word\":\"ministers\",\"start\":4.11,\"end\":4.61}]"}
{"utterance_id": "2277-149874-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.25, 3.52, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":3.25,\"end\":3.52}]"}
{"utterance_id": "5338-284437-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0031.flac", "task_type": "lexical_detection", "query": "outside", "prompt": "After listening to the whole recording, output every instance of the word \"outside\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.2, 3.81, "outside"]], "response_mode": "offline", "answer": "[{\"word\":\"outside\",\"start\":3.2,\"end\":3.81}]"}
{"utterance_id": "6241-61946-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 2.98, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.55,\"end\":2.98}]"}
{"utterance_id": "5895-34622-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.32, 0.73, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":0.32,\"end\":0.73}]"}
{"utterance_id": "8297-275154-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0019.flac", "task_type": "lexical_detection", "query": "set", "prompt": "After listening to the whole recording, output every instance of the word \"set\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.41, 2.82, "set"]], "response_mode": "offline", "answer": "[{\"word\":\"set\",\"start\":2.41,\"end\":2.82}]"}
{"utterance_id": "6313-76958-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0026.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.13, 3.55, "stacy"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":3.13,\"end\":3.55}]"}
{"utterance_id": "2803-154328-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.08, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":0.52,\"end\":1.08}]"}
{"utterance_id": "8842-304647-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0000.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.14, 3.49, "limb"]], "response_mode": "offline", "answer": "[{\"word\":\"limb\",\"start\":3.14,\"end\":3.49}]"}
{"utterance_id": "7850-281318-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 2.09, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":1.66,\"end\":2.09}]"}
{"utterance_id": "8297-275155-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0020.flac", "task_type": "lexical_detection", "query": "answer", "prompt": "After listening to the whole recording, output every instance of the word \"answer\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.01, "answer"]], "response_mode": "offline", "answer": "[{\"word\":\"answer\",\"start\":1.62,\"end\":2.01}]"}
{"utterance_id": "6241-66616-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.22, 3.49, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":3.22,\"end\":3.49}]"}
{"utterance_id": "2803-154328-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.29, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":0.91,\"end\":1.29}]"}
{"utterance_id": "6313-66125-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0025.flac", "task_type": "lexical_counting", "query": "haul", "prompt": "Listen to the entire clip, then report the count sequence for the word \"haul\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.06, "one"], [4.53, 4.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.73,\"end\":1.06},{\"word\":\"two\",\"start\":4.53,\"end\":4.79}]"}
{"utterance_id": "2428-83705-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0033.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.49, 2.08, "one"], [4.13, 4.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.49,\"end\":2.08},{\"word\":\"two\",\"start\":4.13,\"end\":4.97}]"}
{"utterance_id": "251-137823-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.22, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":0.83,\"end\":1.22}]"}
{"utterance_id": "1919-142785-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0053.flac", "task_type": "lexical_detection", "query": "quenelles", "prompt": "Process the full audio offline. Find every instance of the word \"quenelles\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.82, 8.34, "quenelles"]], "response_mode": "offline", "answer": "[{\"word\":\"quenelles\",\"start\":7.82,\"end\":8.34}]"}
{"utterance_id": "3576-138058-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0025.flac", "task_type": "lexical_detection", "query": "cruel", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cruel\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.23, 1.13, "cruel"]], "response_mode": "offline", "answer": "[{\"word\":\"cruel\",\"start\":0.23,\"end\":1.13}]"}
{"utterance_id": "251-137823-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0016.flac", "task_type": "lexical_detection", "query": "bitterly", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bitterly\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.71, 6.25, "bitterly"]], "response_mode": "offline", "answer": "[{\"word\":\"bitterly\",\"start\":5.71,\"end\":6.25}]"}
{"utterance_id": "1462-170138-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0000.flac", "task_type": "lexical_detection", "query": "study", "prompt": "After the audio ends, extract all occurrences of the word \"study\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.15, 10.57, "study"]], "response_mode": "offline", "answer": "[{\"word\":\"study\",\"start\":10.15,\"end\":10.57}]"}
{"utterance_id": "5694-64029-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0001.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.22, 2.88, "nashville"]], "response_mode": "offline", "answer": "[{\"word\":\"nashville\",\"start\":2.22,\"end\":2.88}]"}
{"utterance_id": "1993-147964-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0007.flac", "task_type": "lexical_detection", "query": "liked", "prompt": "After the audio ends, extract all occurrences of the word \"liked\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.74, 6.07, "liked"]], "response_mode": "offline", "answer": "[{\"word\":\"liked\",\"start\":5.74,\"end\":6.07}]"}
{"utterance_id": "3752-4943-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0029.flac", "task_type": "lexical_detection", "query": "flogging", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"flogging\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.08, 2.54, "flogging"]], "response_mode": "offline", "answer": "[{\"word\":\"flogging\",\"start\":2.08,\"end\":2.54}]"}
{"utterance_id": "1919-142785-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0027.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.47, 1.76, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.47,\"end\":1.76}]"}
{"utterance_id": "5338-284437-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.4, 2.92, "trot"], [4.25, 4.99, "button bright's"], [8.6, 9.14, "cap'n bill"]], "response_mode": "offline", "answer": "[{\"word\":\"trot\",\"start\":2.4,\"end\":2.92},{\"word\":\"button bright's\",\"start\":4.25,\"end\":4.99},{\"word\":\"cap'n bill\",\"start\":8.6,\"end\":9.14}]"}
{"utterance_id": "8297-275156-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0004.flac", "task_type": "lexical_detection", "query": "true", "prompt": "When the utterance is complete, report all instances of the word \"true\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 2.91, "true"]], "response_mode": "offline", "answer": "[{\"word\":\"true\",\"start\":2.32,\"end\":2.91}]"}
{"utterance_id": "2428-83705-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0042.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.4, 7.77, "board"]], "response_mode": "offline", "answer": "[{\"word\":\"board\",\"start\":7.4,\"end\":7.77}]"}
{"utterance_id": "5694-64038-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0021.flac", "task_type": "lexical_detection", "query": "jim", "prompt": "Once the audio has finished, identify every instance of the word \"jim\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.58, 1.88, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":1.58,\"end\":1.88}]"}
{"utterance_id": "1919-142785-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0032.flac", "task_type": "lexical_detection", "query": "seasoning", "prompt": "When the utterance is complete, report all instances of the word \"seasoning\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 3.14, "seasoning"]], "response_mode": "offline", "answer": "[{\"word\":\"seasoning\",\"start\":2.66,\"end\":3.14}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "lexical_counting", "query": "himself", "prompt": "Process the full audio offline. For every occurrence of the word \"himself\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.55, 5.05, "one"], [18.04, 18.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.55,\"end\":5.05},{\"word\":\"two\",\"start\":18.04,\"end\":18.68}]"}
{"utterance_id": "6313-66129-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0032.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.74, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":1.2,\"end\":1.74}]"}
{"utterance_id": "6345-93302-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.01, 3.34, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":3.01,\"end\":3.34}]"}
{"utterance_id": "8842-304647-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.4, 0.93, "flesh"], [11.67, 12.04, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":0.4,\"end\":0.93},{\"word\":\"hearts\",\"start\":11.67,\"end\":12.04}]"}
{"utterance_id": "2086-149214-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.52, 10.08, "butterfly"]], "response_mode": "offline", "answer": "[{\"word\":\"butterfly\",\"start\":9.52,\"end\":10.08}]"}
{"utterance_id": "2902-9006-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0016.flac", "task_type": "lexical_detection", "query": "africa", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"africa\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.57, 8.18, "africa"]], "response_mode": "offline", "answer": "[{\"word\":\"africa\",\"start\":7.57,\"end\":8.18}]"}
{"utterance_id": "2035-147961-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.22, 0.56, "pavel"]], "response_mode": "offline", "answer": "[{\"word\":\"pavel\",\"start\":0.22,\"end\":0.56}]"}
{"utterance_id": "6319-275224-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0005.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 2.04, "rose tree"]], "response_mode": "offline", "answer": "[{\"word\":\"rose tree\",\"start\":1.32,\"end\":2.04}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.95, "lips"], [6.23, 6.65, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":4.45,\"end\":4.95},{\"word\":\"face\",\"start\":6.23,\"end\":6.65}]"}
{"utterance_id": "2078-142845-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0004.flac", "task_type": "lexical_counting", "query": "pan", "prompt": "Process the full audio offline. For every occurrence of the word \"pan\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.03, 5.64, "one"], [13.38, 14.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.03,\"end\":5.64},{\"word\":\"two\",\"start\":13.38,\"end\":14.0}]"}
{"utterance_id": "2428-83705-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0019.flac", "task_type": "lexical_detection", "query": "hers", "prompt": "Once the audio has finished, identify every instance of the word \"hers\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.37, 4.89, "hers"]], "response_mode": "offline", "answer": "[{\"word\":\"hers\",\"start\":4.37,\"end\":4.89}]"}
{"utterance_id": "2277-149896-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0020.flac", "task_type": "lexical_detection", "query": "gazed", "prompt": "Process the full audio offline. Find every instance of the word \"gazed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.52, "gazed"]], "response_mode": "offline", "answer": "[{\"word\":\"gazed\",\"start\":2.16,\"end\":2.52}]"}
{"utterance_id": "2078-142845-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.29, 1.97, "yeast"]], "response_mode": "offline", "answer": "[{\"word\":\"yeast\",\"start\":1.29,\"end\":1.97}]"}
{"utterance_id": "6345-93306-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0005.flac", "task_type": "lexical_detection", "query": "house", "prompt": "Process the full audio offline. Find every instance of the word \"house\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.46, "house"]], "response_mode": "offline", "answer": "[{\"word\":\"house\",\"start\":1.85,\"end\":2.46}]"}
{"utterance_id": "652-130726-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0010.flac", "task_type": "lexical_detection", "query": "surroundings", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"surroundings\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.76, "surroundings"]], "response_mode": "offline", "answer": "[{\"word\":\"surroundings\",\"start\":2.04,\"end\":2.76}]"}
{"utterance_id": "1462-170138-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0018.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.46, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":3.24,\"end\":3.46}]"}
{"utterance_id": "3752-4944-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0019.flac", "task_type": "lexical_detection", "query": "attacks", "prompt": "After the audio ends, extract all occurrences of the word \"attacks\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.78, "attacks"]], "response_mode": "offline", "answer": "[{\"word\":\"attacks\",\"start\":1.18,\"end\":1.78}]"}
{"utterance_id": "5895-34622-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.54, 1.02, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":0.54,\"end\":1.02}]"}
{"utterance_id": "8297-275155-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0023.flac", "task_type": "lexical_detection", "query": "father", "prompt": "When the utterance is complete, report all instances of the word \"father\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.19, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.83,\"end\":2.19}]"}
{"utterance_id": "84-121550-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0024.flac", "task_type": "semantic_detection", "query": "money", "prompt": "After listening to the whole recording, output every instance of a money or currency term with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.47, 2.92, "drachm"]], "response_mode": "offline", "answer": "[{\"word\":\"drachm\",\"start\":2.47,\"end\":2.92}]"}
{"utterance_id": "652-129742-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0000.flac", "task_type": "lexical_detection", "query": "asparagus", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"asparagus\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.13, "asparagus"], [2.89, 3.59, "asparagus"]], "response_mode": "offline", "answer": "[{\"word\":\"asparagus\",\"start\":0.51,\"end\":1.13},{\"word\":\"asparagus\",\"start\":2.89,\"end\":3.59}]"}
{"utterance_id": "3000-15664-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0005.flac", "task_type": "lexical_detection", "query": "blanket", "prompt": "After listening to the whole recording, output every instance of the word \"blanket\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 1.02, "blanket"]], "response_mode": "offline", "answer": "[{\"word\":\"blanket\",\"start\":0.61,\"end\":1.02}]"}
{"utterance_id": "3752-4943-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.85, 3.2, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":2.85,\"end\":3.2}]"}
{"utterance_id": "8297-275154-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0011.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.32, 3.7, "brother's"]], "response_mode": "offline", "answer": "[{\"word\":\"brother's\",\"start\":3.32,\"end\":3.7}]"}
{"utterance_id": "6241-61943-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0017.flac", "task_type": "lexical_detection", "query": "addressed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"addressed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.81, "addressed"]], "response_mode": "offline", "answer": "[{\"word\":\"addressed\",\"start\":1.47,\"end\":1.81}]"}
{"utterance_id": "251-137823-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0011.flac", "task_type": "lexical_detection", "query": "nearby", "prompt": "Process the full audio offline. Find every instance of the word \"nearby\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.59, 5.36, "nearby"]], "response_mode": "offline", "answer": "[{\"word\":\"nearby\",\"start\":4.59,\"end\":5.36}]"}
{"utterance_id": "2086-149220-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0044.flac", "task_type": "lexical_detection", "query": "wonderfully", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wonderfully\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 1.02, "wonderfully"]], "response_mode": "offline", "answer": "[{\"word\":\"wonderfully\",\"start\":0.54,\"end\":1.02}]"}
{"utterance_id": "2803-154328-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.52, "john mangles"]], "response_mode": "offline", "answer": "[{\"word\":\"john mangles\",\"start\":0.5,\"end\":1.52}]"}
{"utterance_id": "84-121123-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0007.flac", "task_type": "lexical_detection", "query": "mean", "prompt": "After the audio ends, extract all occurrences of the word \"mean\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.12, "mean"]], "response_mode": "offline", "answer": "[{\"word\":\"mean\",\"start\":0.87,\"end\":1.12}]"}
{"utterance_id": "84-121550-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.49, 2.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.49,\"end\":2.86}]"}
{"utterance_id": "5895-34629-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0025.flac", "task_type": "lexical_detection", "query": "fascinated", "prompt": "Process the full audio offline. Find every instance of the word \"fascinated\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.62, "fascinated"]], "response_mode": "offline", "answer": "[{\"word\":\"fascinated\",\"start\":1.81,\"end\":2.62}]"}
{"utterance_id": "5536-43359-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0016.flac", "task_type": "lexical_detection", "query": "founded", "prompt": "When the utterance is complete, report all instances of the word \"founded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.23, "founded"]], "response_mode": "offline", "answer": "[{\"word\":\"founded\",\"start\":1.78,\"end\":2.23}]"}
{"utterance_id": "5338-284437-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.22, 7.84, "tourmaline"]], "response_mode": "offline", "answer": "[{\"word\":\"tourmaline\",\"start\":7.22,\"end\":7.84}]"}
{"utterance_id": "5895-34629-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0027.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "When the utterance is complete, report all instances of a weather term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.36, "clouds"]], "response_mode": "offline", "answer": "[{\"word\":\"clouds\",\"start\":0.72,\"end\":1.36}]"}
{"utterance_id": "3081-166546-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0016.flac", "task_type": "lexical_detection", "query": "ambitious", "prompt": "Process the full audio offline. Find every instance of the word \"ambitious\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 3.04, "ambitious"]], "response_mode": "offline", "answer": "[{\"word\":\"ambitious\",\"start\":2.32,\"end\":3.04}]"}
{"utterance_id": "652-130726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0004.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.18, 2.5, "food"], [6.02, 7.02, "luncheon dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":2.18,\"end\":2.5},{\"word\":\"luncheon dinner\",\"start\":6.02,\"end\":7.02}]"}
{"utterance_id": "8842-302196-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0011.flac", "task_type": "lexical_detection", "query": "title", "prompt": "After the audio ends, extract all occurrences of the word \"title\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.05, "title"]], "response_mode": "offline", "answer": "[{\"word\":\"title\",\"start\":1.7,\"end\":2.05}]"}
{"utterance_id": "2412-153948-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0013.flac", "task_type": "semantic_counting", "query": "money", "prompt": "When the utterance is complete, report how a money or currency term accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.37, "one"], [5.39, 5.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.95,\"end\":1.37},{\"word\":\"two\",\"start\":5.39,\"end\":5.97}]"}
{"utterance_id": "2078-142845-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0034.flac", "task_type": "lexical_counting", "query": "served", "prompt": "Listen to the full audio. Count every occurrence of the word \"served\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.41, 2.97, "one"], [6.16, 6.78, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.41,\"end\":2.97},{\"word\":\"two\",\"start\":6.16,\"end\":6.78}]"}
{"utterance_id": "8297-275154-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0017.flac", "task_type": "lexical_detection", "query": "why", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"why\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.45, "why"]], "response_mode": "offline", "answer": "[{\"word\":\"why\",\"start\":0.98,\"end\":1.45}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.39, 1.03, "miss woodley"]], "response_mode": "offline", "answer": "[{\"word\":\"miss woodley\",\"start\":0.39,\"end\":1.03}]"}
{"utterance_id": "2803-161169-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0013.flac", "task_type": "lexical_detection", "query": "hills", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"hills\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.63, 1.23, "hills"]], "response_mode": "offline", "answer": "[{\"word\":\"hills\",\"start\":0.63,\"end\":1.23}]"}
{"utterance_id": "6345-93302-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0011.flac", "task_type": "lexical_detection", "query": "cold", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"cold\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.58, 2.03, "cold"]], "response_mode": "offline", "answer": "[{\"word\":\"cold\",\"start\":1.58,\"end\":2.03}]"}
{"utterance_id": "1272-128104-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0009.flac", "task_type": "lexical_detection", "query": "decorative", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"decorative\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.24, 4.8, "decorative"]], "response_mode": "offline", "answer": "[{\"word\":\"decorative\",\"start\":4.24,\"end\":4.8}]"}
{"utterance_id": "84-121550-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0027.flac", "task_type": "lexical_detection", "query": "beatrice", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"beatrice\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.92, 3.64, "beatrice"]], "response_mode": "offline", "answer": "[{\"word\":\"beatrice\",\"start\":2.92,\"end\":3.64}]"}
{"utterance_id": "3170-137482-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0017.flac", "task_type": "lexical_counting", "query": "hour", "prompt": "After listening to the whole recording, output timed count words for the word \"hour\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[6.21, 6.43, "one"], [12.7, 13.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.21,\"end\":6.43},{\"word\":\"two\",\"start\":12.7,\"end\":13.0}]"}
{"utterance_id": "3170-137482-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0039.flac", "task_type": "lexical_detection", "query": "heavenly", "prompt": "After the audio ends, extract all occurrences of the word \"heavenly\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.06, 9.76, "heavenly"]], "response_mode": "offline", "answer": "[{\"word\":\"heavenly\",\"start\":9.06,\"end\":9.76}]"}
{"utterance_id": "2277-149896-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0006.flac", "task_type": "lexical_detection", "query": "carrie", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"carrie\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.71, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":2.4,\"end\":2.71}]"}
{"utterance_id": "251-137823-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 1.77, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":1.48,\"end\":1.77}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "When the utterance is complete, report how a plant name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.32, 0.79, "one"], [5.3, 5.78, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.32,\"end\":0.79},{\"word\":\"two\",\"start\":5.3,\"end\":5.78}]"}
{"utterance_id": "6295-244435-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0009.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.27, 0.6, "one"], [3.8, 4.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.27,\"end\":0.6},{\"word\":\"two\",\"start\":3.8,\"end\":4.08}]"}
{"utterance_id": "6313-66129-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0034.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.69, 1.89, "foot"], [3.05, 3.35, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":1.69,\"end\":1.89},{\"word\":\"body\",\"start\":3.05,\"end\":3.35}]"}
{"utterance_id": "1462-170142-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.08, 1.41, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.08,\"end\":1.41}]"}
{"utterance_id": "3081-166546-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0010.flac", "task_type": "lexical_detection", "query": "foolish", "prompt": "Once the audio has finished, identify every instance of the word \"foolish\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.46, 1.99, "foolish"]], "response_mode": "offline", "answer": "[{\"word\":\"foolish\",\"start\":1.46,\"end\":1.99}]"}
{"utterance_id": "1673-143397-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0007.flac", "task_type": "lexical_detection", "query": "exterminate", "prompt": "After the audio ends, extract all occurrences of the word \"exterminate\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 1.01, "exterminate"], [3.19, 3.88, "exterminate"]], "response_mode": "offline", "answer": "[{\"word\":\"exterminate\",\"start\":0.31,\"end\":1.01},{\"word\":\"exterminate\",\"start\":3.19,\"end\":3.88}]"}
{"utterance_id": "777-126732-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0008.flac", "task_type": "lexical_detection", "query": "stranger", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"stranger\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.64, 4.23, "stranger"]], "response_mode": "offline", "answer": "[{\"word\":\"stranger\",\"start\":3.64,\"end\":4.23}]"}
{"utterance_id": "2035-147961-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.08, 4.51, "parents"]], "response_mode": "offline", "answer": "[{\"word\":\"parents\",\"start\":4.08,\"end\":4.51}]"}
{"utterance_id": "3170-137482-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0016.flac", "task_type": "lexical_detection", "query": "nocturnal", "prompt": "When the utterance is complete, report all instances of the word \"nocturnal\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.59, "nocturnal"]], "response_mode": "offline", "answer": "[{\"word\":\"nocturnal\",\"start\":1.92,\"end\":2.59}]"}
{"utterance_id": "2428-83705-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.88, 11.68, "mary ann's"]], "response_mode": "offline", "answer": "[{\"word\":\"mary ann's\",\"start\":10.88,\"end\":11.68}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.56, 2.95, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.56,\"end\":2.95}]"}
{"utterance_id": "1993-147149-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0010.flac", "task_type": "lexical_detection", "query": "wishing", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"wishing\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.91, "wishing"]], "response_mode": "offline", "answer": "[{\"word\":\"wishing\",\"start\":0.52,\"end\":0.91}]"}
{"utterance_id": "6241-61946-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.62, 0.97, "horses"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":0.62,\"end\":0.97}]"}
{"utterance_id": "2086-149220-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0012.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.84, "chicken"]], "response_mode": "offline", "answer": "[{\"word\":\"chicken\",\"start\":0.51,\"end\":0.84}]"}
{"utterance_id": "7976-110523-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0010.flac", "task_type": "lexical_detection", "query": "tore", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"tore\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.47, 3.78, "tore"]], "response_mode": "offline", "answer": "[{\"word\":\"tore\",\"start\":3.47,\"end\":3.78}]"}
{"utterance_id": "251-137823-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.38, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":2.09,\"end\":2.38}]"}
{"utterance_id": "5536-43363-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0000.flac", "task_type": "lexical_detection", "query": "spirits", "prompt": "When the utterance is complete, report all instances of the word \"spirits\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.74, "spirits"]], "response_mode": "offline", "answer": "[{\"word\":\"spirits\",\"start\":2.08,\"end\":2.74}]"}
{"utterance_id": "1673-143396-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.45, 6.06, "cerinthus"]], "response_mode": "offline", "answer": "[{\"word\":\"cerinthus\",\"start\":5.45,\"end\":6.06}]"}
{"utterance_id": "5895-34629-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0012.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 1.15, "southwark"]], "response_mode": "offline", "answer": "[{\"word\":\"southwark\",\"start\":0.55,\"end\":1.15}]"}
{"utterance_id": "5895-34629-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0007.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.9, 5.15, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":4.9,\"end\":5.15}]"}
{"utterance_id": "84-121550-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0001.flac", "task_type": "lexical_detection", "query": "compared", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"compared\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 6.01, "compared"]], "response_mode": "offline", "answer": "[{\"word\":\"compared\",\"start\":5.55,\"end\":6.01}]"}
{"utterance_id": "652-129742-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.93, 1.63, "french dressing"]], "response_mode": "offline", "answer": "[{\"word\":\"french dressing\",\"start\":0.93,\"end\":1.63}]"}
{"utterance_id": "2035-147961-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0001.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.81, "one"], [3.63, 4.04, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.47,\"end\":0.81},{\"word\":\"two\",\"start\":3.63,\"end\":4.04}]"}
{"utterance_id": "7976-110124-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.42, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":1.05,\"end\":1.42}]"}
{"utterance_id": "3536-23268-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0011.flac", "task_type": "lexical_counting", "query": "motionless", "prompt": "Listen to the entire clip, then report the count sequence for the word \"motionless\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.84, "one"], [3.85, 4.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.27,\"end\":1.84},{\"word\":\"two\",\"start\":3.85,\"end\":4.36}]"}
{"utterance_id": "5338-284437-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.87, 1.48, "coralie"]], "response_mode": "offline", "answer": "[{\"word\":\"coralie\",\"start\":0.87,\"end\":1.48}]"}
{"utterance_id": "422-122949-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0015.flac", "task_type": "lexical_detection", "query": "personal", "prompt": "After the audio ends, extract all occurrences of the word \"personal\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.41, 14.02, "personal"]], "response_mode": "offline", "answer": "[{\"word\":\"personal\",\"start\":13.41,\"end\":14.02}]"}
{"utterance_id": "6345-64257-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0017.flac", "task_type": "lexical_detection", "query": "life", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"life\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.16, "life"]], "response_mode": "offline", "answer": "[{\"word\":\"life\",\"start\":0.78,\"end\":1.16}]"}
{"utterance_id": "7976-110124-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0011.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.84, 4.22, "sister"]], "response_mode": "offline", "answer": "[{\"word\":\"sister\",\"start\":3.84,\"end\":4.22}]"}
{"utterance_id": "6319-275224-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0017.flac", "task_type": "lexical_detection", "query": "orderly", "prompt": "After listening to the whole recording, output every instance of the word \"orderly\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.94, 6.45, "orderly"]], "response_mode": "offline", "answer": "[{\"word\":\"orderly\",\"start\":5.94,\"end\":6.45}]"}
{"utterance_id": "2428-83699-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0005.flac", "task_type": "lexical_detection", "query": "establishment", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"establishment\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.89, 5.69, "establishment"]], "response_mode": "offline", "answer": "[{\"word\":\"establishment\",\"start\":4.89,\"end\":5.69}]"}
{"utterance_id": "7850-286674-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.67, 9.13, "larva"]], "response_mode": "offline", "answer": "[{\"word\":\"larva\",\"start\":8.67,\"end\":9.13}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 3.97, "pair"]], "response_mode": "offline", "answer": "[{\"word\":\"pair\",\"start\":3.75,\"end\":3.97}]"}
{"utterance_id": "1988-24833-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0000.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.61, "kittens"]], "response_mode": "offline", "answer": "[{\"word\":\"kittens\",\"start\":1.11,\"end\":1.61}]"}
{"utterance_id": "6313-76958-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0017.flac", "task_type": "lexical_detection", "query": "lumpy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lumpy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.77, "lumpy"]], "response_mode": "offline", "answer": "[{\"word\":\"lumpy\",\"start\":0.38,\"end\":0.77}]"}
{"utterance_id": "1993-147149-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0016.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "When the utterance is complete, report all instances of an ordinal number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.6, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":2.23,\"end\":2.6}]"}
{"utterance_id": "3170-137482-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0023.flac", "task_type": "lexical_detection", "query": "physician", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"physician\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.39, 4.98, "physician"]], "response_mode": "offline", "answer": "[{\"word\":\"physician\",\"start\":4.39,\"end\":4.98}]"}
{"utterance_id": "6319-275224-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0019.flac", "task_type": "lexical_detection", "query": "gentle", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"gentle\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.06, 3.44, "gentle"]], "response_mode": "offline", "answer": "[{\"word\":\"gentle\",\"start\":3.06,\"end\":3.44}]"}
{"utterance_id": "5536-43363-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.95, 2.24, "food"], [9.07, 9.46, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":1.95,\"end\":2.24},{\"word\":\"food\",\"start\":9.07,\"end\":9.46}]"}
{"utterance_id": "5536-43363-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0012.flac", "task_type": "lexical_detection", "query": "prophecies", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"prophecies\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.95, 2.78, "prophecies"]], "response_mode": "offline", "answer": "[{\"word\":\"prophecies\",\"start\":1.95,\"end\":2.78}]"}
{"utterance_id": "6319-275224-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0001.flac", "task_type": "lexical_detection", "query": "forced", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"forced\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.56, 6.12, "forced"]], "response_mode": "offline", "answer": "[{\"word\":\"forced\",\"start\":5.56,\"end\":6.12}]"}
{"utterance_id": "8842-302196-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0005.flac", "task_type": "lexical_detection", "query": "grace", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"grace\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.8, "grace"]], "response_mode": "offline", "answer": "[{\"word\":\"grace\",\"start\":2.43,\"end\":2.8}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.14, 4.53, "sister's"]], "response_mode": "offline", "answer": "[{\"word\":\"sister's\",\"start\":4.14,\"end\":4.53}]"}
{"utterance_id": "2035-147961-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0034.flac", "task_type": "lexical_detection", "query": "wherever", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wherever\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.37, 0.88, "wherever"]], "response_mode": "offline", "answer": "[{\"word\":\"wherever\",\"start\":0.37,\"end\":0.88}]"}
{"utterance_id": "1988-24833-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Process the full audio offline. Find every instance of a disease name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.65, 4.07, "asthma"]], "response_mode": "offline", "answer": "[{\"word\":\"asthma\",\"start\":3.65,\"end\":4.07}]"}
{"utterance_id": "1988-24833-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0026.flac", "task_type": "lexical_detection", "query": "i've", "prompt": "After listening to the whole recording, output every instance of the word \"i've\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.68, "i've"]], "response_mode": "offline", "answer": "[{\"word\":\"i've\",\"start\":0.52,\"end\":0.68}]"}
{"utterance_id": "422-122949-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0028.flac", "task_type": "lexical_detection", "query": "thou", "prompt": "Once the audio has finished, identify every instance of the word \"thou\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.18, 1.38, "thou"]], "response_mode": "offline", "answer": "[{\"word\":\"thou\",\"start\":1.18,\"end\":1.38}]"}
{"utterance_id": "5536-43363-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0017.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 4.15, "fort ellis"]], "response_mode": "offline", "answer": "[{\"word\":\"fort ellis\",\"start\":3.5,\"end\":4.15}]"}
{"utterance_id": "8842-304647-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0010.flac", "task_type": "lexical_detection", "query": "strain", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"strain\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.24, 8.91, "strain"]], "response_mode": "offline", "answer": "[{\"word\":\"strain\",\"start\":8.24,\"end\":8.91}]"}
{"utterance_id": "251-136532-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0003.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.35, 4.04, "space force"]], "response_mode": "offline", "answer": "[{\"word\":\"space force\",\"start\":3.35,\"end\":4.04}]"}
{"utterance_id": "652-129742-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0015.flac", "task_type": "lexical_detection", "query": "ounces", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ounces\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 2.75, "ounces"]], "response_mode": "offline", "answer": "[{\"word\":\"ounces\",\"start\":2.34,\"end\":2.75}]"}
{"utterance_id": "84-121550-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0000.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.42, "prime"]], "response_mode": "offline", "answer": "[{\"word\":\"prime\",\"start\":2.93,\"end\":3.42}]"}
{"utterance_id": "5536-43358-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0019.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.33, 3.67, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.33,\"end\":3.67}]"}
{"utterance_id": "174-50561-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.43, 4.91, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":4.43,\"end\":4.91}]"}
{"utterance_id": "6241-61943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.35, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":0.84,\"end\":1.35}]"}
{"utterance_id": "6345-93302-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0000.flac", "task_type": "lexical_detection", "query": "poem", "prompt": "After the audio ends, extract all occurrences of the word \"poem\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.64, "poem"]], "response_mode": "offline", "answer": "[{\"word\":\"poem\",\"start\":2.02,\"end\":2.64}]"}
{"utterance_id": "8297-275155-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0017.flac", "task_type": "lexical_detection", "query": "wife", "prompt": "After the audio ends, extract all occurrences of the word \"wife\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.67, 6.16, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":5.67,\"end\":6.16}]"}
{"utterance_id": "422-122949-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.81, 7.18, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"blood\",\"start\":6.81,\"end\":7.18}]"}
{"utterance_id": "6313-66129-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0015.flac", "task_type": "lexical_detection", "query": "unable", "prompt": "After listening to the whole recording, output every instance of the word \"unable\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 1.75, "unable"]], "response_mode": "offline", "answer": "[{\"word\":\"unable\",\"start\":1.32,\"end\":1.75}]"}
{"utterance_id": "7850-281318-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0003.flac", "task_type": "lexical_detection", "query": "home", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"home\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.81, "home"]], "response_mode": "offline", "answer": "[{\"word\":\"home\",\"start\":2.08,\"end\":2.81}]"}
{"utterance_id": "5338-284437-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.19, 0.8, "coralie"]], "response_mode": "offline", "answer": "[{\"word\":\"coralie\",\"start\":0.19,\"end\":0.8}]"}
{"utterance_id": "2428-83705-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0016.flac", "task_type": "lexical_detection", "query": "flow", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"flow\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.48, "flow"]], "response_mode": "offline", "answer": "[{\"word\":\"flow\",\"start\":2.17,\"end\":2.48}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.11, 4.58, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":4.11,\"end\":4.58}]"}
{"utterance_id": "652-130737-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0012.flac", "task_type": "lexical_detection", "query": "ninety", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"ninety\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.53, 2.86, "ninety"]], "response_mode": "offline", "answer": "[{\"word\":\"ninety\",\"start\":2.53,\"end\":2.86}]"}
{"utterance_id": "1462-170142-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0003.flac", "task_type": "lexical_detection", "query": "empty", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"empty\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.42, "empty"]], "response_mode": "offline", "answer": "[{\"word\":\"empty\",\"start\":1.08,\"end\":1.42}]"}
{"utterance_id": "6313-76958-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0025.flac", "task_type": "lexical_detection", "query": "wrangler", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"wrangler\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.93, "wrangler"]], "response_mode": "offline", "answer": "[{\"word\":\"wrangler\",\"start\":1.39,\"end\":1.93}]"}
{"utterance_id": "2035-147961-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0025.flac", "task_type": "lexical_detection", "query": "pavel", "prompt": "After the audio ends, extract all occurrences of the word \"pavel\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.6, "pavel"]], "response_mode": "offline", "answer": "[{\"word\":\"pavel\",\"start\":1.22,\"end\":1.6}]"}
{"utterance_id": "1988-24833-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0018.flac", "task_type": "lexical_detection", "query": "finding", "prompt": "After listening to the whole recording, output every instance of the word \"finding\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.05, 4.42, "finding"]], "response_mode": "offline", "answer": "[{\"word\":\"finding\",\"start\":4.05,\"end\":4.42}]"}
{"utterance_id": "3752-4944-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0068.flac", "task_type": "lexical_detection", "query": "particular", "prompt": "Once the audio has finished, identify every instance of the word \"particular\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.98, 3.5, "particular"]], "response_mode": "offline", "answer": "[{\"word\":\"particular\",\"start\":2.98,\"end\":3.5}]"}
{"utterance_id": "1673-143396-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.65, 2.66, "abbot serapion"]], "response_mode": "offline", "answer": "[{\"word\":\"abbot serapion\",\"start\":1.65,\"end\":2.66}]"}
{"utterance_id": "652-130726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0001.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.68, "o'farrell"], [3.71, 4.22, "fillmore"]], "response_mode": "offline", "answer": "[{\"word\":\"o'farrell\",\"start\":2.25,\"end\":2.68},{\"word\":\"fillmore\",\"start\":3.71,\"end\":4.22}]"}
{"utterance_id": "6319-64726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0007.flac", "task_type": "lexical_detection", "query": "enchanted", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"enchanted\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.93, "enchanted"]], "response_mode": "offline", "answer": "[{\"word\":\"enchanted\",\"start\":1.41,\"end\":1.93}]"}
{"utterance_id": "3536-23268-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.68, 10.18, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":9.68,\"end\":10.18}]"}
{"utterance_id": "3576-138058-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.53, 6.98, "hair"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":6.53,\"end\":6.98}]"}
{"utterance_id": "6295-244435-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0032.flac", "task_type": "lexical_detection", "query": "slopes", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"slopes\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.9, "slopes"]], "response_mode": "offline", "answer": "[{\"word\":\"slopes\",\"start\":2.4,\"end\":2.9}]"}
{"utterance_id": "1272-141231-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0003.flac", "task_type": "lexical_detection", "query": "panic", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"panic\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.52, "panic"]], "response_mode": "offline", "answer": "[{\"word\":\"panic\",\"start\":1.04,\"end\":1.52}]"}
{"utterance_id": "1272-141231-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0000.flac", "task_type": "lexical_detection", "query": "exist", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"exist\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.57, 4.25, "exist"]], "response_mode": "offline", "answer": "[{\"word\":\"exist\",\"start\":3.57,\"end\":4.25}]"}
{"utterance_id": "5694-64038-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.77, 9.12, "seven"]], "response_mode": "offline", "answer": "[{\"word\":\"seven\",\"start\":8.77,\"end\":9.12}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.18, 5.51, "final"]], "response_mode": "offline", "answer": "[{\"word\":\"final\",\"start\":5.18,\"end\":5.51}]"}
{"utterance_id": "2277-149897-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0036.flac", "task_type": "lexical_detection", "query": "spread", "prompt": "Once the audio has finished, identify every instance of the word \"spread\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.23, 1.52, "spread"]], "response_mode": "offline", "answer": "[{\"word\":\"spread\",\"start\":1.23,\"end\":1.52}]"}
{"utterance_id": "2412-153948-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0011.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Once the audio has finished, identify every instance of a weather term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.84, 3.22, "clouds"]], "response_mode": "offline", "answer": "[{\"word\":\"clouds\",\"start\":2.84,\"end\":3.22}]"}
{"utterance_id": "1988-147956-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0027.flac", "task_type": "lexical_detection", "query": "antonia", "prompt": "Once the audio has finished, identify every instance of the word \"antonia\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.88, 1.43, "antonia"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia\",\"start\":0.88,\"end\":1.43}]"}
{"utterance_id": "6319-275224-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "When the utterance is complete, report all instances of a color name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.69, 9.04, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":8.69,\"end\":9.04}]"}
{"utterance_id": "3853-163249-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.29, 0.66, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.29,\"end\":0.66}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.36, "one"], [7.67, 8.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.9,\"end\":2.36},{\"word\":\"two\",\"start\":7.67,\"end\":8.16}]"}
{"utterance_id": "1988-24833-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0019.flac", "task_type": "lexical_detection", "query": "quarter", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"quarter\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.75, 2.03, "quarter"]], "response_mode": "offline", "answer": "[{\"word\":\"quarter\",\"start\":1.75,\"end\":2.03}]"}
{"utterance_id": "6295-244435-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0036.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.99, 4.75, "liveryman"]], "response_mode": "offline", "answer": "[{\"word\":\"liveryman\",\"start\":3.99,\"end\":4.75}]"}
{"utterance_id": "84-121550-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0024.flac", "task_type": "lexical_detection", "query": "virgilius", "prompt": "After the audio ends, extract all occurrences of the word \"virgilius\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.8, "virgilius"]], "response_mode": "offline", "answer": "[{\"word\":\"virgilius\",\"start\":1.02,\"end\":1.8}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[15.37, 15.61, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":15.37,\"end\":15.61}]"}
{"utterance_id": "1988-147956-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0010.flac", "task_type": "lexical_detection", "query": "remembered", "prompt": "After the audio ends, extract all occurrences of the word \"remembered\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.63, 1.26, "remembered"]], "response_mode": "offline", "answer": "[{\"word\":\"remembered\",\"start\":0.63,\"end\":1.26}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "lexical_detection", "query": "wind", "prompt": "When the utterance is complete, report all instances of the word \"wind\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.08, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":3.58,\"end\":4.08}]"}
{"utterance_id": "8842-302196-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.91, 8.39, "dante"]], "response_mode": "offline", "answer": "[{\"word\":\"dante\",\"start\":7.91,\"end\":8.39}]"}
{"utterance_id": "6241-66616-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.88, 5.22, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":4.88,\"end\":5.22}]"}
{"utterance_id": "6345-93306-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0017.flac", "task_type": "lexical_detection", "query": "lock", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"lock\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.73, "lock"]], "response_mode": "offline", "answer": "[{\"word\":\"lock\",\"start\":0.52,\"end\":0.73}]"}
{"utterance_id": "2086-149220-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0007.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.6, 2.34, "chanticleer"], [4.39, 4.81, "chicken"]], "response_mode": "offline", "answer": "[{\"word\":\"chanticleer\",\"start\":1.6,\"end\":2.34},{\"word\":\"chicken\",\"start\":4.39,\"end\":4.81}]"}
{"utterance_id": "5694-64038-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0019.flac", "task_type": "lexical_detection", "query": "bye", "prompt": "Once the audio has finished, identify every instance of the word \"bye\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.84, 3.35, "bye"]], "response_mode": "offline", "answer": "[{\"word\":\"bye\",\"start\":2.84,\"end\":3.35}]"}
{"utterance_id": "3170-137482-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0007.flac", "task_type": "lexical_counting", "query": "bell", "prompt": "Listen to the full audio. Count every occurrence of the word \"bell\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.05, 7.44, "one"], [11.96, 12.21, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.05,\"end\":7.44},{\"word\":\"two\",\"start\":11.96,\"end\":12.21}]"}
{"utterance_id": "174-84280-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0007.flac", "task_type": "lexical_detection", "query": "anxious", "prompt": "When the utterance is complete, report all instances of the word \"anxious\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.34, "anxious"]], "response_mode": "offline", "answer": "[{\"word\":\"anxious\",\"start\":1.88,\"end\":2.34}]"}
{"utterance_id": "8297-275156-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.97, 5.39, "randal's"]], "response_mode": "offline", "answer": "[{\"word\":\"randal's\",\"start\":4.97,\"end\":5.39}]"}
{"utterance_id": "8842-304647-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.84, "organ"], [3.58, 3.91, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"organ\",\"start\":0.53,\"end\":0.84},{\"word\":\"body\",\"start\":3.58,\"end\":3.91}]"}
{"utterance_id": "2035-147961-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0004.flac", "task_type": "lexical_detection", "query": "draw", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"draw\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.2, 5.58, "draw"]], "response_mode": "offline", "answer": "[{\"word\":\"draw\",\"start\":5.2,\"end\":5.58}]"}
{"utterance_id": "1919-142785-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0044.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.49, 5.25, "ingredients"]], "response_mode": "offline", "answer": "[{\"word\":\"ingredients\",\"start\":4.49,\"end\":5.25}]"}
{"utterance_id": "174-168635-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0013.flac", "task_type": "lexical_detection", "query": "beings", "prompt": "After the audio ends, extract all occurrences of the word \"beings\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.83, 3.42, "beings"]], "response_mode": "offline", "answer": "[{\"word\":\"beings\",\"start\":2.83,\"end\":3.42}]"}
{"utterance_id": "6295-244435-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0001.flac", "task_type": "lexical_detection", "query": "colonel", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"colonel\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 0.63, "colonel"]], "response_mode": "offline", "answer": "[{\"word\":\"colonel\",\"start\":0.26,\"end\":0.63}]"}
{"utterance_id": "3081-166546-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0057.flac", "task_type": "lexical_detection", "query": "anybody", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"anybody\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.23, 0.81, "anybody"]], "response_mode": "offline", "answer": "[{\"word\":\"anybody\",\"start\":0.23,\"end\":0.81}]"}
{"utterance_id": "777-126732-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0066.flac", "task_type": "lexical_detection", "query": "willing", "prompt": "Once the audio has finished, identify every instance of the word \"willing\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.76, 3.14, "willing"]], "response_mode": "offline", "answer": "[{\"word\":\"willing\",\"start\":2.76,\"end\":3.14}]"}
{"utterance_id": "7976-110124-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0012.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 3.01, "apples"], [4.55, 5.11, "dessert"]], "response_mode": "offline", "answer": "[{\"word\":\"apples\",\"start\":2.5,\"end\":3.01},{\"word\":\"dessert\",\"start\":4.55,\"end\":5.11}]"}
{"utterance_id": "652-130726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0004.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After listening to the whole recording, output timed count words for a food name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.18, 2.5, "one"], [6.02, 7.02, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.18,\"end\":2.5},{\"word\":\"two\",\"start\":6.02,\"end\":7.02}]"}
{"utterance_id": "3853-163249-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.52, 4.98, "lips"], [6.18, 6.54, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":4.52,\"end\":4.98},{\"word\":\"eyes\",\"start\":6.18,\"end\":6.54}]"}
{"utterance_id": "2412-153954-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0014.flac", "task_type": "semantic_counting", "query": "city", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a city name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[5.01, 5.7, "one"], [7.16, 7.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.01,\"end\":5.7},{\"word\":\"two\",\"start\":7.16,\"end\":7.54}]"}
{"utterance_id": "777-126732-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0077.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.22, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":0.91,\"end\":1.22}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Listen to the full audio. Count every occurrence of a plant name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.69, 3.4, "one"], [5.13, 5.64, "two"], [8.33, 9.14, "three"], [20.67, 21.3, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.69,\"end\":3.4},{\"word\":\"two\",\"start\":5.13,\"end\":5.64},{\"word\":\"three\",\"start\":8.33,\"end\":9.14},{\"word\":\"four\",\"start\":20.67,\"end\":21.3}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen carefully to the entire clip, then report every instance of a time of day as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.54, 9.02, "noontide"]], "response_mode": "offline", "answer": "[{\"word\":\"noontide\",\"start\":8.54,\"end\":9.02}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.11, "hair"], [6.33, 6.76, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":2.87,\"end\":3.11},{\"word\":\"head\",\"start\":6.33,\"end\":6.76}]"}
{"utterance_id": "84-121550-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 3.3, "proserpina"]], "response_mode": "offline", "answer": "[{\"word\":\"proserpina\",\"start\":2.61,\"end\":3.3}]"}
{"utterance_id": "84-121550-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0022.flac", "task_type": "lexical_detection", "query": "pierced", "prompt": "Process the full audio offline. Find every instance of the word \"pierced\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.3, 4.74, "pierced"]], "response_mode": "offline", "answer": "[{\"word\":\"pierced\",\"start\":4.3,\"end\":4.74}]"}
{"utterance_id": "7850-281318-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0021.flac", "task_type": "lexical_detection", "query": "elm", "prompt": "Process the full audio offline. Find every instance of the word \"elm\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.58, "elm"]], "response_mode": "offline", "answer": "[{\"word\":\"elm\",\"start\":3.29,\"end\":3.58}]"}
{"utterance_id": "8297-275154-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0017.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.37, "governess"]], "response_mode": "offline", "answer": "[{\"word\":\"governess\",\"start\":2.65,\"end\":3.37}]"}
{"utterance_id": "7976-105575-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0020.flac", "task_type": "lexical_detection", "query": "lines", "prompt": "After the audio ends, extract all occurrences of the word \"lines\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.88, "lines"]], "response_mode": "offline", "answer": "[{\"word\":\"lines\",\"start\":1.25,\"end\":1.88}]"}
{"utterance_id": "8297-275154-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0004.flac", "task_type": "lexical_detection", "query": "finds", "prompt": "Process the full audio offline. Find every instance of the word \"finds\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.52, 5.01, "finds"]], "response_mode": "offline", "answer": "[{\"word\":\"finds\",\"start\":4.52,\"end\":5.01}]"}
{"utterance_id": "3853-163249-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0010.flac", "task_type": "lexical_detection", "query": "ambition", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"ambition\" with fields word, start, end (seconds). If none, reply [].", "labels": [[8.53, 9.11, "ambition"]], "response_mode": "offline", "answer": "[{\"word\":\"ambition\",\"start\":8.53,\"end\":9.11}]"}
{"utterance_id": "2086-149220-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.78, "hens"], [3.5, 3.73, "hens"], [7.98, 8.42, "fowl"]], "response_mode": "offline", "answer": "[{\"word\":\"hens\",\"start\":1.39,\"end\":1.78},{\"word\":\"hens\",\"start\":3.5,\"end\":3.73},{\"word\":\"fowl\",\"start\":7.98,\"end\":8.42}]"}
{"utterance_id": "6345-93302-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0010.flac", "task_type": "lexical_detection", "query": "inside", "prompt": "Process the full audio offline. Find every instance of the word \"inside\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.88, "inside"]], "response_mode": "offline", "answer": "[{\"word\":\"inside\",\"start\":2.43,\"end\":2.88}]"}
{"utterance_id": "3170-137482-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0027.flac", "task_type": "lexical_detection", "query": "named", "prompt": "After the audio ends, extract all occurrences of the word \"named\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.43, "named"]], "response_mode": "offline", "answer": "[{\"word\":\"named\",\"start\":2.18,\"end\":2.43}]"}
{"utterance_id": "2428-83699-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0040.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.24, 7.63, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":7.24,\"end\":7.63}]"}
{"utterance_id": "2277-149896-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0000.flac", "task_type": "lexical_detection", "query": "fevered", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"fevered\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.73, 1.15, "fevered"]], "response_mode": "offline", "answer": "[{\"word\":\"fevered\",\"start\":0.73,\"end\":1.15}]"}
{"utterance_id": "6319-275224-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0009.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.72, 2.21, "flower"]], "response_mode": "offline", "answer": "[{\"word\":\"flower\",\"start\":1.72,\"end\":2.21}]"}
{"utterance_id": "7850-286674-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0000.flac", "task_type": "lexical_detection", "query": "fond", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"fond\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.39, 5.73, "fond"]], "response_mode": "offline", "answer": "[{\"word\":\"fond\",\"start\":5.39,\"end\":5.73}]"}
{"utterance_id": "8297-275155-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.76, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":0.39,\"end\":0.76}]"}
{"utterance_id": "6241-66616-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.34, 9.76, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":9.34,\"end\":9.76}]"}
{"utterance_id": "1673-143397-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0015.flac", "task_type": "lexical_detection", "query": "provinces", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"provinces\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.57, "provinces"]], "response_mode": "offline", "answer": "[{\"word\":\"provinces\",\"start\":0.91,\"end\":1.57}]"}
{"utterance_id": "3853-163249-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0044.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.23, "david"], [2.57, 3.12, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.76,\"end\":1.23},{\"word\":\"christie\",\"start\":2.57,\"end\":3.12}]"}
{"utterance_id": "2086-149220-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0028.flac", "task_type": "lexical_detection", "query": "glanced", "prompt": "After listening to the whole recording, output every instance of the word \"glanced\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.41, "glanced"]], "response_mode": "offline", "answer": "[{\"word\":\"glanced\",\"start\":1.01,\"end\":1.41}]"}
{"utterance_id": "2428-83699-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0017.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.29, 1.81, "crofton"]], "response_mode": "offline", "answer": "[{\"word\":\"crofton\",\"start\":1.29,\"end\":1.81}]"}
{"utterance_id": "6313-66125-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0011.flac", "task_type": "lexical_detection", "query": "haul", "prompt": "Process the full audio offline. Find every instance of the word \"haul\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.35, 3.66, "haul"]], "response_mode": "offline", "answer": "[{\"word\":\"haul\",\"start\":3.35,\"end\":3.66}]"}
{"utterance_id": "6345-64257-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0003.flac", "task_type": "lexical_detection", "query": "sad", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sad\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.79, 3.38, "sad"]], "response_mode": "offline", "answer": "[{\"word\":\"sad\",\"start\":2.79,\"end\":3.38}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "lexical_counting", "query": "legs", "prompt": "Once the audio has finished, list each occurrence of the word \"legs\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[3.15, 3.52, "one"], [6.67, 7.05, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.15,\"end\":3.52},{\"word\":\"two\",\"start\":6.67,\"end\":7.05}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.71, "kent"], [5.72, 6.13, "rome"], [16.28, 16.82, "iona"]], "response_mode": "offline", "answer": "[{\"word\":\"kent\",\"start\":1.26,\"end\":1.71},{\"word\":\"rome\",\"start\":5.72,\"end\":6.13},{\"word\":\"iona\",\"start\":16.28,\"end\":16.82}]"}
{"utterance_id": "1272-141231-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0029.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.68, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":1.29,\"end\":1.68}]"}
{"utterance_id": "2428-83699-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0000.flac", "task_type": "lexical_detection", "query": "imagine", "prompt": "When the utterance is complete, report all instances of the word \"imagine\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 0.98, "imagine"]], "response_mode": "offline", "answer": "[{\"word\":\"imagine\",\"start\":0.6,\"end\":0.98}]"}
{"utterance_id": "3752-4944-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0042.flac", "task_type": "lexical_detection", "query": "comfortable", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"comfortable\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.35, 3.03, "comfortable"]], "response_mode": "offline", "answer": "[{\"word\":\"comfortable\",\"start\":2.35,\"end\":3.03}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.14, 1.54, "beast"], [13.93, 14.24, "beast"], [21.53, 22.07, "bugbears"]], "response_mode": "offline", "answer": "[{\"word\":\"beast\",\"start\":1.14,\"end\":1.54},{\"word\":\"beast\",\"start\":13.93,\"end\":14.24},{\"word\":\"bugbears\",\"start\":21.53,\"end\":22.07}]"}
{"utterance_id": "6319-64726-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0017.flac", "task_type": "lexical_detection", "query": "length", "prompt": "When the utterance is complete, report all instances of the word \"length\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.67, "length"]], "response_mode": "offline", "answer": "[{\"word\":\"length\",\"start\":3.31,\"end\":3.67}]"}
{"utterance_id": "6319-64726-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0003.flac", "task_type": "lexical_detection", "query": "undo", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"undo\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.51, 2.93, "undo"]], "response_mode": "offline", "answer": "[{\"word\":\"undo\",\"start\":2.51,\"end\":2.93}]"}
{"utterance_id": "1919-142785-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0048.flac", "task_type": "lexical_detection", "query": "french", "prompt": "Once the audio has finished, identify every instance of the word \"french\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 0.99, "french"]], "response_mode": "offline", "answer": "[{\"word\":\"french\",\"start\":0.52,\"end\":0.99}]"}
{"utterance_id": "1993-147149-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0007.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the entire clip, then report the count sequence for a family relation as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.31, "one"], [2.32, 2.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.89,\"end\":1.31},{\"word\":\"two\",\"start\":2.32,\"end\":2.65}]"}
{"utterance_id": "2078-142845-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0027.flac", "task_type": "lexical_detection", "query": "besides", "prompt": "When the utterance is complete, report all instances of the word \"besides\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.71, "besides"]], "response_mode": "offline", "answer": "[{\"word\":\"besides\",\"start\":5.9,\"end\":6.71}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.26, 1.73, "christmas"], [3.3, 4.06, "new year's day"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":1.26,\"end\":1.73},{\"word\":\"new year's day\",\"start\":3.3,\"end\":4.06}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.47, 6.14, "clergymen"]], "response_mode": "offline", "answer": "[{\"word\":\"clergymen\",\"start\":5.47,\"end\":6.14}]"}
{"utterance_id": "2078-142845-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0031.flac", "task_type": "lexical_detection", "query": "inch", "prompt": "Once the audio has finished, identify every instance of the word \"inch\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.25, 6.89, "inch"]], "response_mode": "offline", "answer": "[{\"word\":\"inch\",\"start\":6.25,\"end\":6.89}]"}
{"utterance_id": "2035-147961-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0015.flac", "task_type": "lexical_detection", "query": "gradually", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"gradually\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.12, "gradually"]], "response_mode": "offline", "answer": "[{\"word\":\"gradually\",\"start\":0.51,\"end\":1.12}]"}
{"utterance_id": "2412-153948-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.35, 5.66, "lamb"]], "response_mode": "offline", "answer": "[{\"word\":\"lamb\",\"start\":5.35,\"end\":5.66}]"}
{"utterance_id": "1272-128104-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0014.flac", "task_type": "lexical_detection", "query": "quilter", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"quilter\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.74, 1.31, "quilter"]], "response_mode": "offline", "answer": "[{\"word\":\"quilter\",\"start\":0.74,\"end\":1.31}]"}
{"utterance_id": "2412-153948-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0002.flac", "task_type": "lexical_detection", "query": "honest", "prompt": "After the audio ends, extract all occurrences of the word \"honest\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.08, "honest"]], "response_mode": "offline", "answer": "[{\"word\":\"honest\",\"start\":1.67,\"end\":2.08}]"}
{"utterance_id": "422-122949-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0020.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.26, "one"], [3.44, 3.62, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.03,\"end\":2.26},{\"word\":\"one\",\"start\":3.44,\"end\":3.62}]"}
{"utterance_id": "6313-76958-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0020.flac", "task_type": "lexical_detection", "query": "asked", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"asked\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.15, 1.59, "asked"]], "response_mode": "offline", "answer": "[{\"word\":\"asked\",\"start\":1.15,\"end\":1.59}]"}
{"utterance_id": "3000-15664-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0014.flac", "task_type": "lexical_detection", "query": "bearing", "prompt": "Process the full audio offline. Find every instance of the word \"bearing\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.17, "bearing"]], "response_mode": "offline", "answer": "[{\"word\":\"bearing\",\"start\":2.74,\"end\":3.17}]"}
{"utterance_id": "174-50561-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.71, "singer"]], "response_mode": "offline", "answer": "[{\"word\":\"singer\",\"start\":1.1,\"end\":1.71}]"}
{"utterance_id": "5694-64029-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0020.flac", "task_type": "lexical_counting", "query": "cease", "prompt": "Listen to the full audio. Count every occurrence of the word \"cease\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.79, 8.1, "one"], [10.33, 10.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.79,\"end\":8.1},{\"word\":\"two\",\"start\":10.33,\"end\":10.65}]"}
{"utterance_id": "5694-64029-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0003.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.02, "regiments"]], "response_mode": "offline", "answer": "[{\"word\":\"regiments\",\"start\":1.61,\"end\":2.02}]"}
{"utterance_id": "2078-142845-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0044.flac", "task_type": "lexical_detection", "query": "rusks", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"rusks\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.32, 2.02, "rusks"]], "response_mode": "offline", "answer": "[{\"word\":\"rusks\",\"start\":1.32,\"end\":2.02}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "When the utterance is complete, report all instances of an organization name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.4, "pall mall gazette"], [8.25, 8.78, "spectator"]], "response_mode": "offline", "answer": "[{\"word\":\"pall mall gazette\",\"start\":5.59,\"end\":6.4},{\"word\":\"spectator\",\"start\":8.25,\"end\":8.78}]"}
{"utterance_id": "3000-15664-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0046.flac", "task_type": "lexical_detection", "query": "doomed", "prompt": "When the utterance is complete, report all instances of the word \"doomed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.22, 3.55, "doomed"]], "response_mode": "offline", "answer": "[{\"word\":\"doomed\",\"start\":3.22,\"end\":3.55}]"}
{"utterance_id": "5536-43363-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0003.flac", "task_type": "lexical_detection", "query": "comers", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"comers\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.7, 8.19, "comers"]], "response_mode": "offline", "answer": "[{\"word\":\"comers\",\"start\":7.7,\"end\":8.19}]"}
{"utterance_id": "2035-147960-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0003.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "You will hear a complete utterance. When it ends, list every instance of a weather term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 1.96, "frost"]], "response_mode": "offline", "answer": "[{\"word\":\"frost\",\"start\":1.59,\"end\":1.96}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.64, 8.03, "one"], [12.85, 13.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.64,\"end\":8.03},{\"word\":\"two\",\"start\":12.85,\"end\":13.38}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the full audio. Count every occurrence of a family relation. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.88, 4.6, "one"], [8.89, 9.3, "two"], [16.58, 17.03, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.88,\"end\":4.6},{\"word\":\"two\",\"start\":8.89,\"end\":9.3},{\"word\":\"three\",\"start\":16.58,\"end\":17.03}]"}
{"utterance_id": "7850-281318-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0007.flac", "task_type": "lexical_detection", "query": "hopping", "prompt": "When the utterance is complete, report all instances of the word \"hopping\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 1.99, "hopping"]], "response_mode": "offline", "answer": "[{\"word\":\"hopping\",\"start\":1.59,\"end\":1.99}]"}
{"utterance_id": "6345-93302-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0002.flac", "task_type": "lexical_detection", "query": "weekly", "prompt": "After listening to the whole recording, output every instance of the word \"weekly\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.34, 6.68, "weekly"]], "response_mode": "offline", "answer": "[{\"word\":\"weekly\",\"start\":6.34,\"end\":6.68}]"}
{"utterance_id": "3170-137482-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0038.flac", "task_type": "lexical_detection", "query": "hermit", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hermit\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.84, 5.26, "hermit"]], "response_mode": "offline", "answer": "[{\"word\":\"hermit\",\"start\":4.84,\"end\":5.26}]"}
{"utterance_id": "3576-138058-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0005.flac", "task_type": "lexical_detection", "query": "pressing", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pressing\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.61, 11.01, "pressing"]], "response_mode": "offline", "answer": "[{\"word\":\"pressing\",\"start\":10.61,\"end\":11.01}]"}
{"utterance_id": "6319-64726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0007.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.73, 9.22, "ogre"]], "response_mode": "offline", "answer": "[{\"word\":\"ogre\",\"start\":8.73,\"end\":9.22}]"}
{"utterance_id": "6345-64257-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0016.flac", "task_type": "lexical_detection", "query": "because", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"because\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.05, 4.36, "because"]], "response_mode": "offline", "answer": "[{\"word\":\"because\",\"start\":4.05,\"end\":4.36}]"}
{"utterance_id": "2277-149897-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0009.flac", "task_type": "lexical_detection", "query": "hastened", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"hastened\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.84, 1.3, "hastened"]], "response_mode": "offline", "answer": "[{\"word\":\"hastened\",\"start\":0.84,\"end\":1.3}]"}
{"utterance_id": "3853-163249-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0018.flac", "task_type": "lexical_detection", "query": "sweetheart", "prompt": "Process the full audio offline. Find every instance of the word \"sweetheart\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.78, "sweetheart"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetheart\",\"start\":2.25,\"end\":2.78}]"}
{"utterance_id": "2277-149896-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0030.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Listen carefully to the entire clip, then report every instance of a money or currency term as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.35, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":3.97,\"end\":4.35}]"}
{"utterance_id": "6313-66125-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0008.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.94, 2.35, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":1.94,\"end\":2.35}]"}
{"utterance_id": "3752-4944-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0047.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 1.28, "captain burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"captain burgess\",\"start\":0.46,\"end\":1.28}]"}
{"utterance_id": "2035-147960-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0013.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.7, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.36,\"end\":1.7}]"}
{"utterance_id": "777-126732-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0058.flac", "task_type": "lexical_detection", "query": "secret", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"secret\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.63, 4.04, "secret"]], "response_mode": "offline", "answer": "[{\"word\":\"secret\",\"start\":3.63,\"end\":4.04}]"}
{"utterance_id": "251-137823-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0025.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.65, 3.64, "animal"]], "response_mode": "offline", "answer": "[{\"word\":\"animal\",\"start\":2.65,\"end\":3.64}]"}
{"utterance_id": "7976-110523-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0020.flac", "task_type": "lexical_detection", "query": "precious", "prompt": "Process the full audio offline. Find every instance of the word \"precious\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.17, 7.58, "precious"]], "response_mode": "offline", "answer": "[{\"word\":\"precious\",\"start\":7.17,\"end\":7.58}]"}
{"utterance_id": "3000-15664-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0032.flac", "task_type": "lexical_detection", "query": "wake", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"wake\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.15, 10.59, "wake"]], "response_mode": "offline", "answer": "[{\"word\":\"wake\",\"start\":10.15,\"end\":10.59}]"}
{"utterance_id": "6345-93306-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0013.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[4.7, 5.13, "maid"]], "response_mode": "offline", "answer": "[{\"word\":\"maid\",\"start\":4.7,\"end\":5.13}]"}
{"utterance_id": "8842-302203-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0009.flac", "task_type": "lexical_detection", "query": "joan", "prompt": "After the audio ends, extract all occurrences of the word \"joan\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.8, "joan"]], "response_mode": "offline", "answer": "[{\"word\":\"joan\",\"start\":1.37,\"end\":1.8}]"}
{"utterance_id": "2277-149897-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0028.flac", "task_type": "lexical_detection", "query": "explain", "prompt": "After listening to the whole recording, output every instance of the word \"explain\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.97, "explain"]], "response_mode": "offline", "answer": "[{\"word\":\"explain\",\"start\":0.54,\"end\":0.97}]"}
{"utterance_id": "5895-34629-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.58, 3.2, "shakespeare"]], "response_mode": "offline", "answer": "[{\"word\":\"shakespeare\",\"start\":2.58,\"end\":3.2}]"}
{"utterance_id": "422-122949-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0034.flac", "task_type": "lexical_detection", "query": "dainty", "prompt": "After the audio ends, extract all occurrences of the word \"dainty\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.54, 4.97, "dainty"]], "response_mode": "offline", "answer": "[{\"word\":\"dainty\",\"start\":4.54,\"end\":4.97}]"}
{"utterance_id": "5895-34615-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0020.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.59, 0.86, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":0.59,\"end\":0.86}]"}
{"utterance_id": "5536-43358-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0008.flac", "task_type": "lexical_detection", "query": "solemn", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"solemn\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.13, "solemn"]], "response_mode": "offline", "answer": "[{\"word\":\"solemn\",\"start\":0.7,\"end\":1.13}]"}
{"utterance_id": "2035-147961-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.17, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.0,\"end\":2.17}]"}
{"utterance_id": "2428-83699-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0004.flac", "task_type": "lexical_detection", "query": "trifle", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"trifle\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.42, "trifle"]], "response_mode": "offline", "answer": "[{\"word\":\"trifle\",\"start\":1.02,\"end\":1.42}]"}
{"utterance_id": "5536-43358-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0007.flac", "task_type": "lexical_detection", "query": "knowing", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"knowing\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.91, "knowing"]], "response_mode": "offline", "answer": "[{\"word\":\"knowing\",\"start\":0.54,\"end\":0.91}]"}
{"utterance_id": "8842-302201-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0009.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.51, 0.81, "second"], [2.65, 2.92, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.51,\"end\":0.81},{\"word\":\"third\",\"start\":2.65,\"end\":2.92}]"}
{"utterance_id": "3853-163249-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.35, "lisha"]], "response_mode": "offline", "answer": "[{\"word\":\"lisha\",\"start\":0.86,\"end\":1.35}]"}
{"utterance_id": "6241-66616-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0019.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.12, 3.38, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":3.12,\"end\":3.38}]"}
{"utterance_id": "174-168635-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0006.flac", "task_type": "lexical_counting", "query": "felt", "prompt": "After the audio ends, tally every instance of the word \"felt\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.75, 1.17, "one"], [2.42, 2.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.75,\"end\":1.17},{\"word\":\"two\",\"start\":2.42,\"end\":2.77}]"}
{"utterance_id": "6313-76958-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0018.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.56, 4.86, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":4.56,\"end\":4.86}]"}
{"utterance_id": "8297-275155-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0032.flac", "task_type": "lexical_detection", "query": "doubly", "prompt": "Once the audio has finished, identify every instance of the word \"doubly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.44, 3.81, "doubly"]], "response_mode": "offline", "answer": "[{\"word\":\"doubly\",\"start\":3.44,\"end\":3.81}]"}
{"utterance_id": "6295-64301-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0003.flac", "task_type": "lexical_detection", "query": "pressed", "prompt": "Process the full audio offline. Find every instance of the word \"pressed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 0.96, "pressed"]], "response_mode": "offline", "answer": "[{\"word\":\"pressed\",\"start\":0.65,\"end\":0.96}]"}
{"utterance_id": "652-130726-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.63, 2.03, "charlie's"]], "response_mode": "offline", "answer": "[{\"word\":\"charlie's\",\"start\":1.63,\"end\":2.03}]"}
{"utterance_id": "2412-153948-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.96, 12.46, "eight hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"eight hundred\",\"start\":11.96,\"end\":12.46}]"}
{"utterance_id": "3081-166546-0082", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0082.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.32, 5.05, "miss clarke"]], "response_mode": "offline", "answer": "[{\"word\":\"miss clarke\",\"start\":4.32,\"end\":5.05}]"}
{"utterance_id": "2086-149220-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.68, 3.13, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":2.68,\"end\":3.13}]"}
{"utterance_id": "6313-66125-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0003.flac", "task_type": "lexical_detection", "query": "warned", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"warned\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.27, 2.58, "warned"]], "response_mode": "offline", "answer": "[{\"word\":\"warned\",\"start\":2.27,\"end\":2.58}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the full audio. After it finishes, find every instance of a weather term. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.45, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":1.04,\"end\":1.45}]"}
{"utterance_id": "3170-137482-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0023.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.06, 10.38, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":10.06,\"end\":10.38}]"}
{"utterance_id": "1993-147966-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0006.flac", "task_type": "semantic_detection", "query": "date", "prompt": "You will hear a complete utterance. When it ends, list every instance of a date you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.1, "birthday"]], "response_mode": "offline", "answer": "[{\"word\":\"birthday\",\"start\":0.77,\"end\":1.1}]"}
{"utterance_id": "3752-4944-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0033.flac", "task_type": "lexical_detection", "query": "evidence", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"evidence\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 2.15, "evidence"]], "response_mode": "offline", "answer": "[{\"word\":\"evidence\",\"start\":1.62,\"end\":2.15}]"}
{"utterance_id": "2803-161169-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0000.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a family relation (word, start, end). If none, reply [].", "labels": [[1.18, 1.75, "one"], [3.42, 4.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.18,\"end\":1.75},{\"word\":\"two\",\"start\":3.42,\"end\":4.13}]"}
{"utterance_id": "2035-147961-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.73, "fist"]], "response_mode": "offline", "answer": "[{\"word\":\"fist\",\"start\":2.26,\"end\":2.73}]"}
{"utterance_id": "7850-111771-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0000.flac", "task_type": "lexical_detection", "query": "influence", "prompt": "After listening to the whole recording, output every instance of the word \"influence\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.49, "influence"]], "response_mode": "offline", "answer": "[{\"word\":\"influence\",\"start\":0.75,\"end\":1.49}]"}
{"utterance_id": "3081-166546-0071", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0071.flac", "task_type": "lexical_detection", "query": "figure", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"figure\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.33, 1.75, "figure"]], "response_mode": "offline", "answer": "[{\"word\":\"figure\",\"start\":1.33,\"end\":1.75}]"}
{"utterance_id": "3576-138058-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0033.flac", "task_type": "lexical_counting", "query": "along", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"along\" (word, start, end). If none, reply [].", "labels": [[7.3, 7.7, "one"], [15.68, 16.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.3,\"end\":7.7},{\"word\":\"two\",\"start\":15.68,\"end\":16.0}]"}
{"utterance_id": "6295-64301-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.83, "heart"], [2.79, 3.47, "throat"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":0.45,\"end\":0.83},{\"word\":\"throat\",\"start\":2.79,\"end\":3.47}]"}
{"utterance_id": "174-50561-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0011.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.69, 5.19, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":4.69,\"end\":5.19}]"}
{"utterance_id": "2803-161169-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0007.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "After the audio ends, tally every instance of a plant name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.82, 4.83, "one"], [10.13, 10.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.82,\"end\":4.83},{\"word\":\"two\",\"start\":10.13,\"end\":10.91}]"}
{"utterance_id": "3081-166546-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0044.flac", "task_type": "lexical_detection", "query": "detective", "prompt": "After listening to the whole recording, output every instance of the word \"detective\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.41, "detective"]], "response_mode": "offline", "answer": "[{\"word\":\"detective\",\"start\":1.75,\"end\":2.41}]"}
{"utterance_id": "6241-61946-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0018.flac", "task_type": "lexical_detection", "query": "theirs", "prompt": "Once the audio has finished, identify every instance of the word \"theirs\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.04, 3.53, "theirs"]], "response_mode": "offline", "answer": "[{\"word\":\"theirs\",\"start\":3.04,\"end\":3.53}]"}
{"utterance_id": "777-126732-0070", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0070.flac", "task_type": "lexical_detection", "query": "eating", "prompt": "After listening to the whole recording, output every instance of the word \"eating\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 2.99, "eating"]], "response_mode": "offline", "answer": "[{\"word\":\"eating\",\"start\":2.65,\"end\":2.99}]"}
{"utterance_id": "1919-142785-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0013.flac", "task_type": "lexical_detection", "query": "jug", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"jug\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.26, 5.57, "jug"]], "response_mode": "offline", "answer": "[{\"word\":\"jug\",\"start\":5.26,\"end\":5.57}]"}
{"utterance_id": "2412-153954-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0002.flac", "task_type": "lexical_detection", "query": "streets", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"streets\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.72, "streets"]], "response_mode": "offline", "answer": "[{\"word\":\"streets\",\"start\":0.41,\"end\":0.72}]"}
{"utterance_id": "1673-143397-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.65, "patriarch"]], "response_mode": "offline", "answer": "[{\"word\":\"patriarch\",\"start\":3.97,\"end\":4.65}]"}
{"utterance_id": "3576-138058-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0024.flac", "task_type": "lexical_detection", "query": "convinced", "prompt": "After the audio ends, extract all occurrences of the word \"convinced\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.36, "convinced"]], "response_mode": "offline", "answer": "[{\"word\":\"convinced\",\"start\":2.87,\"end\":3.36}]"}
{"utterance_id": "3752-4943-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0002.flac", "task_type": "lexical_detection", "query": "dismay", "prompt": "After listening to the whole recording, output every instance of the word \"dismay\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 3.93, "dismay"]], "response_mode": "offline", "answer": "[{\"word\":\"dismay\",\"start\":3.48,\"end\":3.93}]"}
{"utterance_id": "3853-163249-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0030.flac", "task_type": "lexical_detection", "query": "sundry", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sundry\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.18, 4.58, "sundry"]], "response_mode": "offline", "answer": "[{\"word\":\"sundry\",\"start\":4.18,\"end\":4.58}]"}
{"utterance_id": "2428-83699-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0002.flac", "task_type": "lexical_detection", "query": "vague", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"vague\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.71, 4.24, "vague"]], "response_mode": "offline", "answer": "[{\"word\":\"vague\",\"start\":3.71,\"end\":4.24}]"}
{"utterance_id": "84-121550-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0030.flac", "task_type": "lexical_detection", "query": "vapours", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"vapours\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.53, 5.06, "vapours"]], "response_mode": "offline", "answer": "[{\"word\":\"vapours\",\"start\":4.53,\"end\":5.06}]"}
{"utterance_id": "6313-76958-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 1.16, "stacy brown's"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy brown's\",\"start\":0.28,\"end\":1.16}]"}
{"utterance_id": "5895-34615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 2.23, "mountebank"]], "response_mode": "offline", "answer": "[{\"word\":\"mountebank\",\"start\":1.46,\"end\":2.23}]"}
{"utterance_id": "2086-149214-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0003.flac", "task_type": "lexical_detection", "query": "essential", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"essential\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.77, 4.19, "essential"]], "response_mode": "offline", "answer": "[{\"word\":\"essential\",\"start\":3.77,\"end\":4.19}]"}
{"utterance_id": "1673-143396-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0012.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.25, "nazareth"]], "response_mode": "offline", "answer": "[{\"word\":\"nazareth\",\"start\":1.7,\"end\":2.25}]"}
{"utterance_id": "1462-170142-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0018.flac", "task_type": "lexical_detection", "query": "deception", "prompt": "When the utterance is complete, report all instances of the word \"deception\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.42, "deception"]], "response_mode": "offline", "answer": "[{\"word\":\"deception\",\"start\":0.73,\"end\":1.42}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_detection", "query": "color", "prompt": "When the utterance is complete, report all instances of a color name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.18, 8.39, "ruby"]], "response_mode": "offline", "answer": "[{\"word\":\"ruby\",\"start\":8.18,\"end\":8.39}]"}
{"utterance_id": "2803-154328-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.06, "glenarvan"]], "response_mode": "offline", "answer": "[{\"word\":\"glenarvan\",\"start\":0.51,\"end\":1.06}]"}
{"utterance_id": "5694-64038-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0011.flac", "task_type": "lexical_detection", "query": "get", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"get\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 0.93, "get"]], "response_mode": "offline", "answer": "[{\"word\":\"get\",\"start\":0.78,\"end\":0.93}]"}
{"utterance_id": "3576-138058-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.26, 1.19, "dulcinea"]], "response_mode": "offline", "answer": "[{\"word\":\"dulcinea\",\"start\":0.26,\"end\":1.19}]"}
{"utterance_id": "6313-66129-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 1.61, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":1.32,\"end\":1.61}]"}
{"utterance_id": "8842-302201-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0006.flac", "task_type": "lexical_counting", "query": "eyes", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"eyes\" (word, start, end). If none, reply [].", "labels": [[1.21, 1.5, "one"], [5.93, 6.23, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.21,\"end\":1.5},{\"word\":\"two\",\"start\":5.93,\"end\":6.23}]"}
{"utterance_id": "6319-275224-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0000.flac", "task_type": "lexical_detection", "query": "dear", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"dear\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.18, 3.58, "dear"]], "response_mode": "offline", "answer": "[{\"word\":\"dear\",\"start\":3.18,\"end\":3.58}]"}
{"utterance_id": "2035-152373-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0013.flac", "task_type": "semantic_counting", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a city name (word, start, end). If none, reply [].", "labels": [[4.98, 5.45, "one"], [6.82, 7.33, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.98,\"end\":5.45},{\"word\":\"two\",\"start\":6.82,\"end\":7.33}]"}
{"utterance_id": "2035-147961-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0021.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.16, 9.62, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":9.16,\"end\":9.62}]"}
{"utterance_id": "6313-76958-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0005.flac", "task_type": "lexical_detection", "query": "gathered", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"gathered\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.99, 3.36, "gathered"]], "response_mode": "offline", "answer": "[{\"word\":\"gathered\",\"start\":2.99,\"end\":3.36}]"}
{"utterance_id": "6313-66125-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.48, 2.82, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":2.48,\"end\":2.82}]"}
{"utterance_id": "2035-152373-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.31, "brothers"]], "response_mode": "offline", "answer": "[{\"word\":\"brothers\",\"start\":1.91,\"end\":2.31}]"}
{"utterance_id": "1919-142785-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0047.flac", "task_type": "lexical_detection", "query": "herbs", "prompt": "When the utterance is complete, report all instances of the word \"herbs\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.4, "herbs"]], "response_mode": "offline", "answer": "[{\"word\":\"herbs\",\"start\":1.03,\"end\":1.4}]"}
{"utterance_id": "7850-286674-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0009.flac", "task_type": "lexical_detection", "query": "whenever", "prompt": "Process the full audio offline. Find every instance of the word \"whenever\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.15, "whenever"]], "response_mode": "offline", "answer": "[{\"word\":\"whenever\",\"start\":0.85,\"end\":1.15}]"}
{"utterance_id": "1462-170138-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0002.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.06, 2.51, "tonight"]], "response_mode": "offline", "answer": "[{\"word\":\"tonight\",\"start\":2.06,\"end\":2.51}]"}
{"utterance_id": "777-126732-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[11.03, 11.38, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":11.03,\"end\":11.38}]"}
{"utterance_id": "2428-83699-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0042.flac", "task_type": "lexical_detection", "query": "cellar", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cellar\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.79, "cellar"]], "response_mode": "offline", "answer": "[{\"word\":\"cellar\",\"start\":1.26,\"end\":1.79}]"}
{"utterance_id": "5694-64029-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0021.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.47, 2.89, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":2.47,\"end\":2.89}]"}
{"utterance_id": "3081-166546-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0018.flac", "task_type": "lexical_detection", "query": "want", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"want\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.02, 1.67, "want"]], "response_mode": "offline", "answer": "[{\"word\":\"want\",\"start\":1.02,\"end\":1.67}]"}
{"utterance_id": "6313-76958-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.67, "stacy"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":0.28,\"end\":0.67}]"}
{"utterance_id": "6345-93302-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0017.flac", "task_type": "lexical_detection", "query": "jealousy", "prompt": "After the audio ends, extract all occurrences of the word \"jealousy\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.25, 3.86, "jealousy"]], "response_mode": "offline", "answer": "[{\"word\":\"jealousy\",\"start\":3.25,\"end\":3.86}]"}
{"utterance_id": "2078-142845-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0047.flac", "task_type": "lexical_detection", "query": "cuisine", "prompt": "Process the full audio offline. Find every instance of the word \"cuisine\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.51, 8.24, "cuisine"]], "response_mode": "offline", "answer": "[{\"word\":\"cuisine\",\"start\":7.51,\"end\":8.24}]"}
{"utterance_id": "2035-147960-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0012.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.34, 3.88, "snake"]], "response_mode": "offline", "answer": "[{\"word\":\"snake\",\"start\":3.34,\"end\":3.88}]"}
{"utterance_id": "1272-141231-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0015.flac", "task_type": "lexical_detection", "query": "brion", "prompt": "After listening to the whole recording, output every instance of the word \"brion\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.62, 4.01, "brion"]], "response_mode": "offline", "answer": "[{\"word\":\"brion\",\"start\":3.62,\"end\":4.01}]"}
{"utterance_id": "1988-24833-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0003.flac", "task_type": "lexical_detection", "query": "fixing", "prompt": "After the audio ends, extract all occurrences of the word \"fixing\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.8, "fixing"]], "response_mode": "offline", "answer": "[{\"word\":\"fixing\",\"start\":2.35,\"end\":2.8}]"}
{"utterance_id": "1919-142785-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0030.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.47, 3.05, "ingredients"]], "response_mode": "offline", "answer": "[{\"word\":\"ingredients\",\"start\":2.47,\"end\":3.05}]"}
{"utterance_id": "6241-66616-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0014.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.38, 3.85, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":3.38,\"end\":3.85}]"}
{"utterance_id": "1993-147966-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0006.flac", "task_type": "lexical_detection", "query": "mistake", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"mistake\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 3.02, "mistake"]], "response_mode": "offline", "answer": "[{\"word\":\"mistake\",\"start\":2.4,\"end\":3.02}]"}
{"utterance_id": "2428-83699-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[2.77, 3.43, "christmases"]], "response_mode": "offline", "answer": "[{\"word\":\"christmases\",\"start\":2.77,\"end\":3.43}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a color name (word, start, end). If none, reply [].", "labels": [[5.94, 6.15, "one"], [9.99, 10.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.94,\"end\":6.15},{\"word\":\"two\",\"start\":9.99,\"end\":10.35}]"}
{"utterance_id": "3170-137482-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0026.flac", "task_type": "lexical_detection", "query": "deal", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"deal\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.36, "deal"]], "response_mode": "offline", "answer": "[{\"word\":\"deal\",\"start\":1.88,\"end\":2.36}]"}
{"utterance_id": "3752-4944-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0006.flac", "task_type": "lexical_detection", "query": "burgess", "prompt": "After listening to the whole recording, output every instance of the word \"burgess\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.58, 2.12, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":1.58,\"end\":2.12}]"}
{"utterance_id": "2035-147961-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0018.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.54, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":3.24,\"end\":3.54}]"}
{"utterance_id": "5536-43359-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0007.flac", "task_type": "lexical_counting", "query": "known", "prompt": "Listen to the entire clip, then report the count sequence for the word \"known\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.09, 6.34, "one"], [14.01, 14.28, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.09,\"end\":6.34},{\"word\":\"two\",\"start\":14.01,\"end\":14.28}]"}
{"utterance_id": "6241-61943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0025.flac", "task_type": "lexical_counting", "query": "grass", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"grass\" with word/start/end. If none, reply [].", "labels": [[2.29, 2.74, "one"], [4.23, 4.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.29,\"end\":2.74},{\"word\":\"two\",\"start\":4.23,\"end\":4.57}]"}
{"utterance_id": "2078-142845-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0011.flac", "task_type": "lexical_detection", "query": "grow", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"grow\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.62, 1.11, "grow"]], "response_mode": "offline", "answer": "[{\"word\":\"grow\",\"start\":0.62,\"end\":1.11}]"}
{"utterance_id": "2035-147961-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0036.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.4, "head"], [3.62, 4.13, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":2.17,\"end\":2.4},{\"word\":\"eyes\",\"start\":3.62,\"end\":4.13}]"}
{"utterance_id": "6313-66125-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.51, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.2,\"end\":1.51}]"}
{"utterance_id": "1988-24833-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.0, 2.37, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":2.0,\"end\":2.37}]"}
{"utterance_id": "3536-8226-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0010.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the entire clip, then report the count sequence for a family relation as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 3.01, "one"], [8.87, 9.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.62,\"end\":3.01},{\"word\":\"two\",\"start\":8.87,\"end\":9.48}]"}
{"utterance_id": "1462-170142-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0041.flac", "task_type": "lexical_detection", "query": "tired", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"tired\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.4, 3.82, "tired"]], "response_mode": "offline", "answer": "[{\"word\":\"tired\",\"start\":3.4,\"end\":3.82}]"}
{"utterance_id": "2086-149220-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0006.flac", "task_type": "lexical_detection", "query": "fountain", "prompt": "After listening to the whole recording, output every instance of the word \"fountain\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.56, "fountain"]], "response_mode": "offline", "answer": "[{\"word\":\"fountain\",\"start\":0.99,\"end\":1.56}]"}
{"utterance_id": "777-126732-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0009.flac", "task_type": "lexical_detection", "query": "terrorist", "prompt": "After listening to the whole recording, output every instance of the word \"terrorist\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.4, "terrorist"]], "response_mode": "offline", "answer": "[{\"word\":\"terrorist\",\"start\":0.93,\"end\":1.4}]"}
{"utterance_id": "2428-83705-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0030.flac", "task_type": "lexical_detection", "query": "inexplicable", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"inexplicable\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.85, 8.83, "inexplicable"]], "response_mode": "offline", "answer": "[{\"word\":\"inexplicable\",\"start\":7.85,\"end\":8.83}]"}
{"utterance_id": "6295-64301-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0026.flac", "task_type": "lexical_detection", "query": "throne", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"throne\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.28, 4.8, "throne"]], "response_mode": "offline", "answer": "[{\"word\":\"throne\",\"start\":4.28,\"end\":4.8}]"}
{"utterance_id": "422-122949-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0009.flac", "task_type": "lexical_detection", "query": "cases", "prompt": "After the audio ends, extract all occurrences of the word \"cases\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.18, 17.64, "cases"]], "response_mode": "offline", "answer": "[{\"word\":\"cases\",\"start\":17.18,\"end\":17.64}]"}
{"utterance_id": "7976-110523-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.36, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":0.86,\"end\":1.36}]"}
{"utterance_id": "2035-147960-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0004.flac", "task_type": "lexical_detection", "query": "gravel", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"gravel\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.38, 1.79, "gravel"]], "response_mode": "offline", "answer": "[{\"word\":\"gravel\",\"start\":1.38,\"end\":1.79}]"}
{"utterance_id": "6295-244435-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0013.flac", "task_type": "lexical_detection", "query": "emotion", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"emotion\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.92, 2.49, "emotion"]], "response_mode": "offline", "answer": "[{\"word\":\"emotion\",\"start\":1.92,\"end\":2.49}]"}
{"utterance_id": "251-136532-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0021.flac", "task_type": "lexical_detection", "query": "believe", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"believe\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.85, "believe"]], "response_mode": "offline", "answer": "[{\"word\":\"believe\",\"start\":1.35,\"end\":1.85}]"}
{"utterance_id": "2277-149897-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0001.flac", "task_type": "lexical_detection", "query": "hardly", "prompt": "After listening to the whole recording, output every instance of the word \"hardly\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.86, "hardly"]], "response_mode": "offline", "answer": "[{\"word\":\"hardly\",\"start\":0.55,\"end\":0.86}]"}
{"utterance_id": "1988-24833-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0022.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.88, 1.11, "one"], [2.28, 2.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.88,\"end\":1.11},{\"word\":\"two\",\"start\":2.28,\"end\":2.66}]"}
{"utterance_id": "2902-9008-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0005.flac", "task_type": "lexical_counting", "query": "off", "prompt": "Once the audio has finished, list each occurrence of the word \"off\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.29, 1.54, "one"], [3.86, 4.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.29,\"end\":1.54},{\"word\":\"two\",\"start\":3.86,\"end\":4.06}]"}
{"utterance_id": "1462-170142-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0029.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.19, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":0.73,\"end\":1.19}]"}
{"utterance_id": "6295-244435-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0012.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Process the full audio offline. Find every instance of a state or province name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.74, 4.36, "south"]], "response_mode": "offline", "answer": "[{\"word\":\"south\",\"start\":3.74,\"end\":4.36}]"}
{"utterance_id": "3081-166546-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0032.flac", "task_type": "lexical_detection", "query": "it's", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"it's\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.41, 0.69, "it's"]], "response_mode": "offline", "answer": "[{\"word\":\"it's\",\"start\":0.41,\"end\":0.69}]"}
{"utterance_id": "8297-275155-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0025.flac", "task_type": "lexical_detection", "query": "notice", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"notice\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.66, "notice"]], "response_mode": "offline", "answer": "[{\"word\":\"notice\",\"start\":2.05,\"end\":2.66}]"}
{"utterance_id": "3536-23268-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0006.flac", "task_type": "lexical_detection", "query": "promise", "prompt": "Once the audio has finished, identify every instance of the word \"promise\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.74, 4.17, "promise"]], "response_mode": "offline", "answer": "[{\"word\":\"promise\",\"start\":3.74,\"end\":4.17}]"}
{"utterance_id": "6241-66616-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0004.flac", "task_type": "lexical_detection", "query": "newsome's", "prompt": "After the audio ends, extract all occurrences of the word \"newsome's\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.64, "newsome's"]], "response_mode": "offline", "answer": "[{\"word\":\"newsome's\",\"start\":2.23,\"end\":2.64}]"}
{"utterance_id": "3536-8226-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0024.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.64, 5.18, "child"]], "response_mode": "offline", "answer": "[{\"word\":\"child\",\"start\":4.64,\"end\":5.18}]"}
{"utterance_id": "2035-147961-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0037.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.32, "cow"]], "response_mode": "offline", "answer": "[{\"word\":\"cow\",\"start\":2.0,\"end\":2.32}]"}
{"utterance_id": "6345-64257-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0001.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a weather term with fields word, start, end (seconds). If none, reply [].", "labels": [[12.33, 12.7, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":12.33,\"end\":12.7}]"}
{"utterance_id": "84-121550-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0020.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[5.52, 6.04, "august"]], "response_mode": "offline", "answer": "[{\"word\":\"august\",\"start\":5.52,\"end\":6.04}]"}
{"utterance_id": "1993-147149-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0011.flac", "task_type": "lexical_detection", "query": "lifted", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"lifted\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.19, 1.56, "lifted"]], "response_mode": "offline", "answer": "[{\"word\":\"lifted\",\"start\":1.19,\"end\":1.56}]"}
{"utterance_id": "5338-24640-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.1, "mister morton"]], "response_mode": "offline", "answer": "[{\"word\":\"mister morton\",\"start\":0.53,\"end\":1.1}]"}
{"utterance_id": "3081-166546-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0062.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.86, 1.68, "mister slater"]], "response_mode": "offline", "answer": "[{\"word\":\"mister slater\",\"start\":0.86,\"end\":1.68}]"}
{"utterance_id": "1988-148538-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0009.flac", "task_type": "lexical_detection", "query": "american", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"american\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.97, 1.5, "american"]], "response_mode": "offline", "answer": "[{\"word\":\"american\",\"start\":0.97,\"end\":1.5}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "lexical_counting", "query": "northern", "prompt": "When the utterance is complete, report how the word \"northern\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.33, 17.72, "one"], [31.1, 31.44, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":17.33,\"end\":17.72},{\"word\":\"two\",\"start\":31.1,\"end\":31.44}]"}
{"utterance_id": "6295-244435-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0015.flac", "task_type": "lexical_detection", "query": "wisely", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wisely\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 2.04, "wisely"]], "response_mode": "offline", "answer": "[{\"word\":\"wisely\",\"start\":1.27,\"end\":2.04}]"}
{"utterance_id": "3536-23268-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0028.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.42, 6.72, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":6.42,\"end\":6.72}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 3.15, "kaliko"], [6.83, 7.41, "ruggedo s"], [11.26, 11.66, "ruggedo"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":2.62,\"end\":3.15},{\"word\":\"ruggedo s\",\"start\":6.83,\"end\":7.41},{\"word\":\"ruggedo\",\"start\":11.26,\"end\":11.66}]"}
{"utterance_id": "1988-147956-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0026.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Once the audio has finished, list each occurrence of a plant name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.28, 1.79, "one"], [5.31, 5.61, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.28,\"end\":1.79},{\"word\":\"two\",\"start\":5.31,\"end\":5.61}]"}
{"utterance_id": "7976-105575-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0027.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.38, 4.84, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":4.38,\"end\":4.84}]"}
{"utterance_id": "6241-61946-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0014.flac", "task_type": "lexical_detection", "query": "hamlet", "prompt": "Process the full audio offline. Find every instance of the word \"hamlet\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.98, "hamlet"]], "response_mode": "offline", "answer": "[{\"word\":\"hamlet\",\"start\":4.45,\"end\":4.98}]"}
{"utterance_id": "6313-66129-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.82, 5.17, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":4.82,\"end\":5.17}]"}
{"utterance_id": "7976-110124-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.68, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"father's\",\"start\":1.28,\"end\":1.68}]"}
{"utterance_id": "6345-93306-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[15.55, 16.09, "housemaid"]], "response_mode": "offline", "answer": "[{\"word\":\"housemaid\",\"start\":15.55,\"end\":16.09}]"}
{"utterance_id": "7850-281318-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 3.11, "feathers"]], "response_mode": "offline", "answer": "[{\"word\":\"feathers\",\"start\":2.5,\"end\":3.11}]"}
{"utterance_id": "2412-153954-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.29, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.11,\"end\":1.29}]"}
{"utterance_id": "5694-64038-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0022.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.51, "soldiers"]], "response_mode": "offline", "answer": "[{\"word\":\"soldiers\",\"start\":1.01,\"end\":1.51}]"}
{"utterance_id": "6313-66129-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0004.flac", "task_type": "lexical_detection", "query": "announced", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"announced\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.82, 2.18, "announced"]], "response_mode": "offline", "answer": "[{\"word\":\"announced\",\"start\":1.82,\"end\":2.18}]"}
{"utterance_id": "2086-149214-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.62, 0.93, "author"]], "response_mode": "offline", "answer": "[{\"word\":\"author\",\"start\":0.62,\"end\":0.93}]"}
{"utterance_id": "7976-105575-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0014.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the full audio. After it finishes, find every instance of a time of day. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.24, "evening"], [6.41, 7.07, "midnight"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":0.84,\"end\":1.24},{\"word\":\"midnight\",\"start\":6.41,\"end\":7.07}]"}
{"utterance_id": "777-126732-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0042.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.17, "backs"]], "response_mode": "offline", "answer": "[{\"word\":\"backs\",\"start\":1.79,\"end\":2.17}]"}
{"utterance_id": "2428-83705-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0023.flac", "task_type": "lexical_detection", "query": "mamma", "prompt": "Process the full audio offline. Find every instance of the word \"mamma\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.39, "mamma"]], "response_mode": "offline", "answer": "[{\"word\":\"mamma\",\"start\":1.03,\"end\":1.39}]"}
{"utterance_id": "2078-142845-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0030.flac", "task_type": "lexical_counting", "query": "bread", "prompt": "Process the full audio offline. For every occurrence of the word \"bread\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.92, "one"], [6.43, 6.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.54,\"end\":1.92},{\"word\":\"two\",\"start\":6.43,\"end\":6.89}]"}
{"utterance_id": "422-122949-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.32, 15.72, "limbs"]], "response_mode": "offline", "answer": "[{\"word\":\"limbs\",\"start\":15.32,\"end\":15.72}]"}
{"utterance_id": "6295-64301-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0017.flac", "task_type": "lexical_detection", "query": "overcome", "prompt": "After listening to the whole recording, output every instance of the word \"overcome\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.91, "overcome"]], "response_mode": "offline", "answer": "[{\"word\":\"overcome\",\"start\":1.18,\"end\":1.91}]"}
{"utterance_id": "3576-138058-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0014.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.55, 6.16, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":5.55,\"end\":6.16}]"}
{"utterance_id": "5694-64029-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0002.flac", "task_type": "lexical_detection", "query": "murfreesboro", "prompt": "After the audio ends, extract all occurrences of the word \"murfreesboro\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.5, "murfreesboro"]], "response_mode": "offline", "answer": "[{\"word\":\"murfreesboro\",\"start\":1.64,\"end\":2.5}]"}
{"utterance_id": "2277-149874-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.46, 0.76, "minnie"]], "response_mode": "offline", "answer": "[{\"word\":\"minnie\",\"start\":0.46,\"end\":0.76}]"}
{"utterance_id": "5536-43363-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0013.flac", "task_type": "lexical_detection", "query": "medicine", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"medicine\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.31, 1.75, "medicine"]], "response_mode": "offline", "answer": "[{\"word\":\"medicine\",\"start\":1.31,\"end\":1.75}]"}
{"utterance_id": "2428-83705-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.35, 5.99, "mary ann's"]], "response_mode": "offline", "answer": "[{\"word\":\"mary ann's\",\"start\":5.35,\"end\":5.99}]"}
{"utterance_id": "777-126732-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0028.flac", "task_type": "lexical_detection", "query": "blind", "prompt": "After the audio ends, extract all occurrences of the word \"blind\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.3, 9.79, "blind"]], "response_mode": "offline", "answer": "[{\"word\":\"blind\",\"start\":9.3,\"end\":9.79}]"}
{"utterance_id": "6295-244435-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0037.flac", "task_type": "lexical_detection", "query": "got", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"got\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.97, "got"]], "response_mode": "offline", "answer": "[{\"word\":\"got\",\"start\":0.67,\"end\":0.97}]"}
{"utterance_id": "84-121550-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0028.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.8, 5.21, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.8,\"end\":5.21}]"}
{"utterance_id": "1462-170145-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 2.14, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":1.51,\"end\":2.14}]"}
{"utterance_id": "5694-64038-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0022.flac", "task_type": "lexical_detection", "query": "forgotten", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"forgotten\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.94, 3.49, "forgotten"]], "response_mode": "offline", "answer": "[{\"word\":\"forgotten\",\"start\":2.94,\"end\":3.49}]"}
{"utterance_id": "2086-149220-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0037.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.58, 2.19, "hepzibah's"]], "response_mode": "offline", "answer": "[{\"word\":\"hepzibah's\",\"start\":1.58,\"end\":2.19}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "lexical_detection", "query": "satisfy", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"satisfy\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.7, 7.31, "satisfy"]], "response_mode": "offline", "answer": "[{\"word\":\"satisfy\",\"start\":6.7,\"end\":7.31}]"}
{"utterance_id": "1673-143397-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0006.flac", "task_type": "lexical_detection", "query": "asia", "prompt": "Process the full audio offline. Find every instance of the word \"asia\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.11, 2.59, "asia"]], "response_mode": "offline", "answer": "[{\"word\":\"asia\",\"start\":2.11,\"end\":2.59}]"}
{"utterance_id": "5895-34629-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0001.flac", "task_type": "lexical_detection", "query": "delight", "prompt": "After listening to the whole recording, output every instance of the word \"delight\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.94, 2.37, "delight"]], "response_mode": "offline", "answer": "[{\"word\":\"delight\",\"start\":1.94,\"end\":2.37}]"}
{"utterance_id": "3853-163249-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.71, 4.25, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"christie\",\"start\":3.71,\"end\":4.25}]"}
{"utterance_id": "1673-143397-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0003.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After listening to the whole recording, output timed count words for a person's name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.24, 1.64, "one"], [3.32, 3.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.24,\"end\":1.64},{\"word\":\"two\",\"start\":3.32,\"end\":3.77}]"}
{"utterance_id": "3536-23268-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 1.0, "dorriforth"]], "response_mode": "offline", "answer": "[{\"word\":\"dorriforth\",\"start\":0.4,\"end\":1.0}]"}
{"utterance_id": "5338-284437-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0014.flac", "task_type": "lexical_detection", "query": "giant", "prompt": "After listening to the whole recording, output every instance of the word \"giant\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.74, "giant"]], "response_mode": "offline", "answer": "[{\"word\":\"giant\",\"start\":0.98,\"end\":1.74}]"}
{"utterance_id": "2412-153954-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0015.flac", "task_type": "lexical_detection", "query": "suffice", "prompt": "After listening to the whole recording, output every instance of the word \"suffice\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.94, "suffice"]], "response_mode": "offline", "answer": "[{\"word\":\"suffice\",\"start\":0.41,\"end\":0.94}]"}
{"utterance_id": "5895-34629-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.64, 1.91, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":1.64,\"end\":1.91}]"}
{"utterance_id": "2078-142845-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0027.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After the audio ends, tally every instance of a food name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.43, 1.91, "one"], [3.29, 3.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.43,\"end\":1.91},{\"word\":\"two\",\"start\":3.29,\"end\":3.97}]"}
{"utterance_id": "6319-64726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0002.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the full audio. Count every occurrence of a body part. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.43, 5.79, "one"], [9.0, 9.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.43,\"end\":5.79},{\"word\":\"two\",\"start\":9.0,\"end\":9.34}]"}
{"utterance_id": "2035-147961-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0032.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.96, 10.66, "monastery"]], "response_mode": "offline", "answer": "[{\"word\":\"monastery\",\"start\":9.96,\"end\":10.66}]"}
{"utterance_id": "652-130726-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0022.flac", "task_type": "lexical_detection", "query": "listen", "prompt": "Once the audio has finished, identify every instance of the word \"listen\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.09, 9.37, "listen"]], "response_mode": "offline", "answer": "[{\"word\":\"listen\",\"start\":9.09,\"end\":9.37}]"}
{"utterance_id": "5536-43358-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.21, 9.62, "plants"]], "response_mode": "offline", "answer": "[{\"word\":\"plants\",\"start\":9.21,\"end\":9.62}]"}
{"utterance_id": "3000-15664-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0015.flac", "task_type": "lexical_counting", "query": "approaching", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"approaching\" with word/start/end. If none, reply [].", "labels": [[0.47, 0.97, "one"], [11.29, 11.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.47,\"end\":0.97},{\"word\":\"two\",\"start\":11.29,\"end\":11.83}]"}
{"utterance_id": "8842-302201-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0002.flac", "task_type": "lexical_detection", "query": "addressed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"addressed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.89, 4.26, "addressed"]], "response_mode": "offline", "answer": "[{\"word\":\"addressed\",\"start\":3.89,\"end\":4.26}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "After the audio ends, tally every instance of a profession. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.31, 5.87, "one"], [6.96, 7.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.31,\"end\":5.87},{\"word\":\"two\",\"start\":6.96,\"end\":7.38}]"}
{"utterance_id": "2803-161169-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.33, 4.96, "thousands"]], "response_mode": "offline", "answer": "[{\"word\":\"thousands\",\"start\":4.33,\"end\":4.96}]"}
{"utterance_id": "1988-24833-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.26, 0.48, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.26,\"end\":0.48}]"}
{"utterance_id": "6313-66125-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0008.flac", "task_type": "lexical_detection", "query": "gazing", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"gazing\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.68, 3.07, "gazing"]], "response_mode": "offline", "answer": "[{\"word\":\"gazing\",\"start\":2.68,\"end\":3.07}]"}
{"utterance_id": "2277-149896-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0031.flac", "task_type": "lexical_detection", "query": "anyhow", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"anyhow\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.2, 1.73, "anyhow"]], "response_mode": "offline", "answer": "[{\"word\":\"anyhow\",\"start\":1.2,\"end\":1.73}]"}
{"utterance_id": "2078-142845-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0030.flac", "task_type": "semantic_counting", "query": "food", "prompt": "When the utterance is complete, report how a food name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.92, "one"], [4.44, 5.07, "two"], [6.43, 6.89, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.54,\"end\":1.92},{\"word\":\"two\",\"start\":4.44,\"end\":5.07},{\"word\":\"three\",\"start\":6.43,\"end\":6.89}]"}
{"utterance_id": "3536-23268-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.63, 3.89, "fruit"]], "response_mode": "offline", "answer": "[{\"word\":\"fruit\",\"start\":3.63,\"end\":3.89}]"}
{"utterance_id": "1919-142785-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0061.flac", "task_type": "lexical_detection", "query": "rings", "prompt": "Process the full audio offline. Find every instance of the word \"rings\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.88, 4.37, "rings"]], "response_mode": "offline", "answer": "[{\"word\":\"rings\",\"start\":3.88,\"end\":4.37}]"}
{"utterance_id": "2086-149220-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0045.flac", "task_type": "lexical_detection", "query": "mysterious", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mysterious\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.04, 3.6, "mysterious"]], "response_mode": "offline", "answer": "[{\"word\":\"mysterious\",\"start\":3.04,\"end\":3.6}]"}
{"utterance_id": "3576-138058-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0037.flac", "task_type": "lexical_detection", "query": "captains", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"captains\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.13, "captains"]], "response_mode": "offline", "answer": "[{\"word\":\"captains\",\"start\":0.6,\"end\":1.13}]"}
{"utterance_id": "2803-161169-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0009.flac", "task_type": "lexical_detection", "query": "ferns", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ferns\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.19, 1.0, "ferns"]], "response_mode": "offline", "answer": "[{\"word\":\"ferns\",\"start\":0.19,\"end\":1.0}]"}
{"utterance_id": "2803-161169-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0000.flac", "task_type": "lexical_detection", "query": "putting", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"putting\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.02, 6.35, "putting"]], "response_mode": "offline", "answer": "[{\"word\":\"putting\",\"start\":6.02,\"end\":6.35}]"}
{"utterance_id": "6313-66129-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0035.flac", "task_type": "lexical_detection", "query": "instead", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"instead\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.38, "instead"]], "response_mode": "offline", "answer": "[{\"word\":\"instead\",\"start\":1.05,\"end\":1.38}]"}
{"utterance_id": "652-130726-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.55, 3.94, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":3.55,\"end\":3.94}]"}
{"utterance_id": "3752-4943-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.68, 3.33, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.68,\"end\":3.33}]"}
{"utterance_id": "5536-43363-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0010.flac", "task_type": "lexical_detection", "query": "prophets", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"prophets\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.66, 5.29, "prophets"]], "response_mode": "offline", "answer": "[{\"word\":\"prophets\",\"start\":4.66,\"end\":5.29}]"}
{"utterance_id": "6313-76958-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0013.flac", "task_type": "lexical_detection", "query": "outfit", "prompt": "Once the audio has finished, identify every instance of the word \"outfit\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.04, 1.45, "outfit"]], "response_mode": "offline", "answer": "[{\"word\":\"outfit\",\"start\":1.04,\"end\":1.45}]"}
{"utterance_id": "2803-154328-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0009.flac", "task_type": "lexical_detection", "query": "whichever", "prompt": "When the utterance is complete, report all instances of the word \"whichever\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.23, "whichever"]], "response_mode": "offline", "answer": "[{\"word\":\"whichever\",\"start\":1.44,\"end\":2.23}]"}
{"utterance_id": "7850-281318-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0010.flac", "task_type": "lexical_counting", "query": "magpie", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"magpie\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.55, 2.23, "one"], [3.59, 4.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.55,\"end\":2.23},{\"word\":\"two\",\"start\":3.59,\"end\":4.32}]"}
{"utterance_id": "3081-166546-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0049.flac", "task_type": "lexical_detection", "query": "position", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"position\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.24, 1.91, "position"]], "response_mode": "offline", "answer": "[{\"word\":\"position\",\"start\":1.24,\"end\":1.91}]"}
{"utterance_id": "2086-149220-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.32, "hens"]], "response_mode": "offline", "answer": "[{\"word\":\"hens\",\"start\":1.86,\"end\":2.32}]"}
{"utterance_id": "2412-153948-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.12, 1.44, "sheep"], [14.12, 14.65, "ewes"]], "response_mode": "offline", "answer": "[{\"word\":\"sheep\",\"start\":1.12,\"end\":1.44},{\"word\":\"ewes\",\"start\":14.12,\"end\":14.65}]"}
{"utterance_id": "6295-244435-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.29, 0.79, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":0.29,\"end\":0.79}]"}
{"utterance_id": "8297-275154-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0009.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.47, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":1.16,\"end\":1.47}]"}
{"utterance_id": "5895-34629-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.17, 6.77, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":6.17,\"end\":6.77}]"}
{"utterance_id": "1272-141231-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After the audio ends, extract all occurrences of a disease name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 0.83, "cut"]], "response_mode": "offline", "answer": "[{\"word\":\"cut\",\"start\":0.59,\"end\":0.83}]"}
{"utterance_id": "3081-166546-0089", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0089.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.35, 5.57, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.35,\"end\":5.57}]"}
{"utterance_id": "251-137823-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.37, 2.2, "mister faber"]], "response_mode": "offline", "answer": "[{\"word\":\"mister faber\",\"start\":1.37,\"end\":2.2}]"}
{"utterance_id": "8297-275156-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.5, 4.13, "catherine"]], "response_mode": "offline", "answer": "[{\"word\":\"catherine\",\"start\":3.5,\"end\":4.13}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "lexical_counting", "query": "ruggedo", "prompt": "When the utterance is complete, report how the word \"ruggedo\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.83, 7.25, "one"], [11.26, 11.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.83,\"end\":7.25},{\"word\":\"two\",\"start\":11.26,\"end\":11.66}]"}
{"utterance_id": "6313-66129-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0033.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a person's name with word/start/end. If none, reply [].", "labels": [[0.93, 1.3, "one"], [6.75, 7.33, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.93,\"end\":1.3},{\"word\":\"two\",\"start\":6.75,\"end\":7.33}]"}
{"utterance_id": "2428-83699-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0041.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 1.22, "mister christopher"]], "response_mode": "offline", "answer": "[{\"word\":\"mister christopher\",\"start\":0.39,\"end\":1.22}]"}
{"utterance_id": "6241-66616-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0023.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.76, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.55,\"end\":0.76}]"}
{"utterance_id": "5694-64038-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0018.flac", "task_type": "lexical_detection", "query": "privations", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"privations\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 2.18, "privations"]], "response_mode": "offline", "answer": "[{\"word\":\"privations\",\"start\":1.46,\"end\":2.18}]"}
{"utterance_id": "652-129742-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0001.flac", "task_type": "lexical_counting", "query": "nest", "prompt": "Listen to the full audio. Count every occurrence of the word \"nest\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.31, "one"], [7.19, 7.53, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.96,\"end\":1.31},{\"word\":\"two\",\"start\":7.19,\"end\":7.53}]"}
{"utterance_id": "3576-138058-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0026.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 1.55, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":0.62,\"end\":1.55}]"}
{"utterance_id": "251-136532-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0017.flac", "task_type": "lexical_detection", "query": "cash", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"cash\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.88, "cash"]], "response_mode": "offline", "answer": "[{\"word\":\"cash\",\"start\":3.4,\"end\":3.88}]"}
{"utterance_id": "6319-275224-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.54, 3.05, "gardener"]], "response_mode": "offline", "answer": "[{\"word\":\"gardener\",\"start\":2.54,\"end\":3.05}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.25, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":0.93,\"end\":1.25}]"}
{"utterance_id": "6313-76958-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0029.flac", "task_type": "lexical_detection", "query": "meals", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"meals\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.03, 4.46, "meals"]], "response_mode": "offline", "answer": "[{\"word\":\"meals\",\"start\":4.03,\"end\":4.46}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.41, 3.76, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.41,\"end\":3.76}]"}
{"utterance_id": "6313-76958-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 1.99, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":1.55,\"end\":1.99}]"}
{"utterance_id": "2086-149220-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0038.flac", "task_type": "lexical_detection", "query": "system", "prompt": "When the utterance is complete, report all instances of the word \"system\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.44, 3.92, "system"]], "response_mode": "offline", "answer": "[{\"word\":\"system\",\"start\":3.44,\"end\":3.92}]"}
{"utterance_id": "174-168635-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0022.flac", "task_type": "lexical_detection", "query": "child's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"child's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.68, "child's"]], "response_mode": "offline", "answer": "[{\"word\":\"child's\",\"start\":1.18,\"end\":1.68}]"}
{"utterance_id": "7850-281318-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0016.flac", "task_type": "lexical_detection", "query": "screamed", "prompt": "Once the audio has finished, identify every instance of the word \"screamed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.92, 2.4, "screamed"]], "response_mode": "offline", "answer": "[{\"word\":\"screamed\",\"start\":1.92,\"end\":2.4}]"}
{"utterance_id": "6241-66616-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.24, 4.84, "wabigoon's"]], "response_mode": "offline", "answer": "[{\"word\":\"wabigoon's\",\"start\":4.24,\"end\":4.84}]"}
{"utterance_id": "5536-43363-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.27, 0.54, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":0.27,\"end\":0.54}]"}
{"utterance_id": "7850-73752-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 2.89, "miss temple"]], "response_mode": "offline", "answer": "[{\"word\":\"miss temple\",\"start\":2.13,\"end\":2.89}]"}
{"utterance_id": "6313-66125-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.81, 4.25, "lige"]], "response_mode": "offline", "answer": "[{\"word\":\"lige\",\"start\":3.81,\"end\":4.25}]"}
{"utterance_id": "84-121123-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0002.flac", "task_type": "lexical_detection", "query": "purple", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"purple\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.39, 8.83, "purple"]], "response_mode": "offline", "answer": "[{\"word\":\"purple\",\"start\":8.39,\"end\":8.83}]"}
{"utterance_id": "6241-66616-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 1.91, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.66,\"end\":1.91}]"}
{"utterance_id": "1993-147965-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.65, 0.94, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":0.65,\"end\":0.94}]"}
{"utterance_id": "2035-147960-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.57, 0.87, "snake"]], "response_mode": "offline", "answer": "[{\"word\":\"snake\",\"start\":0.57,\"end\":0.87}]"}
{"utterance_id": "8842-304647-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0004.flac", "task_type": "lexical_detection", "query": "disease", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"disease\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.9, 7.48, "disease"]], "response_mode": "offline", "answer": "[{\"word\":\"disease\",\"start\":6.9,\"end\":7.48}]"}
{"utterance_id": "3576-138058-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0030.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.91, 3.21, "three"], [4.96, 5.28, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.91,\"end\":3.21},{\"word\":\"three\",\"start\":4.96,\"end\":5.28}]"}
{"utterance_id": "1462-170142-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.48, 3.99, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":3.48,\"end\":3.99}]"}
{"utterance_id": "777-126732-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0057.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.76, 10.21, "husband's"]], "response_mode": "offline", "answer": "[{\"word\":\"husband's\",\"start\":9.76,\"end\":10.21}]"}
{"utterance_id": "777-126732-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 0.92, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":0.31,\"end\":0.92}]"}
{"utterance_id": "5694-64029-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0009.flac", "task_type": "lexical_detection", "query": "quick", "prompt": "After the audio ends, extract all occurrences of the word \"quick\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.89, "quick"]], "response_mode": "offline", "answer": "[{\"word\":\"quick\",\"start\":0.55,\"end\":0.89}]"}
{"utterance_id": "2277-149874-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.06, 2.64, "hanson"]], "response_mode": "offline", "answer": "[{\"word\":\"hanson\",\"start\":2.06,\"end\":2.64}]"}
{"utterance_id": "777-126732-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0020.flac", "task_type": "lexical_detection", "query": "ass", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"ass\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.76, "ass"]], "response_mode": "offline", "answer": "[{\"word\":\"ass\",\"start\":1.1,\"end\":1.76}]"}
{"utterance_id": "1673-143397-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0010.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.57, 4.91, "monk"], [5.92, 6.32, "pope"]], "response_mode": "offline", "answer": "[{\"word\":\"monk\",\"start\":4.57,\"end\":4.91},{\"word\":\"pope\",\"start\":5.92,\"end\":6.32}]"}
{"utterance_id": "777-126732-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0017.flac", "task_type": "lexical_detection", "query": "perfectly", "prompt": "Process the full audio offline. Find every instance of the word \"perfectly\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.81, "perfectly"]], "response_mode": "offline", "answer": "[{\"word\":\"perfectly\",\"start\":1.39,\"end\":1.81}]"}
{"utterance_id": "2086-149220-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0049.flac", "task_type": "lexical_detection", "query": "depths", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"depths\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.33, "depths"]], "response_mode": "offline", "answer": "[{\"word\":\"depths\",\"start\":1.96,\"end\":2.33}]"}
{"utterance_id": "3536-23268-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.52, "missus horton"], [9.98, 10.57, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"missus horton\",\"start\":3.81,\"end\":4.52},{\"word\":\"miss milner\",\"start\":9.98,\"end\":10.57}]"}
{"utterance_id": "1272-141231-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0013.flac", "task_type": "lexical_detection", "query": "twenties", "prompt": "Process the full audio offline. Find every instance of the word \"twenties\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 1.43, "twenties"]], "response_mode": "offline", "answer": "[{\"word\":\"twenties\",\"start\":0.68,\"end\":1.43}]"}
{"utterance_id": "3081-166546-0081", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0081.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After listening to the whole recording, output every instance of a disease name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.55, "wound"]], "response_mode": "offline", "answer": "[{\"word\":\"wound\",\"start\":1.9,\"end\":2.55}]"}
{"utterance_id": "3853-163249-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0050.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 1.21, "mister power"]], "response_mode": "offline", "answer": "[{\"word\":\"mister power\",\"start\":0.53,\"end\":1.21}]"}
{"utterance_id": "2277-149874-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0014.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.16, 6.56, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":6.16,\"end\":6.56}]"}
{"utterance_id": "6241-66616-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0019.flac", "task_type": "lexical_detection", "query": "tears", "prompt": "When the utterance is complete, report all instances of the word \"tears\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.17, "tears"]], "response_mode": "offline", "answer": "[{\"word\":\"tears\",\"start\":0.81,\"end\":1.17}]"}
{"utterance_id": "174-84280-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.14, 2.55, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":2.14,\"end\":2.55}]"}
{"utterance_id": "8842-302201-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0009.flac", "task_type": "lexical_detection", "query": "angel", "prompt": "After listening to the whole recording, output every instance of the word \"angel\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.23, "angel"]], "response_mode": "offline", "answer": "[{\"word\":\"angel\",\"start\":1.74,\"end\":2.23}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "lexical_counting", "query": "beast", "prompt": "After listening to the whole recording, output timed count words for the word \"beast\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.14, 1.54, "one"], [13.93, 14.24, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.14,\"end\":1.54},{\"word\":\"two\",\"start\":13.93,\"end\":14.24}]"}
{"utterance_id": "1272-141231-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0016.flac", "task_type": "lexical_detection", "query": "seconds", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"seconds\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.77, 1.32, "seconds"]], "response_mode": "offline", "answer": "[{\"word\":\"seconds\",\"start\":0.77,\"end\":1.32}]"}
{"utterance_id": "1462-170142-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.85, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":0.47,\"end\":0.85}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After listening to the whole recording, output timed count words for a person's name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.76, 1.17, "one"], [6.19, 6.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.76,\"end\":1.17},{\"word\":\"two\",\"start\":6.19,\"end\":6.79}]"}
{"utterance_id": "6319-64726-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0019.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[8.65, 9.11, "almoner"]], "response_mode": "offline", "answer": "[{\"word\":\"almoner\",\"start\":8.65,\"end\":9.11}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.17, "newsome's"], [6.19, 6.79, "minnetaki"]], "response_mode": "offline", "answer": "[{\"word\":\"newsome's\",\"start\":0.76,\"end\":1.17},{\"word\":\"minnetaki\",\"start\":6.19,\"end\":6.79}]"}
{"utterance_id": "3752-4943-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0010.flac", "task_type": "lexical_detection", "query": "sorry", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sorry\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 1.05, "sorry"]], "response_mode": "offline", "answer": "[{\"word\":\"sorry\",\"start\":0.65,\"end\":1.05}]"}
{"utterance_id": "6313-66129-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.5, 2.74, "neck"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":2.5,\"end\":2.74}]"}
{"utterance_id": "2412-153947-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.23, 1.57, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":1.23,\"end\":1.57}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.54, "profession"], [10.12, 10.75, "officer"], [17.88, 18.38, "officers"]], "response_mode": "offline", "answer": "[{\"word\":\"profession\",\"start\":2.9,\"end\":3.54},{\"word\":\"officer\",\"start\":10.12,\"end\":10.75},{\"word\":\"officers\",\"start\":17.88,\"end\":18.38}]"}
{"utterance_id": "7976-110523-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0003.flac", "task_type": "lexical_detection", "query": "consented", "prompt": "Process the full audio offline. Find every instance of the word \"consented\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.47, "consented"]], "response_mode": "offline", "answer": "[{\"word\":\"consented\",\"start\":1.83,\"end\":2.47}]"}
{"utterance_id": "2035-147961-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0029.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.44, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":1.1,\"end\":1.44}]"}
{"utterance_id": "2412-153948-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 1.98, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":1.82,\"end\":1.98}]"}
{"utterance_id": "1462-170142-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.83, 7.25, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":6.83,\"end\":7.25}]"}
{"utterance_id": "7976-110523-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0018.flac", "task_type": "lexical_detection", "query": "get", "prompt": "Once the audio has finished, identify every instance of the word \"get\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.3, 1.46, "get"]], "response_mode": "offline", "answer": "[{\"word\":\"get\",\"start\":1.3,\"end\":1.46}]"}
{"utterance_id": "6345-93306-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0003.flac", "task_type": "lexical_detection", "query": "quiet", "prompt": "Once the audio has finished, identify every instance of the word \"quiet\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.22, 1.61, "quiet"]], "response_mode": "offline", "answer": "[{\"word\":\"quiet\",\"start\":1.22,\"end\":1.61}]"}
{"utterance_id": "3081-166546-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0023.flac", "task_type": "lexical_detection", "query": "evidently", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"evidently\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.19, 6.89, "evidently"]], "response_mode": "offline", "answer": "[{\"word\":\"evidently\",\"start\":6.19,\"end\":6.89}]"}
{"utterance_id": "5694-64029-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0020.flac", "task_type": "lexical_detection", "query": "ranks", "prompt": "After the audio ends, extract all occurrences of the word \"ranks\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.86, 5.32, "ranks"]], "response_mode": "offline", "answer": "[{\"word\":\"ranks\",\"start\":4.86,\"end\":5.32}]"}
{"utterance_id": "251-137823-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0015.flac", "task_type": "lexical_detection", "query": "let's", "prompt": "After the audio ends, extract all occurrences of the word \"let's\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.11, 0.35, "let's"]], "response_mode": "offline", "answer": "[{\"word\":\"let's\",\"start\":0.11,\"end\":0.35}]"}
{"utterance_id": "3081-166546-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0047.flac", "task_type": "lexical_detection", "query": "you're", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"you're\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.16, 2.45, "you're"]], "response_mode": "offline", "answer": "[{\"word\":\"you're\",\"start\":2.16,\"end\":2.45}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "After the audio ends, tally every instance of a plant name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[6.75, 7.2, "one"], [8.49, 9.23, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.75,\"end\":7.2},{\"word\":\"two\",\"start\":8.49,\"end\":9.23}]"}
{"utterance_id": "84-121550-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0019.flac", "task_type": "lexical_detection", "query": "beheld", "prompt": "Process the full audio offline. Find every instance of the word \"beheld\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.67, 4.27, "beheld"]], "response_mode": "offline", "answer": "[{\"word\":\"beheld\",\"start\":3.67,\"end\":4.27}]"}
{"utterance_id": "5338-284437-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.98, 1.74, "giant"]], "response_mode": "offline", "answer": "[{\"word\":\"giant\",\"start\":0.98,\"end\":1.74}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.26, "monastery"]], "response_mode": "offline", "answer": "[{\"word\":\"monastery\",\"start\":2.46,\"end\":3.26}]"}
{"utterance_id": "5694-64038-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0012.flac", "task_type": "lexical_detection", "query": "stayed", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"stayed\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.09, 1.4, "stayed"]], "response_mode": "offline", "answer": "[{\"word\":\"stayed\",\"start\":1.09,\"end\":1.4}]"}
{"utterance_id": "6319-64726-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.4, 7.67, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.4,\"end\":7.67}]"}
{"utterance_id": "84-121123-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0017.flac", "task_type": "lexical_detection", "query": "keep", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"keep\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.01, 7.25, "keep"]], "response_mode": "offline", "answer": "[{\"word\":\"keep\",\"start\":7.01,\"end\":7.25}]"}
{"utterance_id": "174-168635-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.45, 2.4, "jean valjean"]], "response_mode": "offline", "answer": "[{\"word\":\"jean valjean\",\"start\":1.45,\"end\":2.4}]"}
{"utterance_id": "652-130726-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.3, 0.99, "john tait"]], "response_mode": "offline", "answer": "[{\"word\":\"john tait\",\"start\":0.3,\"end\":0.99}]"}
{"utterance_id": "2412-153948-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0003.flac", "task_type": "lexical_detection", "query": "gather", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"gather\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.43, 6.7, "gather"]], "response_mode": "offline", "answer": "[{\"word\":\"gather\",\"start\":6.43,\"end\":6.7}]"}
{"utterance_id": "8297-275155-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0005.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.22, 7.64, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":7.22,\"end\":7.64}]"}
{"utterance_id": "2078-142845-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0051.flac", "task_type": "lexical_detection", "query": "tin", "prompt": "When the utterance is complete, report all instances of the word \"tin\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.12, 2.51, "tin"]], "response_mode": "offline", "answer": "[{\"word\":\"tin\",\"start\":2.12,\"end\":2.51}]"}
{"utterance_id": "1919-142785-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0004.flac", "task_type": "lexical_detection", "query": "yolks", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"yolks\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.89, 1.21, "yolks"]], "response_mode": "offline", "answer": "[{\"word\":\"yolks\",\"start\":0.89,\"end\":1.21}]"}
{"utterance_id": "2428-83705-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0021.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.37, "papa"]], "response_mode": "offline", "answer": "[{\"word\":\"papa\",\"start\":1.05,\"end\":1.37}]"}
{"utterance_id": "5338-284437-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.7, 6.11, "trot's"]], "response_mode": "offline", "answer": "[{\"word\":\"trot's\",\"start\":5.7,\"end\":6.11}]"}
{"utterance_id": "1919-142785-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0056.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.95, "bread crumbs"]], "response_mode": "offline", "answer": "[{\"word\":\"bread crumbs\",\"start\":1.08,\"end\":1.95}]"}
{"utterance_id": "3752-4944-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0056.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.77, 3.25, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":2.77,\"end\":3.25}]"}
{"utterance_id": "6319-275224-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0012.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.71, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":2.95,\"end\":3.71}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[3.78, 4.07, "aunt"], [9.59, 10.18, "spouse"]], "response_mode": "offline", "answer": "[{\"word\":\"aunt\",\"start\":3.78,\"end\":4.07},{\"word\":\"spouse\",\"start\":9.59,\"end\":10.18}]"}
{"utterance_id": "3081-166546-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0009.flac", "task_type": "lexical_detection", "query": "suicide", "prompt": "Once the audio has finished, identify every instance of the word \"suicide\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.59, 3.53, "suicide"]], "response_mode": "offline", "answer": "[{\"word\":\"suicide\",\"start\":2.59,\"end\":3.53}]"}
{"utterance_id": "6345-93306-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.82, 4.17, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":3.82,\"end\":4.17}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "lexical_counting", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"food\" (word, start, end). If none, reply [].", "labels": [[9.57, 9.91, "one"], [14.29, 14.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.57,\"end\":9.91},{\"word\":\"two\",\"start\":14.29,\"end\":14.68}]"}
{"utterance_id": "1462-170138-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0000.flac", "task_type": "lexical_counting", "query": "history", "prompt": "Process the full audio offline. For every occurrence of the word \"history\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.25, "one"], [5.76, 6.23, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.75,\"end\":4.25},{\"word\":\"two\",\"start\":5.76,\"end\":6.23}]"}
{"utterance_id": "251-136532-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0016.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Listen to the entire clip, then report the count sequence for an animal name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 3.01, "one"], [4.25, 4.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.29,\"end\":3.01},{\"word\":\"two\",\"start\":4.25,\"end\":4.74}]"}
{"utterance_id": "5895-34615-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.35, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.81,\"end\":1.35}]"}
{"utterance_id": "84-121123-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[3.69, 3.96, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"half\",\"start\":3.69,\"end\":3.96}]"}
{"utterance_id": "2086-149220-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0000.flac", "task_type": "lexical_counting", "query": "partly", "prompt": "Process the full audio offline. For every occurrence of the word \"partly\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.28, 6.63, "one"], [8.5, 8.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.28,\"end\":6.63},{\"word\":\"two\",\"start\":8.5,\"end\":8.79}]"}
{"utterance_id": "251-118436-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0019.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.12, 2.63, "silver"]], "response_mode": "offline", "answer": "[{\"word\":\"silver\",\"start\":2.12,\"end\":2.63}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the full audio. After it finishes, find every instance of a date. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.52, 7.26, "april twelfth"], [8.88, 9.79, "april twentieth"]], "response_mode": "offline", "answer": "[{\"word\":\"april twelfth\",\"start\":6.52,\"end\":7.26},{\"word\":\"april twentieth\",\"start\":8.88,\"end\":9.79}]"}
{"utterance_id": "652-129742-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.22, "mayonnaise"]], "response_mode": "offline", "answer": "[{\"word\":\"mayonnaise\",\"start\":0.59,\"end\":1.22}]"}
{"utterance_id": "2428-83705-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0020.flac", "task_type": "lexical_detection", "query": "comfortably", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"comfortably\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.7, "comfortably"]], "response_mode": "offline", "answer": "[{\"word\":\"comfortably\",\"start\":2.98,\"end\":3.7}]"}
{"utterance_id": "1272-141231-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0023.flac", "task_type": "lexical_detection", "query": "points", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"points\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.89, 6.41, "points"]], "response_mode": "offline", "answer": "[{\"word\":\"points\",\"start\":5.89,\"end\":6.41}]"}
{"utterance_id": "1272-135031-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0022.flac", "task_type": "lexical_detection", "query": "kaliko", "prompt": "After the audio ends, extract all occurrences of the word \"kaliko\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.48, "kaliko"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":1.81,\"end\":2.48}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.29, "shasta"], [4.32, 4.73, "shasta"], [7.81, 8.59, "mc cloud"]], "response_mode": "offline", "answer": "[{\"word\":\"shasta\",\"start\":0.86,\"end\":1.29},{\"word\":\"shasta\",\"start\":4.32,\"end\":4.73},{\"word\":\"mc cloud\",\"start\":7.81,\"end\":8.59}]"}
{"utterance_id": "6319-64726-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.83, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.54,\"end\":1.83}]"}
{"utterance_id": "6295-64301-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0024.flac", "task_type": "lexical_detection", "query": "hath", "prompt": "After listening to the whole recording, output every instance of the word \"hath\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 2.87, "hath"], [5.16, 5.44, "hath"]], "response_mode": "offline", "answer": "[{\"word\":\"hath\",\"start\":2.52,\"end\":2.87},{\"word\":\"hath\",\"start\":5.16,\"end\":5.44}]"}
{"utterance_id": "6295-244435-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0004.flac", "task_type": "lexical_detection", "query": "sharply", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sharply\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.37, 8.92, "sharply"]], "response_mode": "offline", "answer": "[{\"word\":\"sharply\",\"start\":8.37,\"end\":8.92}]"}
{"utterance_id": "2035-147961-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0025.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.77, 1.22, "wolves"]], "response_mode": "offline", "answer": "[{\"word\":\"wolves\",\"start\":0.77,\"end\":1.22}]"}
{"utterance_id": "2803-161169-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.62, 8.03, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":7.62,\"end\":8.03}]"}
{"utterance_id": "5895-34622-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0000.flac", "task_type": "lexical_detection", "query": "stories", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"stories\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.31, 2.98, "stories"]], "response_mode": "offline", "answer": "[{\"word\":\"stories\",\"start\":2.31,\"end\":2.98}]"}
{"utterance_id": "8842-304647-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0013.flac", "task_type": "lexical_detection", "query": "liberty", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"liberty\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.94, 8.4, "liberty"]], "response_mode": "offline", "answer": "[{\"word\":\"liberty\",\"start\":7.94,\"end\":8.4}]"}
{"utterance_id": "3576-138058-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0030.flac", "task_type": "lexical_counting", "query": "three", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"three\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[2.91, 3.21, "one"], [4.96, 5.28, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.91,\"end\":3.21},{\"word\":\"two\",\"start\":4.96,\"end\":5.28}]"}
{"utterance_id": "2078-142845-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0022.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 2.04, "rolls"]], "response_mode": "offline", "answer": "[{\"word\":\"rolls\",\"start\":1.22,\"end\":2.04}]"}
{"utterance_id": "1988-24833-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0024.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.57, "coffee"]], "response_mode": "offline", "answer": "[{\"word\":\"coffee\",\"start\":1.17,\"end\":1.57}]"}
{"utterance_id": "8297-275155-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 5.1, "sydenham"]], "response_mode": "offline", "answer": "[{\"word\":\"sydenham\",\"start\":4.47,\"end\":5.1}]"}
{"utterance_id": "2803-154328-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.75, 6.14, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":5.75,\"end\":6.14}]"}
{"utterance_id": "2902-9008-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0002.flac", "task_type": "lexical_detection", "query": "spare", "prompt": "Once the audio has finished, identify every instance of the word \"spare\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.64, 5.98, "spare"]], "response_mode": "offline", "answer": "[{\"word\":\"spare\",\"start\":5.64,\"end\":5.98}]"}
{"utterance_id": "422-122949-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0010.flac", "task_type": "lexical_detection", "query": "blending", "prompt": "When the utterance is complete, report all instances of the word \"blending\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.12, 6.61, "blending"]], "response_mode": "offline", "answer": "[{\"word\":\"blending\",\"start\":6.12,\"end\":6.61}]"}
{"utterance_id": "777-126732-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0056.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Listen to the full audio. Count every occurrence of a color name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.14, 3.43, "one"], [6.79, 7.04, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.14,\"end\":3.43},{\"word\":\"two\",\"start\":6.79,\"end\":7.04}]"}
{"utterance_id": "1673-143396-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0002.flac", "task_type": "lexical_detection", "query": "charitable", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"charitable\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.72, "charitable"]], "response_mode": "offline", "answer": "[{\"word\":\"charitable\",\"start\":1.21,\"end\":1.72}]"}
{"utterance_id": "422-122949-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0027.flac", "task_type": "lexical_detection", "query": "limbs", "prompt": "When the utterance is complete, report all instances of the word \"limbs\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.32, 15.72, "limbs"]], "response_mode": "offline", "answer": "[{\"word\":\"limbs\",\"start\":15.32,\"end\":15.72}]"}
{"utterance_id": "1673-143397-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.5, 0.96, "vatican"]], "response_mode": "offline", "answer": "[{\"word\":\"vatican\",\"start\":0.5,\"end\":0.96}]"}
{"utterance_id": "3536-23268-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0020.flac", "task_type": "lexical_detection", "query": "reality", "prompt": "After the audio ends, extract all occurrences of the word \"reality\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.49, "reality"]], "response_mode": "offline", "answer": "[{\"word\":\"reality\",\"start\":3.86,\"end\":4.49}]"}
{"utterance_id": "777-126732-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0005.flac", "task_type": "lexical_detection", "query": "joint", "prompt": "After listening to the whole recording, output every instance of the word \"joint\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.85, "joint"]], "response_mode": "offline", "answer": "[{\"word\":\"joint\",\"start\":2.4,\"end\":2.85}]"}
{"utterance_id": "2078-142845-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0035.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.17, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.88,\"end\":2.17}]"}
{"utterance_id": "1919-142785-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0035.flac", "task_type": "lexical_detection", "query": "mackerel", "prompt": "When the utterance is complete, report all instances of the word \"mackerel\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 3.13, "mackerel"]], "response_mode": "offline", "answer": "[{\"word\":\"mackerel\",\"start\":2.71,\"end\":3.13}]"}
{"utterance_id": "2086-149220-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0017.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Process the full audio offline. For every occurrence of an animal name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.78, "one"], [3.5, 3.73, "two"], [7.98, 8.42, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.39,\"end\":1.78},{\"word\":\"two\",\"start\":3.5,\"end\":3.73},{\"word\":\"three\",\"start\":7.98,\"end\":8.42}]"}
{"utterance_id": "3752-4943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 4.97, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":4.47,\"end\":4.97}]"}
{"utterance_id": "8297-275155-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0008.flac", "task_type": "lexical_detection", "query": "added", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"added\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.99, "added"]], "response_mode": "offline", "answer": "[{\"word\":\"added\",\"start\":0.52,\"end\":0.99}]"}
{"utterance_id": "2412-153947-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.5, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":1.17,\"end\":1.5}]"}
{"utterance_id": "8297-275154-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0021.flac", "task_type": "lexical_detection", "query": "train", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"train\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.45, "train"]], "response_mode": "offline", "answer": "[{\"word\":\"train\",\"start\":2.84,\"end\":3.45}]"}
{"utterance_id": "8297-275154-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0027.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.86, "servants"]], "response_mode": "offline", "answer": "[{\"word\":\"servants\",\"start\":2.14,\"end\":2.86}]"}
{"utterance_id": "7850-286674-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0007.flac", "task_type": "lexical_detection", "query": "lower", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"lower\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.38, "lower"]], "response_mode": "offline", "answer": "[{\"word\":\"lower\",\"start\":1.04,\"end\":1.38}]"}
{"utterance_id": "2803-161169-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0003.flac", "task_type": "lexical_detection", "query": "contents", "prompt": "Once the audio has finished, identify every instance of the word \"contents\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.0, 7.92, "contents"]], "response_mode": "offline", "answer": "[{\"word\":\"contents\",\"start\":7.0,\"end\":7.92}]"}
{"utterance_id": "5536-43359-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.98, 4.38, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":3.98,\"end\":4.38}]"}
{"utterance_id": "6241-61943-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.03, 6.6, "valkyrie"]], "response_mode": "offline", "answer": "[{\"word\":\"valkyrie\",\"start\":6.03,\"end\":6.6}]"}
{"utterance_id": "3853-163249-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0045.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.64, 3.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.64,\"end\":3.86}]"}
{"utterance_id": "1673-143397-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0013.flac", "task_type": "lexical_detection", "query": "garrison", "prompt": "Process the full audio offline. Find every instance of the word \"garrison\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.41, 4.87, "garrison"]], "response_mode": "offline", "answer": "[{\"word\":\"garrison\",\"start\":4.41,\"end\":4.87}]"}
{"utterance_id": "1272-141231-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0019.flac", "task_type": "lexical_detection", "query": "final", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"final\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.92, 3.33, "final"]], "response_mode": "offline", "answer": "[{\"word\":\"final\",\"start\":2.92,\"end\":3.33}]"}
{"utterance_id": "1919-142785-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.03, 7.64, "twenty four"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty four\",\"start\":7.03,\"end\":7.64}]"}
{"utterance_id": "3536-23268-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0029.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.26, 6.59, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":6.26,\"end\":6.59}]"}
{"utterance_id": "1993-147964-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0009.flac", "task_type": "lexical_detection", "query": "produced", "prompt": "Once the audio has finished, identify every instance of the word \"produced\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.02, 2.46, "produced"]], "response_mode": "offline", "answer": "[{\"word\":\"produced\",\"start\":2.02,\"end\":2.46}]"}
{"utterance_id": "2803-161169-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.09, 5.67, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":5.09,\"end\":5.67}]"}
{"utterance_id": "7976-110523-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0006.flac", "task_type": "lexical_detection", "query": "hansel", "prompt": "After listening to the whole recording, output every instance of the word \"hansel\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.81, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":1.44,\"end\":1.81}]"}
{"utterance_id": "2035-147961-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.56, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.31,\"end\":1.56}]"}
{"utterance_id": "174-50561-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0012.flac", "task_type": "lexical_detection", "query": "wandering", "prompt": "After the audio ends, extract all occurrences of the word \"wandering\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 1.1, "wandering"]], "response_mode": "offline", "answer": "[{\"word\":\"wandering\",\"start\":0.65,\"end\":1.1}]"}
{"utterance_id": "174-168635-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0008.flac", "task_type": "lexical_detection", "query": "gulf", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"gulf\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.21, 4.67, "gulf"], [9.32, 9.73, "gulf"]], "response_mode": "offline", "answer": "[{\"word\":\"gulf\",\"start\":4.21,\"end\":4.67},{\"word\":\"gulf\",\"start\":9.32,\"end\":9.73}]"}
{"utterance_id": "251-136532-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.32, 0.83, "tony's"]], "response_mode": "offline", "answer": "[{\"word\":\"tony's\",\"start\":0.32,\"end\":0.83}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[18.27, 18.67, "artist"]], "response_mode": "offline", "answer": "[{\"word\":\"artist\",\"start\":18.27,\"end\":18.67}]"}
{"utterance_id": "1919-142785-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0023.flac", "task_type": "lexical_detection", "query": "store", "prompt": "Process the full audio offline. Find every instance of the word \"store\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.86, "store"]], "response_mode": "offline", "answer": "[{\"word\":\"store\",\"start\":0.53,\"end\":0.86}]"}
{"utterance_id": "2035-152373-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.9, "prelate"]], "response_mode": "offline", "answer": "[{\"word\":\"prelate\",\"start\":1.41,\"end\":1.9}]"}
{"utterance_id": "2902-9006-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 4.04, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":3.59,\"end\":4.04}]"}
{"utterance_id": "6313-66125-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0010.flac", "task_type": "lexical_detection", "query": "end", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"end\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.21, 3.58, "end"]], "response_mode": "offline", "answer": "[{\"word\":\"end\",\"start\":3.21,\"end\":3.58}]"}
{"utterance_id": "2412-153954-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0021.flac", "task_type": "lexical_detection", "query": "watches", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"watches\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.14, 5.75, "watches"]], "response_mode": "offline", "answer": "[{\"word\":\"watches\",\"start\":5.14,\"end\":5.75}]"}
{"utterance_id": "5895-34615-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0021.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Process the full audio offline. For every occurrence of a body part, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.4, 4.84, "one"], [6.88, 7.22, "two"], [11.51, 11.77, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.4,\"end\":4.84},{\"word\":\"two\",\"start\":6.88,\"end\":7.22},{\"word\":\"three\",\"start\":11.51,\"end\":11.77}]"}
{"utterance_id": "777-126732-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.73, "apostle"]], "response_mode": "offline", "answer": "[{\"word\":\"apostle\",\"start\":2.06,\"end\":2.73}]"}
{"utterance_id": "5338-284437-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0011.flac", "task_type": "lexical_detection", "query": "coralie", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"coralie\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.48, "coralie"]], "response_mode": "offline", "answer": "[{\"word\":\"coralie\",\"start\":0.87,\"end\":1.48}]"}
{"utterance_id": "5694-64029-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0006.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After listening to the whole recording, output every instance of a date with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.5, 6.04, "march"]], "response_mode": "offline", "answer": "[{\"word\":\"march\",\"start\":5.5,\"end\":6.04}]"}
{"utterance_id": "2412-153954-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0013.flac", "task_type": "lexical_detection", "query": "rounded", "prompt": "When the utterance is complete, report all instances of the word \"rounded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.04, "rounded"]], "response_mode": "offline", "answer": "[{\"word\":\"rounded\",\"start\":1.7,\"end\":2.04}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "lexical_detection", "query": "transformed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"transformed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.45, 7.28, "transformed"]], "response_mode": "offline", "answer": "[{\"word\":\"transformed\",\"start\":6.45,\"end\":7.28}]"}
{"utterance_id": "5694-64025-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0017.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.94, 1.3, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":0.94,\"end\":1.3}]"}
{"utterance_id": "2035-152373-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0006.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the full audio. After it finishes, find every instance of a disease name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.49, "epidemic"]], "response_mode": "offline", "answer": "[{\"word\":\"epidemic\",\"start\":3.86,\"end\":4.49}]"}
{"utterance_id": "3170-137482-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0007.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "After listening to the whole recording, output every instance of a time of day with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.38, 13.79, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":13.38,\"end\":13.79}]"}
{"utterance_id": "3081-166546-0076", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0076.flac", "task_type": "lexical_detection", "query": "difference", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"difference\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.83, "difference"]], "response_mode": "offline", "answer": "[{\"word\":\"difference\",\"start\":1.23,\"end\":1.83}]"}
{"utterance_id": "5338-24615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0002.flac", "task_type": "lexical_detection", "query": "rushed", "prompt": "Process the full audio offline. Find every instance of the word \"rushed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.76, 9.1, "rushed"]], "response_mode": "offline", "answer": "[{\"word\":\"rushed\",\"start\":8.76,\"end\":9.1}]"}
{"utterance_id": "3853-163249-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0000.flac", "task_type": "lexical_detection", "query": "letty", "prompt": "Once the audio has finished, identify every instance of the word \"letty\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.96, 1.24, "letty"]], "response_mode": "offline", "answer": "[{\"word\":\"letty\",\"start\":0.96,\"end\":1.24}]"}
{"utterance_id": "5694-64029-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0000.flac", "task_type": "lexical_detection", "query": "months", "prompt": "When the utterance is complete, report all instances of the word \"months\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.66, "months"]], "response_mode": "offline", "answer": "[{\"word\":\"months\",\"start\":1.2,\"end\":1.66}]"}
{"utterance_id": "2035-152373-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0014.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.84, 3.6, "christianity"]], "response_mode": "offline", "answer": "[{\"word\":\"christianity\",\"start\":2.84,\"end\":3.6}]"}
{"utterance_id": "2803-161169-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0005.flac", "task_type": "lexical_counting", "query": "side", "prompt": "Listen to the entire clip, then report the count sequence for the word \"side\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.34, 8.71, "one"], [10.61, 11.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.34,\"end\":8.71},{\"word\":\"two\",\"start\":10.61,\"end\":11.12}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.48, 2.34, "missus sterling"]], "response_mode": "offline", "answer": "[{\"word\":\"missus sterling\",\"start\":1.48,\"end\":2.34}]"}
{"utterance_id": "777-126732-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0066.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.8, 11.16, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":10.8,\"end\":11.16}]"}
{"utterance_id": "1993-147965-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0001.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[0.44, 0.83, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":0.44,\"end\":0.83}]"}
{"utterance_id": "6241-66616-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.15, 5.58, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":5.15,\"end\":5.58}]"}
{"utterance_id": "777-126732-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0024.flac", "task_type": "lexical_detection", "query": "gift", "prompt": "After the audio ends, extract all occurrences of the word \"gift\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.89, "gift"]], "response_mode": "offline", "answer": "[{\"word\":\"gift\",\"start\":1.52,\"end\":1.89}]"}
{"utterance_id": "3170-137482-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0010.flac", "task_type": "lexical_counting", "query": "few", "prompt": "After listening to the whole recording, output timed count words for the word \"few\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.01, 1.32, "one"], [6.95, 7.14, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.01,\"end\":1.32},{\"word\":\"two\",\"start\":6.95,\"end\":7.14}]"}
{"utterance_id": "1988-24833-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.11, "tom"], [2.28, 2.66, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.88,\"end\":1.11},{\"word\":\"hilda\",\"start\":2.28,\"end\":2.66}]"}
{"utterance_id": "6241-61946-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0022.flac", "task_type": "lexical_detection", "query": "slack", "prompt": "After the audio ends, extract all occurrences of the word \"slack\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.37, 7.66, "slack"]], "response_mode": "offline", "answer": "[{\"word\":\"slack\",\"start\":7.37,\"end\":7.66}]"}
{"utterance_id": "7976-110523-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[4.87, 5.26, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":4.87,\"end\":5.26}]"}
{"utterance_id": "2277-149896-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0032.flac", "task_type": "lexical_detection", "query": "situation", "prompt": "After listening to the whole recording, output every instance of the word \"situation\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.35, 5.12, "situation"]], "response_mode": "offline", "answer": "[{\"word\":\"situation\",\"start\":4.35,\"end\":5.12}]"}
{"utterance_id": "7850-281318-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0017.flac", "task_type": "lexical_detection", "query": "pigeon", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pigeon\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.58, "pigeon"]], "response_mode": "offline", "answer": "[{\"word\":\"pigeon\",\"start\":1.01,\"end\":1.58}]"}
{"utterance_id": "652-130726-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.44, 5.76, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":5.44,\"end\":5.76}]"}
{"utterance_id": "3853-163249-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0049.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.94, 3.25, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":2.94,\"end\":3.25}]"}
{"utterance_id": "1673-143397-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0018.flac", "task_type": "lexical_detection", "query": "avenue", "prompt": "After the audio ends, extract all occurrences of the word \"avenue\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.16, "avenue"]], "response_mode": "offline", "answer": "[{\"word\":\"avenue\",\"start\":1.73,\"end\":2.16}]"}
{"utterance_id": "3170-137482-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0002.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.67, 14.01, "purple"]], "response_mode": "offline", "answer": "[{\"word\":\"purple\",\"start\":13.67,\"end\":14.01}]"}
{"utterance_id": "3752-4944-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0021.flac", "task_type": "lexical_detection", "query": "lights", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"lights\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.37, 3.67, "lights"]], "response_mode": "offline", "answer": "[{\"word\":\"lights\",\"start\":3.37,\"end\":3.67}]"}
{"utterance_id": "2412-153954-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0014.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.01, 5.7, "villages"], [7.16, 7.54, "town"]], "response_mode": "offline", "answer": "[{\"word\":\"villages\",\"start\":5.01,\"end\":5.7},{\"word\":\"town\",\"start\":7.16,\"end\":7.54}]"}
{"utterance_id": "2412-153948-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0013.flac", "task_type": "lexical_detection", "query": "capital", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"capital\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.39, 5.97, "capital"]], "response_mode": "offline", "answer": "[{\"word\":\"capital\",\"start\":5.39,\"end\":5.97}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[0.19, 0.41, "one"], [2.41, 2.55, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.19,\"end\":0.41},{\"word\":\"two\",\"start\":2.41,\"end\":2.55}]"}
{"utterance_id": "5694-64029-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0024.flac", "task_type": "semantic_counting", "query": "country", "prompt": "Listen to the entire clip, then report the count sequence for a country name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.27, 4.74, "one"], [5.74, 6.27, "two"], [8.23, 8.81, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.27,\"end\":4.74},{\"word\":\"two\",\"start\":5.74,\"end\":6.27},{\"word\":\"three\",\"start\":8.23,\"end\":8.81}]"}
{"utterance_id": "1673-143396-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0013.flac", "task_type": "lexical_detection", "query": "expire", "prompt": "Once the audio has finished, identify every instance of the word \"expire\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[14.52, 15.21, "expire"]], "response_mode": "offline", "answer": "[{\"word\":\"expire\",\"start\":14.52,\"end\":15.21}]"}
{"utterance_id": "3752-4943-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.04, "christ"]], "response_mode": "offline", "answer": "[{\"word\":\"christ\",\"start\":1.44,\"end\":2.04}]"}
{"utterance_id": "8842-302196-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0003.flac", "task_type": "lexical_counting", "query": "fresh", "prompt": "When the utterance is complete, report how the word \"fresh\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 2.83, "one"], [4.79, 5.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.5,\"end\":2.83},{\"word\":\"two\",\"start\":4.79,\"end\":5.12}]"}
{"utterance_id": "6345-93302-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0001.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[0.52, 0.98, "one"], [4.44, 4.78, "two"], [7.69, 8.22, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.98},{\"word\":\"two\",\"start\":4.44,\"end\":4.78},{\"word\":\"three\",\"start\":7.69,\"end\":8.22}]"}
{"utterance_id": "2428-83705-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0037.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.04, 8.64, "snellings"]], "response_mode": "offline", "answer": "[{\"word\":\"snellings\",\"start\":8.04,\"end\":8.64}]"}
{"utterance_id": "777-126732-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0066.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.22, 2.5, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":2.22,\"end\":2.5}]"}
{"utterance_id": "6345-64257-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0005.flac", "task_type": "lexical_detection", "query": "beautiful", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"beautiful\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.67, 5.19, "beautiful"]], "response_mode": "offline", "answer": "[{\"word\":\"beautiful\",\"start\":4.67,\"end\":5.19}]"}
{"utterance_id": "3752-4943-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0003.flac", "task_type": "lexical_detection", "query": "wretched", "prompt": "After the audio ends, extract all occurrences of the word \"wretched\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.47, 2.85, "wretched"]], "response_mode": "offline", "answer": "[{\"word\":\"wretched\",\"start\":2.47,\"end\":2.85}]"}
{"utterance_id": "2803-154328-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0016.flac", "task_type": "lexical_detection", "query": "object", "prompt": "After the audio ends, extract all occurrences of the word \"object\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 2.01, "object"]], "response_mode": "offline", "answer": "[{\"word\":\"object\",\"start\":1.21,\"end\":2.01}]"}
{"utterance_id": "3752-4943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.16, 6.49, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":6.16,\"end\":6.49}]"}
{"utterance_id": "6313-66129-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.94, 11.42, "flanks"]], "response_mode": "offline", "answer": "[{\"word\":\"flanks\",\"start\":10.94,\"end\":11.42}]"}
{"utterance_id": "5694-64029-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0027.flac", "task_type": "lexical_detection", "query": "pretty", "prompt": "After listening to the whole recording, output every instance of the word \"pretty\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.22, "pretty"]], "response_mode": "offline", "answer": "[{\"word\":\"pretty\",\"start\":1.02,\"end\":1.22}]"}
{"utterance_id": "2277-149897-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.03, 2.43, "manager"]], "response_mode": "offline", "answer": "[{\"word\":\"manager\",\"start\":2.03,\"end\":2.43}]"}
{"utterance_id": "3752-4943-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0017.flac", "task_type": "lexical_detection", "query": "hard", "prompt": "Once the audio has finished, identify every instance of the word \"hard\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.46, 0.82, "hard"]], "response_mode": "offline", "answer": "[{\"word\":\"hard\",\"start\":0.46,\"end\":0.82}]"}
{"utterance_id": "7850-281318-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0019.flac", "task_type": "lexical_detection", "query": "finish", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"finish\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.6, "finish"]], "response_mode": "offline", "answer": "[{\"word\":\"finish\",\"start\":3.26,\"end\":3.6}]"}
{"utterance_id": "6319-275224-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0018.flac", "task_type": "lexical_detection", "query": "hurried", "prompt": "Once the audio has finished, identify every instance of the word \"hurried\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.27, 4.66, "hurried"]], "response_mode": "offline", "answer": "[{\"word\":\"hurried\",\"start\":4.27,\"end\":4.66}]"}
{"utterance_id": "1988-148538-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0015.flac", "task_type": "lexical_counting", "query": "same", "prompt": "Once the audio has finished, list each occurrence of the word \"same\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[3.93, 4.31, "one"], [8.15, 8.45, "two"], [12.93, 13.22, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.93,\"end\":4.31},{\"word\":\"two\",\"start\":8.15,\"end\":8.45},{\"word\":\"three\",\"start\":12.93,\"end\":13.22}]"}
{"utterance_id": "84-121123-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0024.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the full audio. Count every occurrence of a body part. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.64, 8.08, "one"], [9.56, 10.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.64,\"end\":8.08},{\"word\":\"two\",\"start\":9.56,\"end\":10.0}]"}
{"utterance_id": "7850-281318-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0009.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.51, 4.98, "elm tree"]], "response_mode": "offline", "answer": "[{\"word\":\"elm tree\",\"start\":4.51,\"end\":4.98}]"}
{"utterance_id": "2428-83699-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.03, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":1.69,\"end\":2.03}]"}
{"utterance_id": "5694-64038-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0013.flac", "task_type": "lexical_detection", "query": "coast", "prompt": "After the audio ends, extract all occurrences of the word \"coast\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.4, "coast"]], "response_mode": "offline", "answer": "[{\"word\":\"coast\",\"start\":2.06,\"end\":2.4}]"}
{"utterance_id": "8842-304647-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0000.flac", "task_type": "lexical_counting", "query": "thy", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"thy\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.95, 1.15, "one"], [7.1, 7.28, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.95,\"end\":1.15},{\"word\":\"two\",\"start\":7.1,\"end\":7.28}]"}
{"utterance_id": "6313-76958-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0013.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an animal name with word/start/end. If none, reply [].", "labels": [[0.38, 0.78, "one"], [9.34, 9.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.38,\"end\":0.78},{\"word\":\"two\",\"start\":9.34,\"end\":9.8}]"}
{"utterance_id": "6345-93306-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[12.34, 12.69, "nine"], [19.21, 19.42, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"nine\",\"start\":12.34,\"end\":12.69},{\"word\":\"two\",\"start\":19.21,\"end\":19.42}]"}
{"utterance_id": "8297-275154-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0008.flac", "task_type": "lexical_detection", "query": "distress", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"distress\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.71, 1.34, "distress"]], "response_mode": "offline", "answer": "[{\"word\":\"distress\",\"start\":0.71,\"end\":1.34}]"}
{"utterance_id": "8297-275156-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0009.flac", "task_type": "lexical_detection", "query": "delicacy", "prompt": "Process the full audio offline. Find every instance of the word \"delicacy\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 2.29, "delicacy"]], "response_mode": "offline", "answer": "[{\"word\":\"delicacy\",\"start\":1.57,\"end\":2.29}]"}
{"utterance_id": "777-126732-0063", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0063.flac", "task_type": "lexical_detection", "query": "footsteps", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"footsteps\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.56, "footsteps"]], "response_mode": "offline", "answer": "[{\"word\":\"footsteps\",\"start\":2.98,\"end\":3.56}]"}
{"utterance_id": "251-136532-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.96, "martha"]], "response_mode": "offline", "answer": "[{\"word\":\"martha\",\"start\":0.42,\"end\":0.96}]"}
{"utterance_id": "2078-142845-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0005.flac", "task_type": "lexical_detection", "query": "afterwards", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"afterwards\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.17, 9.79, "afterwards"]], "response_mode": "offline", "answer": "[{\"word\":\"afterwards\",\"start\":9.17,\"end\":9.79}]"}
{"utterance_id": "6295-244435-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0009.flac", "task_type": "lexical_detection", "query": "volunteers", "prompt": "After listening to the whole recording, output every instance of the word \"volunteers\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 2.05, "volunteers"]], "response_mode": "offline", "answer": "[{\"word\":\"volunteers\",\"start\":1.23,\"end\":2.05}]"}
{"utterance_id": "174-50561-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.03, 1.44, "singer"]], "response_mode": "offline", "answer": "[{\"word\":\"singer\",\"start\":1.03,\"end\":1.44}]"}
{"utterance_id": "2803-154328-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0020.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 4.08, "one"], [7.49, 8.24, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.03,\"end\":4.08},{\"word\":\"two\",\"start\":7.49,\"end\":8.24}]"}
{"utterance_id": "7976-105575-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0022.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.38, 5.05, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":4.38,\"end\":5.05}]"}
{"utterance_id": "6319-64726-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0016.flac", "task_type": "lexical_detection", "query": "extremely", "prompt": "Once the audio has finished, identify every instance of the word \"extremely\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.27, 7.91, "extremely"]], "response_mode": "offline", "answer": "[{\"word\":\"extremely\",\"start\":7.27,\"end\":7.91}]"}
{"utterance_id": "1462-170138-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0011.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.53, 10.13, "primroses"]], "response_mode": "offline", "answer": "[{\"word\":\"primroses\",\"start\":9.53,\"end\":10.13}]"}
{"utterance_id": "3853-163249-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.37, "david"], [7.34, 7.62, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.86,\"end\":1.37},{\"word\":\"god\",\"start\":7.34,\"end\":7.62}]"}
{"utterance_id": "3081-166546-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0011.flac", "task_type": "lexical_detection", "query": "sooner", "prompt": "Once the audio has finished, identify every instance of the word \"sooner\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.42, 4.97, "sooner"]], "response_mode": "offline", "answer": "[{\"word\":\"sooner\",\"start\":4.42,\"end\":4.97}]"}
{"utterance_id": "6241-61943-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0010.flac", "task_type": "lexical_detection", "query": "schools", "prompt": "After the audio ends, extract all occurrences of the word \"schools\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.0, "schools"]], "response_mode": "offline", "answer": "[{\"word\":\"schools\",\"start\":2.46,\"end\":3.0}]"}
{"utterance_id": "84-121550-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.12, 5.47, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":5.12,\"end\":5.47}]"}
{"utterance_id": "5694-64029-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.44, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":1.01,\"end\":1.44}]"}
{"utterance_id": "422-122949-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0033.flac", "task_type": "lexical_detection", "query": "table", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"table\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.67, 5.11, "table"]], "response_mode": "offline", "answer": "[{\"word\":\"table\",\"start\":4.67,\"end\":5.11}]"}
{"utterance_id": "2803-154328-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.59, "lady helena"]], "response_mode": "offline", "answer": "[{\"word\":\"lady helena\",\"start\":2.81,\"end\":3.59}]"}
{"utterance_id": "6319-64726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.81, "one"], [12.9, 13.13, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.81},{\"word\":\"one\",\"start\":12.9,\"end\":13.13}]"}
{"utterance_id": "6241-61946-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.6, 2.02, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":1.6,\"end\":2.02}]"}
{"utterance_id": "6313-66125-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0018.flac", "task_type": "lexical_detection", "query": "lige", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"lige\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 0.78, "lige"]], "response_mode": "offline", "answer": "[{\"word\":\"lige\",\"start\":0.35,\"end\":0.78}]"}
{"utterance_id": "6313-76958-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0018.flac", "task_type": "lexical_detection", "query": "indignation", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"indignation\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.31, 2.95, "indignation"]], "response_mode": "offline", "answer": "[{\"word\":\"indignation\",\"start\":2.31,\"end\":2.95}]"}
{"utterance_id": "174-50561-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0014.flac", "task_type": "lexical_detection", "query": "ladies", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"ladies\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.27, "ladies"]], "response_mode": "offline", "answer": "[{\"word\":\"ladies\",\"start\":0.69,\"end\":1.27}]"}
{"utterance_id": "2086-149220-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0000.flac", "task_type": "lexical_detection", "query": "formerly", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"formerly\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.4, "formerly"]], "response_mode": "offline", "answer": "[{\"word\":\"formerly\",\"start\":1.01,\"end\":1.4}]"}
{"utterance_id": "7976-110523-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0007.flac", "task_type": "lexical_detection", "query": "heavy", "prompt": "After listening to the whole recording, output every instance of the word \"heavy\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.12, "heavy"]], "response_mode": "offline", "answer": "[{\"word\":\"heavy\",\"start\":1.77,\"end\":2.12}]"}
{"utterance_id": "174-168635-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0018.flac", "task_type": "lexical_counting", "query": "memory", "prompt": "Listen to the entire clip, then report the count sequence for the word \"memory\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.82, 13.29, "one"], [24.93, 25.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":12.82,\"end\":13.29},{\"word\":\"two\",\"start\":24.93,\"end\":25.47}]"}
{"utterance_id": "7976-110523-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0019.flac", "task_type": "lexical_detection", "query": "oven", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"oven\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.11, 2.42, "oven"]], "response_mode": "offline", "answer": "[{\"word\":\"oven\",\"start\":2.11,\"end\":2.42}]"}
{"utterance_id": "6313-66125-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.35, 0.78, "lige"]], "response_mode": "offline", "answer": "[{\"word\":\"lige\",\"start\":0.35,\"end\":0.78}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.14, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":1.79,\"end\":2.14}]"}
{"utterance_id": "1993-147149-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.1, "alice"]], "response_mode": "offline", "answer": "[{\"word\":\"alice\",\"start\":0.66,\"end\":1.1}]"}
{"utterance_id": "2428-83705-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0005.flac", "task_type": "lexical_detection", "query": "instance", "prompt": "After listening to the whole recording, output every instance of the word \"instance\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.89, "instance"]], "response_mode": "offline", "answer": "[{\"word\":\"instance\",\"start\":0.41,\"end\":0.89}]"}
{"utterance_id": "8842-302201-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.2, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.75,\"end\":2.2}]"}
{"utterance_id": "7850-281318-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0002.flac", "task_type": "lexical_detection", "query": "tell", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"tell\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.54, "tell"]], "response_mode": "offline", "answer": "[{\"word\":\"tell\",\"start\":3.26,\"end\":3.54}]"}
{"utterance_id": "7850-73752-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0005.flac", "task_type": "lexical_detection", "query": "indeed", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"indeed\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.58, 0.91, "indeed"]], "response_mode": "offline", "answer": "[{\"word\":\"indeed\",\"start\":0.58,\"end\":0.91}]"}
{"utterance_id": "174-168635-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0007.flac", "task_type": "lexical_detection", "query": "hovel", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hovel\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.83, 9.22, "hovel"]], "response_mode": "offline", "answer": "[{\"word\":\"hovel\",\"start\":8.83,\"end\":9.22}]"}
{"utterance_id": "3752-4943-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.53, "mister north"]], "response_mode": "offline", "answer": "[{\"word\":\"mister north\",\"start\":1.76,\"end\":2.53}]"}
{"utterance_id": "2428-83705-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0041.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.32, 0.76, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.32,\"end\":0.76}]"}
{"utterance_id": "777-126732-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0023.flac", "task_type": "lexical_detection", "query": "personally", "prompt": "After listening to the whole recording, output every instance of the word \"personally\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.33, 3.75, "personally"]], "response_mode": "offline", "answer": "[{\"word\":\"personally\",\"start\":3.33,\"end\":3.75}]"}
{"utterance_id": "8842-302201-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0011.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.9, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"third\",\"start\":0.49,\"end\":0.9}]"}
{"utterance_id": "8297-275154-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0001.flac", "task_type": "lexical_detection", "query": "master", "prompt": "Process the full audio offline. Find every instance of the word \"master\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.65, "master"]], "response_mode": "offline", "answer": "[{\"word\":\"master\",\"start\":1.11,\"end\":1.65}]"}
{"utterance_id": "2277-149897-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0005.flac", "task_type": "lexical_detection", "query": "denial", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"denial\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.26, 5.77, "denial"]], "response_mode": "offline", "answer": "[{\"word\":\"denial\",\"start\":5.26,\"end\":5.77}]"}
{"utterance_id": "84-121123-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.1, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"five\",\"start\":0.87,\"end\":1.1}]"}
{"utterance_id": "6319-64726-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.08, 1.54, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.08,\"end\":1.54}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 3.32, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.71,\"end\":3.32}]"}
{"utterance_id": "652-130726-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0009.flac", "task_type": "lexical_detection", "query": "top", "prompt": "After the audio ends, extract all occurrences of the word \"top\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.7, 2.96, "top"]], "response_mode": "offline", "answer": "[{\"word\":\"top\",\"start\":2.7,\"end\":2.96}]"}
{"utterance_id": "1673-143397-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[1.05, 1.54, "bishops"]], "response_mode": "offline", "answer": "[{\"word\":\"bishops\",\"start\":1.05,\"end\":1.54}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.77, 14.29, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":13.77,\"end\":14.29}]"}
{"utterance_id": "5895-34629-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0010.flac", "task_type": "lexical_detection", "query": "performances", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"performances\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.07, 1.99, "performances"]], "response_mode": "offline", "answer": "[{\"word\":\"performances\",\"start\":1.07,\"end\":1.99}]"}
{"utterance_id": "3000-15664-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0018.flac", "task_type": "lexical_detection", "query": "glacier", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"glacier\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.18, 2.76, "glacier"]], "response_mode": "offline", "answer": "[{\"word\":\"glacier\",\"start\":2.18,\"end\":2.76}]"}
{"utterance_id": "422-122949-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0014.flac", "task_type": "lexical_counting", "query": "new", "prompt": "Listen to the entire clip, then report the count sequence for the word \"new\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.01, "one"], [2.36, 2.51, "two"], [22.22, 22.44, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.82,\"end\":1.01},{\"word\":\"two\",\"start\":2.36,\"end\":2.51},{\"word\":\"three\",\"start\":22.22,\"end\":22.44}]"}
{"utterance_id": "7850-73752-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0006.flac", "task_type": "lexical_detection", "query": "waters", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"waters\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.71, 7.18, "waters"]], "response_mode": "offline", "answer": "[{\"word\":\"waters\",\"start\":6.71,\"end\":7.18}]"}
{"utterance_id": "2078-142845-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0002.flac", "task_type": "lexical_detection", "query": "bread", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"bread\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.97, 2.64, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":1.97,\"end\":2.64}]"}
{"utterance_id": "1988-24833-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0003.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.89, "cats"], [4.03, 4.31, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cats\",\"start\":1.29,\"end\":1.89},{\"word\":\"cat\",\"start\":4.03,\"end\":4.31}]"}
{"utterance_id": "3853-163249-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.98, 6.56, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":5.98,\"end\":6.56}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "lexical_detection", "query": "field", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"field\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.47, 7.98, "field"]], "response_mode": "offline", "answer": "[{\"word\":\"field\",\"start\":7.47,\"end\":7.98}]"}
{"utterance_id": "3000-15664-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.97, 1.61, "profession"]], "response_mode": "offline", "answer": "[{\"word\":\"profession\",\"start\":0.97,\"end\":1.61}]"}
{"utterance_id": "1272-135031-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0013.flac", "task_type": "lexical_detection", "query": "metal", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"metal\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.71, "metal"]], "response_mode": "offline", "answer": "[{\"word\":\"metal\",\"start\":2.38,\"end\":2.71}]"}
{"utterance_id": "5338-24615-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 2.47, "captain waverley"]], "response_mode": "offline", "answer": "[{\"word\":\"captain waverley\",\"start\":1.49,\"end\":2.47}]"}
{"utterance_id": "3081-166546-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0064.flac", "task_type": "lexical_detection", "query": "sweetwater", "prompt": "Once the audio has finished, identify every instance of the word \"sweetwater\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.35, 1.16, "sweetwater"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":0.35,\"end\":1.16}]"}
{"utterance_id": "1673-143397-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0012.flac", "task_type": "lexical_detection", "query": "bishops", "prompt": "Process the full audio offline. Find every instance of the word \"bishops\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.54, "bishops"]], "response_mode": "offline", "answer": "[{\"word\":\"bishops\",\"start\":1.05,\"end\":1.54}]"}
{"utterance_id": "1462-170145-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0005.flac", "task_type": "lexical_detection", "query": "boots", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"boots\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.37, "boots"]], "response_mode": "offline", "answer": "[{\"word\":\"boots\",\"start\":1.93,\"end\":2.37}]"}
{"utterance_id": "3081-166546-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0043.flac", "task_type": "lexical_detection", "query": "fellow", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"fellow\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.0, 1.47, "fellow"]], "response_mode": "offline", "answer": "[{\"word\":\"fellow\",\"start\":1.0,\"end\":1.47}]"}
{"utterance_id": "2803-161169-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0014.flac", "task_type": "lexical_detection", "query": "useful", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"useful\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.18, 6.61, "useful"]], "response_mode": "offline", "answer": "[{\"word\":\"useful\",\"start\":6.18,\"end\":6.61}]"}
{"utterance_id": "2035-147960-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.1, 2.64, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":2.1,\"end\":2.64}]"}
{"utterance_id": "3536-23268-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0004.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[1.92, 2.23, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":1.92,\"end\":2.23}]"}
{"utterance_id": "5895-34615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0004.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a body part: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[8.98, 9.51, "one"], [13.69, 14.41, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.98,\"end\":9.51},{\"word\":\"two\",\"start\":13.69,\"end\":14.41}]"}
{"utterance_id": "6319-275224-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.47, 3.84, "fir"]], "response_mode": "offline", "answer": "[{\"word\":\"fir\",\"start\":3.47,\"end\":3.84}]"}
{"utterance_id": "2803-161169-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0007.flac", "task_type": "lexical_detection", "query": "trunks", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"trunks\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.15, 4.83, "trunks"]], "response_mode": "offline", "answer": "[{\"word\":\"trunks\",\"start\":4.15,\"end\":4.83}]"}
{"utterance_id": "1919-142785-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0049.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.22, 11.64, "udder"]], "response_mode": "offline", "answer": "[{\"word\":\"udder\",\"start\":11.22,\"end\":11.64}]"}
{"utterance_id": "2277-149896-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0001.flac", "task_type": "semantic_detection", "query": "money", "prompt": "After listening to the whole recording, output every instance of a money or currency term with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.24, 1.6, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":1.24,\"end\":1.6}]"}
{"utterance_id": "2277-149874-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0012.flac", "task_type": "lexical_detection", "query": "minnie", "prompt": "After listening to the whole recording, output every instance of the word \"minnie\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.76, "minnie"]], "response_mode": "offline", "answer": "[{\"word\":\"minnie\",\"start\":0.46,\"end\":0.76}]"}
{"utterance_id": "6313-66125-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0023.flac", "task_type": "lexical_detection", "query": "rescue", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"rescue\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.39, 7.8, "rescue"]], "response_mode": "offline", "answer": "[{\"word\":\"rescue\",\"start\":7.39,\"end\":7.8}]"}
{"utterance_id": "2035-152373-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0013.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.98, 5.45, "iona"], [6.82, 7.33, "tara"]], "response_mode": "offline", "answer": "[{\"word\":\"iona\",\"start\":4.98,\"end\":5.45},{\"word\":\"tara\",\"start\":6.82,\"end\":7.33}]"}
{"utterance_id": "3853-163249-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0055.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.5, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.96,\"end\":1.5}]"}
{"utterance_id": "1919-142785-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0020.flac", "task_type": "lexical_detection", "query": "preserved", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"preserved\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 4.12, "preserved"]], "response_mode": "offline", "answer": "[{\"word\":\"preserved\",\"start\":3.59,\"end\":4.12}]"}
{"utterance_id": "6345-64257-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0001.flac", "task_type": "lexical_detection", "query": "detail", "prompt": "When the utterance is complete, report all instances of the word \"detail\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.13, 7.68, "detail"]], "response_mode": "offline", "answer": "[{\"word\":\"detail\",\"start\":7.13,\"end\":7.68}]"}
{"utterance_id": "6313-66125-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0027.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[5.36, 5.82, "one"], [8.54, 8.81, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.36,\"end\":5.82},{\"word\":\"two\",\"start\":8.54,\"end\":8.81}]"}
{"utterance_id": "8297-275155-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0003.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 1.93, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.55,\"end\":1.93}]"}
{"utterance_id": "3752-4943-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0000.flac", "task_type": "lexical_detection", "query": "transported", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"transported\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.65, "transported"]], "response_mode": "offline", "answer": "[{\"word\":\"transported\",\"start\":2.9,\"end\":3.65}]"}
{"utterance_id": "1988-148538-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.64, "plato"]], "response_mode": "offline", "answer": "[{\"word\":\"plato\",\"start\":3.1,\"end\":3.64}]"}
{"utterance_id": "6313-66125-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0014.flac", "task_type": "lexical_detection", "query": "movement", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"movement\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.36, 0.77, "movement"]], "response_mode": "offline", "answer": "[{\"word\":\"movement\",\"start\":0.36,\"end\":0.77}]"}
{"utterance_id": "3081-166546-0087", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0087.flac", "task_type": "lexical_detection", "query": "protruded", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"protruded\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.84, "protruded"]], "response_mode": "offline", "answer": "[{\"word\":\"protruded\",\"start\":1.17,\"end\":1.84}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.91, 9.41, "husband's"]], "response_mode": "offline", "answer": "[{\"word\":\"husband's\",\"start\":8.91,\"end\":9.41}]"}
{"utterance_id": "3170-137482-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.68, 10.22, "chest"]], "response_mode": "offline", "answer": "[{\"word\":\"chest\",\"start\":9.68,\"end\":10.22}]"}
{"utterance_id": "7976-110124-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0020.flac", "task_type": "lexical_detection", "query": "father's", "prompt": "When the utterance is complete, report all instances of the word \"father's\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.68, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"father's\",\"start\":1.28,\"end\":1.68}]"}
{"utterance_id": "6319-64726-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.03, 2.59, "rosy"]], "response_mode": "offline", "answer": "[{\"word\":\"rosy\",\"start\":2.03,\"end\":2.59}]"}
{"utterance_id": "1673-143397-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0009.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the full audio. After it finishes, find every instance of a country name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.89, 3.35, "egypt"]], "response_mode": "offline", "answer": "[{\"word\":\"egypt\",\"start\":2.89,\"end\":3.35}]"}
{"utterance_id": "422-122949-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0015.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.56, 2.91, "mother"], [10.69, 11.22, "child"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":2.56,\"end\":2.91},{\"word\":\"child\",\"start\":10.69,\"end\":11.22}]"}
{"utterance_id": "1272-135031-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0004.flac", "task_type": "lexical_detection", "query": "disgrace", "prompt": "After listening to the whole recording, output every instance of the word \"disgrace\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.94, "disgrace"]], "response_mode": "offline", "answer": "[{\"word\":\"disgrace\",\"start\":1.34,\"end\":1.94}]"}
{"utterance_id": "777-126732-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0068.flac", "task_type": "lexical_detection", "query": "talked", "prompt": "When the utterance is complete, report all instances of the word \"talked\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.68, 1.97, "talked"]], "response_mode": "offline", "answer": "[{\"word\":\"talked\",\"start\":1.68,\"end\":1.97}]"}
{"utterance_id": "5895-34629-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0017.flac", "task_type": "lexical_detection", "query": "remained", "prompt": "After the audio ends, extract all occurrences of the word \"remained\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.53, 3.93, "remained"]], "response_mode": "offline", "answer": "[{\"word\":\"remained\",\"start\":3.53,\"end\":3.93}]"}
{"utterance_id": "7976-110124-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0001.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Once the audio has finished, identify every instance of a money or currency term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.87, 6.23, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":5.87,\"end\":6.23}]"}
{"utterance_id": "6313-66129-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0002.flac", "task_type": "lexical_detection", "query": "nourishing", "prompt": "After the audio ends, extract all occurrences of the word \"nourishing\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.69, "nourishing"]], "response_mode": "offline", "answer": "[{\"word\":\"nourishing\",\"start\":1.27,\"end\":1.69}]"}
{"utterance_id": "2035-147961-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0011.flac", "task_type": "lexical_counting", "query": "bright", "prompt": "Once the audio has finished, list each occurrence of the word \"bright\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.69, 1.96, "one"], [4.98, 5.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.69,\"end\":1.96},{\"word\":\"two\",\"start\":4.98,\"end\":5.38}]"}
{"utterance_id": "174-50561-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0008.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.01, 15.32, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":15.01,\"end\":15.32}]"}
{"utterance_id": "251-137823-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0013.flac", "task_type": "lexical_detection", "query": "we'll", "prompt": "When the utterance is complete, report all instances of the word \"we'll\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.21, 3.52, "we'll"]], "response_mode": "offline", "answer": "[{\"word\":\"we'll\",\"start\":3.21,\"end\":3.52}]"}
{"utterance_id": "1462-170142-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.83, 3.5, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander\",\"start\":2.83,\"end\":3.5}]"}
{"utterance_id": "3170-137482-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0006.flac", "task_type": "lexical_detection", "query": "enjoying", "prompt": "After listening to the whole recording, output every instance of the word \"enjoying\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.22, 5.72, "enjoying"]], "response_mode": "offline", "answer": "[{\"word\":\"enjoying\",\"start\":5.22,\"end\":5.72}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "semantic_counting", "query": "country", "prompt": "Once the audio has finished, list each occurrence of a country name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[12.61, 13.0, "one"], [18.75, 19.02, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":12.61,\"end\":13.0},{\"word\":\"two\",\"start\":18.75,\"end\":19.02}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.92, 5.6, "abbess"]], "response_mode": "offline", "answer": "[{\"word\":\"abbess\",\"start\":4.92,\"end\":5.6}]"}
{"utterance_id": "5536-43358-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0002.flac", "task_type": "lexical_detection", "query": "imperfect", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"imperfect\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.94, 4.52, "imperfect"]], "response_mode": "offline", "answer": "[{\"word\":\"imperfect\",\"start\":3.94,\"end\":4.52}]"}
{"utterance_id": "6313-76958-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0024.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.26, "wrangler"], [2.38, 2.83, "foreman"]], "response_mode": "offline", "answer": "[{\"word\":\"wrangler\",\"start\":0.91,\"end\":1.26},{\"word\":\"foreman\",\"start\":2.38,\"end\":2.83}]"}
{"utterance_id": "174-168635-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.52, "fifty"]], "response_mode": "offline", "answer": "[{\"word\":\"fifty\",\"start\":2.07,\"end\":2.52}]"}
{"utterance_id": "84-121550-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0010.flac", "task_type": "lexical_detection", "query": "unto", "prompt": "After listening to the whole recording, output every instance of the word \"unto\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.57, 4.84, "unto"]], "response_mode": "offline", "answer": "[{\"word\":\"unto\",\"start\":4.57,\"end\":4.84}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 1.84, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.5,\"end\":1.84}]"}
{"utterance_id": "2428-83705-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0021.flac", "task_type": "lexical_detection", "query": "papa", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"papa\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.05, 1.37, "papa"]], "response_mode": "offline", "answer": "[{\"word\":\"papa\",\"start\":1.05,\"end\":1.37}]"}
{"utterance_id": "7850-286674-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0010.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.4, 1.57, "one"], [7.02, 7.28, "two"], [8.76, 9.33, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.4,\"end\":1.57},{\"word\":\"two\",\"start\":7.02,\"end\":7.28},{\"word\":\"three\",\"start\":8.76,\"end\":9.33}]"}
{"utterance_id": "6241-61943-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0011.flac", "task_type": "lexical_detection", "query": "population", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"population\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.09, 1.71, "population"]], "response_mode": "offline", "answer": "[{\"word\":\"population\",\"start\":1.09,\"end\":1.71}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.42, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":1.17,\"end\":1.42}]"}
{"utterance_id": "1272-135031-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0024.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 3.15, "one"], [6.83, 7.41, "two"], [11.26, 11.66, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.62,\"end\":3.15},{\"word\":\"two\",\"start\":6.83,\"end\":7.41},{\"word\":\"three\",\"start\":11.26,\"end\":11.66}]"}
{"utterance_id": "1673-143396-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0004.flac", "task_type": "lexical_detection", "query": "religion", "prompt": "Process the full audio offline. Find every instance of the word \"religion\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.0, 7.36, "religion"]], "response_mode": "offline", "answer": "[{\"word\":\"religion\",\"start\":7.0,\"end\":7.36}]"}
{"utterance_id": "174-168635-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0015.flac", "task_type": "lexical_detection", "query": "pray", "prompt": "After listening to the whole recording, output every instance of the word \"pray\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.74, "pray"]], "response_mode": "offline", "answer": "[{\"word\":\"pray\",\"start\":3.24,\"end\":3.74}]"}
{"utterance_id": "652-130726-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0020.flac", "task_type": "lexical_detection", "query": "market", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"market\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.23, 1.56, "market"]], "response_mode": "offline", "answer": "[{\"word\":\"market\",\"start\":1.23,\"end\":1.56}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[21.03, 21.17, "one"], [26.29, 26.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":21.03,\"end\":21.17},{\"word\":\"two\",\"start\":26.29,\"end\":26.9}]"}
{"utterance_id": "6295-244435-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.05, 5.73, "charleston"]], "response_mode": "offline", "answer": "[{\"word\":\"charleston\",\"start\":5.05,\"end\":5.73}]"}
{"utterance_id": "2428-83705-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0006.flac", "task_type": "lexical_detection", "query": "accident", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"accident\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.66, 1.14, "accident"]], "response_mode": "offline", "answer": "[{\"word\":\"accident\",\"start\":0.66,\"end\":1.14}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "When the utterance is complete, report how a plant name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[21.95, 22.42, "one"], [25.99, 26.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":21.95,\"end\":22.42},{\"word\":\"two\",\"start\":25.99,\"end\":26.54}]"}
{"utterance_id": "174-84280-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0008.flac", "task_type": "lexical_detection", "query": "faith", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"faith\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.47, 3.88, "faith"]], "response_mode": "offline", "answer": "[{\"word\":\"faith\",\"start\":3.47,\"end\":3.88}]"}
{"utterance_id": "6241-66616-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0011.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.17, 6.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":6.17,\"end\":6.34}]"}
{"utterance_id": "174-50561-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0016.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[12.62, 13.15, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":12.62,\"end\":13.15}]"}
{"utterance_id": "2078-142845-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0042.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 3.83, "seventeen thirty two"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen thirty two\",\"start\":2.16,\"end\":3.83}]"}
{"utterance_id": "6241-61943-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0012.flac", "task_type": "lexical_detection", "query": "vessel", "prompt": "Once the audio has finished, identify every instance of the word \"vessel\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.02, 5.49, "vessel"]], "response_mode": "offline", "answer": "[{\"word\":\"vessel\",\"start\":5.02,\"end\":5.49}]"}
{"utterance_id": "3752-4944-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0010.flac", "task_type": "lexical_detection", "query": "married", "prompt": "Process the full audio offline. Find every instance of the word \"married\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.12, "married"]], "response_mode": "offline", "answer": "[{\"word\":\"married\",\"start\":0.75,\"end\":1.12}]"}
{"utterance_id": "652-130737-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0002.flac", "task_type": "lexical_counting", "query": "wines", "prompt": "After the audio ends, tally every instance of the word \"wines\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.86, 3.35, "one"], [6.59, 7.05, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.86,\"end\":3.35},{\"word\":\"two\",\"start\":6.59,\"end\":7.05}]"}
{"utterance_id": "6313-66129-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0014.flac", "task_type": "lexical_detection", "query": "tent", "prompt": "Once the audio has finished, identify every instance of the word \"tent\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.22, 7.52, "tent"]], "response_mode": "offline", "answer": "[{\"word\":\"tent\",\"start\":7.22,\"end\":7.52}]"}
{"utterance_id": "174-84280-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0005.flac", "task_type": "lexical_detection", "query": "presence", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"presence\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.0, 6.58, "presence"]], "response_mode": "offline", "answer": "[{\"word\":\"presence\",\"start\":6.0,\"end\":6.58}]"}
{"utterance_id": "6295-64301-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0004.flac", "task_type": "lexical_detection", "query": "earth", "prompt": "Process the full audio offline. Find every instance of the word \"earth\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.77, "earth"]], "response_mode": "offline", "answer": "[{\"word\":\"earth\",\"start\":0.5,\"end\":0.77}]"}
{"utterance_id": "174-50561-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0008.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.38, 4.87, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":4.38,\"end\":4.87}]"}
{"utterance_id": "5536-43363-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.73, 5.08, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":4.73,\"end\":5.08}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.33, "feast"], [7.34, 8.13, "breakfasts"]], "response_mode": "offline", "answer": "[{\"word\":\"feast\",\"start\":1.86,\"end\":2.33},{\"word\":\"breakfasts\",\"start\":7.34,\"end\":8.13}]"}
{"utterance_id": "6313-76958-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0014.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.99, 2.32, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":1.99,\"end\":2.32}]"}
{"utterance_id": "2078-142845-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0037.flac", "task_type": "lexical_counting", "query": "flour", "prompt": "After listening to the whole recording, output timed count words for the word \"flour\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.67, 2.25, "one"], [12.85, 13.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.67,\"end\":2.25},{\"word\":\"two\",\"start\":12.85,\"end\":13.37}]"}
{"utterance_id": "6241-66616-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.75, 4.2, "newsome"]], "response_mode": "offline", "answer": "[{\"word\":\"newsome\",\"start\":3.75,\"end\":4.2}]"}
{"utterance_id": "652-130726-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0034.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.96, 6.38, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":5.96,\"end\":6.38}]"}
{"utterance_id": "2277-149874-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0000.flac", "task_type": "lexical_detection", "query": "apartments", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"apartments\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.42, 2.97, "apartments"]], "response_mode": "offline", "answer": "[{\"word\":\"apartments\",\"start\":2.42,\"end\":2.97}]"}
{"utterance_id": "5338-24615-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 6.11, "tully veolan"]], "response_mode": "offline", "answer": "[{\"word\":\"tully veolan\",\"start\":5.14,\"end\":6.11}]"}
{"utterance_id": "2902-9008-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0007.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[4.51, 4.8, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":4.51,\"end\":4.8}]"}
{"utterance_id": "8842-304647-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0003.flac", "task_type": "lexical_detection", "query": "unworthy", "prompt": "Process the full audio offline. Find every instance of the word \"unworthy\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.68, 3.24, "unworthy"]], "response_mode": "offline", "answer": "[{\"word\":\"unworthy\",\"start\":2.68,\"end\":3.24}]"}
{"utterance_id": "5895-34622-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0023.flac", "task_type": "lexical_detection", "query": "piece", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"piece\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.8, "piece"]], "response_mode": "offline", "answer": "[{\"word\":\"piece\",\"start\":2.37,\"end\":2.8}]"}
{"utterance_id": "5536-43363-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.84, "many"], [2.02, 2.32, "one"], [3.44, 3.86, "once"]], "response_mode": "offline", "answer": "[{\"word\":\"many\",\"start\":0.57,\"end\":0.84},{\"word\":\"one\",\"start\":2.02,\"end\":2.32},{\"word\":\"once\",\"start\":3.44,\"end\":3.86}]"}
{"utterance_id": "1919-142785-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0010.flac", "task_type": "lexical_detection", "query": "parts", "prompt": "After listening to the whole recording, output every instance of the word \"parts\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 3.34, "parts"]], "response_mode": "offline", "answer": "[{\"word\":\"parts\",\"start\":3.03,\"end\":3.34}]"}
{"utterance_id": "2277-149896-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0029.flac", "task_type": "lexical_detection", "query": "thing", "prompt": "After listening to the whole recording, output every instance of the word \"thing\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.68, 2.08, "thing"]], "response_mode": "offline", "answer": "[{\"word\":\"thing\",\"start\":1.68,\"end\":2.08}]"}
{"utterance_id": "3752-4944-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.27, 10.64, "north"]], "response_mode": "offline", "answer": "[{\"word\":\"north\",\"start\":10.27,\"end\":10.64}]"}
{"utterance_id": "2078-142845-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0050.flac", "task_type": "lexical_detection", "query": "melted", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"melted\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.85, 7.44, "melted"]], "response_mode": "offline", "answer": "[{\"word\":\"melted\",\"start\":6.85,\"end\":7.44}]"}
{"utterance_id": "2277-149874-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.12, "minnie"], [6.94, 7.3, "drouet s"]], "response_mode": "offline", "answer": "[{\"word\":\"minnie\",\"start\":0.81,\"end\":1.12},{\"word\":\"drouet s\",\"start\":6.94,\"end\":7.3}]"}
{"utterance_id": "2803-154320-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.93, "god"], [3.65, 4.18, "john"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":0.42,\"end\":0.93},{\"word\":\"john\",\"start\":3.65,\"end\":4.18}]"}
{"utterance_id": "5694-64029-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0015.flac", "task_type": "lexical_detection", "query": "general", "prompt": "After listening to the whole recording, output every instance of the word \"general\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.27, "general"]], "response_mode": "offline", "answer": "[{\"word\":\"general\",\"start\":1.87,\"end\":2.27}]"}
{"utterance_id": "174-84280-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0012.flac", "task_type": "lexical_detection", "query": "net", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"net\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.74, 6.19, "net"]], "response_mode": "offline", "answer": "[{\"word\":\"net\",\"start\":5.74,\"end\":6.19}]"}
{"utterance_id": "5694-64025-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0001.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.33, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.97,\"end\":1.33}]"}
{"utterance_id": "84-121550-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0018.flac", "task_type": "lexical_detection", "query": "newly", "prompt": "When the utterance is complete, report all instances of the word \"newly\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.97, 8.38, "newly"]], "response_mode": "offline", "answer": "[{\"word\":\"newly\",\"start\":7.97,\"end\":8.38}]"}
{"utterance_id": "1673-143396-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.96, 7.5, "christ"]], "response_mode": "offline", "answer": "[{\"word\":\"christ\",\"start\":6.96,\"end\":7.5}]"}
{"utterance_id": "251-136532-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0017.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 1.37, "one"], [5.24, 5.79, "two"], [7.18, 7.72, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.36,\"end\":1.37},{\"word\":\"two\",\"start\":5.24,\"end\":5.79},{\"word\":\"three\",\"start\":7.18,\"end\":7.72}]"}
{"utterance_id": "84-121550-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.02, 1.8, "virgilius"]], "response_mode": "offline", "answer": "[{\"word\":\"virgilius\",\"start\":1.02,\"end\":1.8}]"}
{"utterance_id": "174-168635-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0020.flac", "task_type": "lexical_detection", "query": "walked", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"walked\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.62, 2.09, "walked"]], "response_mode": "offline", "answer": "[{\"word\":\"walked\",\"start\":1.62,\"end\":2.09}]"}
{"utterance_id": "5536-43359-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.12, 3.58, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":3.12,\"end\":3.58}]"}
{"utterance_id": "84-121123-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0027.flac", "task_type": "lexical_detection", "query": "excuse", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"excuse\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.18, "excuse"]], "response_mode": "offline", "answer": "[{\"word\":\"excuse\",\"start\":1.77,\"end\":2.18}]"}
{"utterance_id": "3752-4944-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0047.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the entire clip, then report the count sequence for a body part as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.97, 3.33, "one"], [6.33, 6.72, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.97,\"end\":3.33},{\"word\":\"two\",\"start\":6.33,\"end\":6.72}]"}
{"utterance_id": "3000-15664-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0042.flac", "task_type": "lexical_detection", "query": "lassen's", "prompt": "Once the audio has finished, identify every instance of the word \"lassen's\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.06, 1.56, "lassen's"]], "response_mode": "offline", "answer": "[{\"word\":\"lassen's\",\"start\":1.06,\"end\":1.56}]"}
{"utterance_id": "3752-4944-0069", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0069.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.91, 5.37, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":4.91,\"end\":5.37}]"}
{"utterance_id": "3000-15664-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0010.flac", "task_type": "lexical_detection", "query": "responsive", "prompt": "Once the audio has finished, identify every instance of the word \"responsive\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.65, 2.33, "responsive"]], "response_mode": "offline", "answer": "[{\"word\":\"responsive\",\"start\":1.65,\"end\":2.33}]"}
{"utterance_id": "2902-9008-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0013.flac", "task_type": "lexical_detection", "query": "creature", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"creature\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.56, "creature"]], "response_mode": "offline", "answer": "[{\"word\":\"creature\",\"start\":1.08,\"end\":1.56}]"}
{"utterance_id": "3752-4944-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 2.44, "john rex"]], "response_mode": "offline", "answer": "[{\"word\":\"john rex\",\"start\":1.5,\"end\":2.44}]"}
{"utterance_id": "3536-23268-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.58, "miss milner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss milner\",\"start\":0.94,\"end\":1.58}]"}
{"utterance_id": "6241-66616-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0018.flac", "task_type": "lexical_detection", "query": "wabi", "prompt": "When the utterance is complete, report all instances of the word \"wabi\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.84, 3.14, "wabi"]], "response_mode": "offline", "answer": "[{\"word\":\"wabi\",\"start\":2.84,\"end\":3.14}]"}
{"utterance_id": "5536-43358-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0009.flac", "task_type": "lexical_detection", "query": "fellows", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"fellows\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.32, 7.84, "fellows"]], "response_mode": "offline", "answer": "[{\"word\":\"fellows\",\"start\":7.32,\"end\":7.84}]"}
{"utterance_id": "6241-66616-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0000.flac", "task_type": "lexical_detection", "query": "revealed", "prompt": "When the utterance is complete, report all instances of the word \"revealed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.27, 8.78, "revealed"]], "response_mode": "offline", "answer": "[{\"word\":\"revealed\",\"start\":8.27,\"end\":8.78}]"}
{"utterance_id": "1993-147149-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0024.flac", "task_type": "lexical_detection", "query": "possesses", "prompt": "After the audio ends, extract all occurrences of the word \"possesses\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.45, "possesses"]], "response_mode": "offline", "answer": "[{\"word\":\"possesses\",\"start\":1.9,\"end\":2.45}]"}
{"utterance_id": "2803-161169-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0010.flac", "task_type": "lexical_detection", "query": "wood", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"wood\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.98, 4.47, "wood"]], "response_mode": "offline", "answer": "[{\"word\":\"wood\",\"start\":3.98,\"end\":4.47}]"}
{"utterance_id": "1919-142785-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0039.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.37, "marjoram"]], "response_mode": "offline", "answer": "[{\"word\":\"marjoram\",\"start\":1.78,\"end\":2.37}]"}
{"utterance_id": "6295-244435-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0028.flac", "task_type": "lexical_detection", "query": "sight", "prompt": "When the utterance is complete, report all instances of the word \"sight\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.39, "sight"]], "response_mode": "offline", "answer": "[{\"word\":\"sight\",\"start\":1.83,\"end\":2.39}]"}
{"utterance_id": "1673-143397-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0019.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Once the audio has finished, list each occurrence of a profession as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[6.38, 6.84, "one"], [14.17, 14.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.38,\"end\":6.84},{\"word\":\"two\",\"start\":14.17,\"end\":14.68}]"}
{"utterance_id": "5694-64029-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0014.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the full audio. After it finishes, find every instance of a date. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.0, 3.55, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":3.0,\"end\":3.55}]"}
{"utterance_id": "2086-149220-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0048.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.42, 4.89, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":4.42,\"end\":4.89}]"}
{"utterance_id": "2428-83699-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0013.flac", "task_type": "lexical_detection", "query": "bought", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bought\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.68, 0.91, "bought"]], "response_mode": "offline", "answer": "[{\"word\":\"bought\",\"start\":0.68,\"end\":0.91}]"}
{"utterance_id": "6313-66125-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.36, 5.82, "grasp"], [8.54, 8.81, "limbs"]], "response_mode": "offline", "answer": "[{\"word\":\"grasp\",\"start\":5.36,\"end\":5.82},{\"word\":\"limbs\",\"start\":8.54,\"end\":8.81}]"}
{"utterance_id": "5338-284437-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 1.6, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.48,\"end\":1.6}]"}
{"utterance_id": "777-126732-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.96, "prophet"]], "response_mode": "offline", "answer": "[{\"word\":\"prophet\",\"start\":2.44,\"end\":2.96}]"}
{"utterance_id": "2078-142845-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.22, 1.92, "seventeen seventeen"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen seventeen\",\"start\":0.22,\"end\":1.92}]"}
{"utterance_id": "2902-9006-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0017.flac", "task_type": "lexical_counting", "query": "gothic", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"gothic\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[10.93, 11.4, "one"], [15.65, 16.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.93,\"end\":11.4},{\"word\":\"two\",\"start\":15.65,\"end\":16.08}]"}
{"utterance_id": "5694-64038-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0017.flac", "task_type": "lexical_detection", "query": "emotions", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"emotions\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.81, 5.25, "emotions"]], "response_mode": "offline", "answer": "[{\"word\":\"emotions\",\"start\":4.81,\"end\":5.25}]"}
{"utterance_id": "6313-66129-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0029.flac", "task_type": "lexical_detection", "query": "leaning", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"leaning\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.24, 2.59, "leaning"]], "response_mode": "offline", "answer": "[{\"word\":\"leaning\",\"start\":2.24,\"end\":2.59}]"}
{"utterance_id": "5895-34615-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0001.flac", "task_type": "lexical_detection", "query": "perfect", "prompt": "When the utterance is complete, report all instances of the word \"perfect\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.19, "perfect"]], "response_mode": "offline", "answer": "[{\"word\":\"perfect\",\"start\":0.79,\"end\":1.19}]"}
{"utterance_id": "5694-64025-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0007.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.79, 3.44, "columbia"]], "response_mode": "offline", "answer": "[{\"word\":\"columbia\",\"start\":2.79,\"end\":3.44}]"}
{"utterance_id": "2277-149896-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.17, 3.49, "wife's"]], "response_mode": "offline", "answer": "[{\"word\":\"wife's\",\"start\":3.17,\"end\":3.49}]"}
{"utterance_id": "6319-275224-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.57, 11.12, "roots"]], "response_mode": "offline", "answer": "[{\"word\":\"roots\",\"start\":10.57,\"end\":11.12}]"}
{"utterance_id": "3000-15664-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0013.flac", "task_type": "lexical_detection", "query": "silence", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"silence\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.11, 8.79, "silence"]], "response_mode": "offline", "answer": "[{\"word\":\"silence\",\"start\":8.11,\"end\":8.79}]"}
{"utterance_id": "1673-143397-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0008.flac", "task_type": "lexical_detection", "query": "pillars", "prompt": "When the utterance is complete, report all instances of the word \"pillars\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.28, 2.72, "pillars"]], "response_mode": "offline", "answer": "[{\"word\":\"pillars\",\"start\":2.28,\"end\":2.72}]"}
{"utterance_id": "2277-149896-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0028.flac", "task_type": "lexical_detection", "query": "hat", "prompt": "When the utterance is complete, report all instances of the word \"hat\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.38, "hat"]], "response_mode": "offline", "answer": "[{\"word\":\"hat\",\"start\":1.06,\"end\":1.38}]"}
{"utterance_id": "3752-4944-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0011.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 0.65, "devil"]], "response_mode": "offline", "answer": "[{\"word\":\"devil\",\"start\":0.31,\"end\":0.65}]"}
{"utterance_id": "3752-4944-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0035.flac", "task_type": "lexical_detection", "query": "sleeps", "prompt": "When the utterance is complete, report all instances of the word \"sleeps\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.73, "sleeps"]], "response_mode": "offline", "answer": "[{\"word\":\"sleeps\",\"start\":0.36,\"end\":0.73}]"}
{"utterance_id": "3752-4944-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0003.flac", "task_type": "lexical_detection", "query": "colony", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"colony\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.36, 1.78, "colony"]], "response_mode": "offline", "answer": "[{\"word\":\"colony\",\"start\":1.36,\"end\":1.78}]"}
{"utterance_id": "2035-152373-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0004.flac", "task_type": "lexical_detection", "query": "kings", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"kings\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.36, "kings"]], "response_mode": "offline", "answer": "[{\"word\":\"kings\",\"start\":1.02,\"end\":1.36}]"}
{"utterance_id": "2078-142845-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0031.flac", "task_type": "lexical_counting", "query": "toast", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"toast\" (word, start, end). If none, reply [].", "labels": [[8.29, 8.64, "one"], [15.56, 16.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.29,\"end\":8.64},{\"word\":\"two\",\"start\":15.56,\"end\":16.2}]"}
{"utterance_id": "8842-304647-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0009.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.03, 10.64, "crickets"]], "response_mode": "offline", "answer": "[{\"word\":\"crickets\",\"start\":10.03,\"end\":10.64}]"}
{"utterance_id": "3752-4944-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0002.flac", "task_type": "lexical_detection", "query": "shadow", "prompt": "After the audio ends, extract all occurrences of the word \"shadow\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.57, "shadow"]], "response_mode": "offline", "answer": "[{\"word\":\"shadow\",\"start\":1.16,\"end\":1.57}]"}
{"utterance_id": "7976-110124-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0018.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.54, 1.88, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.54,\"end\":1.88}]"}
{"utterance_id": "84-121550-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.74, 6.2, "son"]], "response_mode": "offline", "answer": "[{\"word\":\"son\",\"start\":5.74,\"end\":6.2}]"}
{"utterance_id": "7850-281318-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0004.flac", "task_type": "lexical_detection", "query": "things", "prompt": "When the utterance is complete, report all instances of the word \"things\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.84, "things"]], "response_mode": "offline", "answer": "[{\"word\":\"things\",\"start\":1.4,\"end\":1.84}]"}
{"utterance_id": "3000-15664-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0028.flac", "task_type": "lexical_detection", "query": "lake", "prompt": "Once the audio has finished, identify every instance of the word \"lake\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.68, 4.03, "lake"]], "response_mode": "offline", "answer": "[{\"word\":\"lake\",\"start\":3.68,\"end\":4.03}]"}
{"utterance_id": "1919-142785-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0062.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.06, 1.34, "half"], [2.41, 2.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"half\",\"start\":1.06,\"end\":1.34},{\"word\":\"two\",\"start\":2.41,\"end\":2.65}]"}
{"utterance_id": "422-122949-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0012.flac", "task_type": "lexical_detection", "query": "higher", "prompt": "Once the audio has finished, identify every instance of the word \"higher\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.9, 3.38, "higher"]], "response_mode": "offline", "answer": "[{\"word\":\"higher\",\"start\":2.9,\"end\":3.38}]"}
{"utterance_id": "777-126732-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0047.flac", "task_type": "lexical_detection", "query": "women", "prompt": "Once the audio has finished, identify every instance of the word \"women\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.07, 4.34, "women"]], "response_mode": "offline", "answer": "[{\"word\":\"women\",\"start\":4.07,\"end\":4.34}]"}
{"utterance_id": "1673-143397-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0005.flac", "task_type": "lexical_counting", "query": "obstacle", "prompt": "Process the full audio offline. For every occurrence of the word \"obstacle\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.22, 4.71, "one"], [7.92, 8.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.22,\"end\":4.71},{\"word\":\"two\",\"start\":7.92,\"end\":8.38}]"}
{"utterance_id": "5694-64029-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0024.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.27, 4.74, "country"], [5.74, 6.27, "country"], [8.23, 8.81, "country"]], "response_mode": "offline", "answer": "[{\"word\":\"country\",\"start\":4.27,\"end\":4.74},{\"word\":\"country\",\"start\":5.74,\"end\":6.27},{\"word\":\"country\",\"start\":8.23,\"end\":8.81}]"}
{"utterance_id": "1988-148538-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0000.flac", "task_type": "lexical_detection", "query": "communities", "prompt": "Once the audio has finished, identify every instance of the word \"communities\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.0, 1.6, "communities"]], "response_mode": "offline", "answer": "[{\"word\":\"communities\",\"start\":1.0,\"end\":1.6}]"}
{"utterance_id": "5694-64029-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.89, "oakley"]], "response_mode": "offline", "answer": "[{\"word\":\"oakley\",\"start\":0.42,\"end\":0.89}]"}
{"utterance_id": "5338-24615-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0003.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.54, 10.83, "dog"]], "response_mode": "offline", "answer": "[{\"word\":\"dog\",\"start\":10.54,\"end\":10.83}]"}
{"utterance_id": "6313-66129-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0016.flac", "task_type": "lexical_detection", "query": "anxiety", "prompt": "After listening to the whole recording, output every instance of the word \"anxiety\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.15, "anxiety"]], "response_mode": "offline", "answer": "[{\"word\":\"anxiety\",\"start\":1.6,\"end\":2.15}]"}
{"utterance_id": "6345-93302-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0005.flac", "task_type": "lexical_detection", "query": "january", "prompt": "Once the audio has finished, identify every instance of the word \"january\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.91, 3.81, "january"]], "response_mode": "offline", "answer": "[{\"word\":\"january\",\"start\":2.91,\"end\":3.81}]"}
{"utterance_id": "1673-143396-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.73, 9.16, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":8.73,\"end\":9.16}]"}
{"utterance_id": "3536-8226-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0027.flac", "task_type": "lexical_detection", "query": "work", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"work\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.15, "work"]], "response_mode": "offline", "answer": "[{\"word\":\"work\",\"start\":0.87,\"end\":1.15}]"}
{"utterance_id": "3000-15664-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0004.flac", "task_type": "lexical_detection", "query": "better", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"better\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.47, "better"]], "response_mode": "offline", "answer": "[{\"word\":\"better\",\"start\":1.0,\"end\":1.47}]"}
{"utterance_id": "6313-66129-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0014.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.2, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":3.81,\"end\":4.2}]"}
{"utterance_id": "7976-105575-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0027.flac", "task_type": "lexical_detection", "query": "afterward", "prompt": "Process the full audio offline. Find every instance of the word \"afterward\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.17, 5.68, "afterward"]], "response_mode": "offline", "answer": "[{\"word\":\"afterward\",\"start\":5.17,\"end\":5.68}]"}
{"utterance_id": "1462-170142-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0001.flac", "task_type": "semantic_counting", "query": "city", "prompt": "After listening to the whole recording, output timed count words for a city name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.12, 1.73, "one"], [8.13, 9.1, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.12,\"end\":1.73},{\"word\":\"two\",\"start\":8.13,\"end\":9.1}]"}
{"utterance_id": "2277-149896-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0012.flac", "task_type": "lexical_detection", "query": "able", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"able\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.03, "able"]], "response_mode": "offline", "answer": "[{\"word\":\"able\",\"start\":0.79,\"end\":1.03}]"}
{"utterance_id": "652-129742-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0003.flac", "task_type": "semantic_counting", "query": "food", "prompt": "When the utterance is complete, report how a food name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.39, "one"], [5.28, 5.75, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.39},{\"word\":\"two\",\"start\":5.28,\"end\":5.75}]"}
{"utterance_id": "2428-83699-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0036.flac", "task_type": "lexical_detection", "query": "information", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"information\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 1.07, "information"]], "response_mode": "offline", "answer": "[{\"word\":\"information\",\"start\":0.55,\"end\":1.07}]"}
{"utterance_id": "2277-149896-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0023.flac", "task_type": "lexical_detection", "query": "sit", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sit\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.02, 3.22, "sit"]], "response_mode": "offline", "answer": "[{\"word\":\"sit\",\"start\":3.02,\"end\":3.22}]"}
{"utterance_id": "2035-152373-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0005.flac", "task_type": "lexical_counting", "query": "ancient", "prompt": "Process the full audio offline. For every occurrence of the word \"ancient\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.14, "one"], [17.22, 17.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.73,\"end\":2.14},{\"word\":\"two\",\"start\":17.22,\"end\":17.57}]"}
{"utterance_id": "1462-170142-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0011.flac", "task_type": "lexical_detection", "query": "pulled", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"pulled\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.65, 0.94, "pulled"]], "response_mode": "offline", "answer": "[{\"word\":\"pulled\",\"start\":0.65,\"end\":0.94}]"}
{"utterance_id": "1919-142785-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0053.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.14, 1.9, "one"], [7.82, 8.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.14,\"end\":1.9},{\"word\":\"two\",\"start\":7.82,\"end\":8.34}]"}
{"utterance_id": "5895-34622-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0013.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 3.38, "administrator"]], "response_mode": "offline", "answer": "[{\"word\":\"administrator\",\"start\":2.71,\"end\":3.38}]"}
{"utterance_id": "7850-111771-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0006.flac", "task_type": "lexical_detection", "query": "lee", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"lee\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.13, 1.37, "lee"]], "response_mode": "offline", "answer": "[{\"word\":\"lee\",\"start\":1.13,\"end\":1.37}]"}
{"utterance_id": "777-126732-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0051.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.66, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.4,\"end\":1.66}]"}
{"utterance_id": "251-136532-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0018.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.81, 5.64, "archaeologist"]], "response_mode": "offline", "answer": "[{\"word\":\"archaeologist\",\"start\":4.81,\"end\":5.64}]"}
{"utterance_id": "8297-275155-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0018.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.18, "captain"]], "response_mode": "offline", "answer": "[{\"word\":\"captain\",\"start\":0.7,\"end\":1.18}]"}
{"utterance_id": "1919-142785-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0021.flac", "task_type": "lexical_detection", "query": "pickle", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"pickle\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.66, 1.06, "pickle"]], "response_mode": "offline", "answer": "[{\"word\":\"pickle\",\"start\":0.66,\"end\":1.06}]"}
{"utterance_id": "1462-170142-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0025.flac", "task_type": "lexical_detection", "query": "hilda's", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hilda's\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.94, "hilda's"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda's\",\"start\":0.52,\"end\":0.94}]"}
{"utterance_id": "3081-166546-0075", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0075.flac", "task_type": "lexical_detection", "query": "times", "prompt": "After listening to the whole recording, output every instance of the word \"times\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.94, "times"]], "response_mode": "offline", "answer": "[{\"word\":\"times\",\"start\":1.26,\"end\":1.94}]"}
{"utterance_id": "1919-142785-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0053.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.9, "ingredients"], [7.82, 8.34, "quenelles"]], "response_mode": "offline", "answer": "[{\"word\":\"ingredients\",\"start\":1.14,\"end\":1.9},{\"word\":\"quenelles\",\"start\":7.82,\"end\":8.34}]"}
{"utterance_id": "84-121550-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0035.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.43, "one"], [6.31, 7.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.98,\"end\":3.43},{\"word\":\"two\",\"start\":6.31,\"end\":7.01}]"}
{"utterance_id": "1272-135031-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.81, 2.48, "kaliko"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":1.81,\"end\":2.48}]"}
{"utterance_id": "652-130737-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0000.flac", "task_type": "lexical_detection", "query": "brandy", "prompt": "Process the full audio offline. Find every instance of the word \"brandy\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.33, 4.06, "brandy"]], "response_mode": "offline", "answer": "[{\"word\":\"brandy\",\"start\":3.33,\"end\":4.06}]"}
{"utterance_id": "3853-163249-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0036.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Process the full audio offline. Find every instance of a time of day and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[20.32, 20.6, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":20.32,\"end\":20.6}]"}
{"utterance_id": "8297-275156-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0013.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[7.06, 7.65, "captain's"]], "response_mode": "offline", "answer": "[{\"word\":\"captain's\",\"start\":7.06,\"end\":7.65}]"}
{"utterance_id": "8842-302201-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0009.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an ordinal number with word/start/end. If none, reply [].", "labels": [[0.51, 0.81, "one"], [2.65, 2.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.81},{\"word\":\"two\",\"start\":2.65,\"end\":2.92}]"}
{"utterance_id": "8842-302203-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0001.flac", "task_type": "lexical_detection", "query": "excellent", "prompt": "After the audio ends, extract all occurrences of the word \"excellent\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.17, 6.62, "excellent"]], "response_mode": "offline", "answer": "[{\"word\":\"excellent\",\"start\":6.17,\"end\":6.62}]"}
{"utterance_id": "5536-43363-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0019.flac", "task_type": "lexical_detection", "query": "prophet", "prompt": "When the utterance is complete, report all instances of the word \"prophet\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.14, "prophet"]], "response_mode": "offline", "answer": "[{\"word\":\"prophet\",\"start\":1.63,\"end\":2.14}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.59, 6.9, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"black\",\"start\":6.59,\"end\":6.9}]"}
{"utterance_id": "6345-64257-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0010.flac", "task_type": "lexical_detection", "query": "utterly", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"utterly\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.68, 2.28, "utterly"]], "response_mode": "offline", "answer": "[{\"word\":\"utterly\",\"start\":1.68,\"end\":2.28}]"}
{"utterance_id": "3752-4944-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0004.flac", "task_type": "lexical_detection", "query": "gravely", "prompt": "After listening to the whole recording, output every instance of the word \"gravely\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.98, "gravely"]], "response_mode": "offline", "answer": "[{\"word\":\"gravely\",\"start\":1.46,\"end\":1.98}]"}
{"utterance_id": "3536-23268-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0028.flac", "task_type": "lexical_detection", "query": "reply", "prompt": "Process the full audio offline. Find every instance of the word \"reply\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.43, "reply"]], "response_mode": "offline", "answer": "[{\"word\":\"reply\",\"start\":0.87,\"end\":1.43}]"}
{"utterance_id": "2035-147961-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0021.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.32, 6.69, "horses"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":6.32,\"end\":6.69}]"}
{"utterance_id": "2428-83705-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 2.08, "mary ann"], [4.13, 4.97, "mister whiting"]], "response_mode": "offline", "answer": "[{\"word\":\"mary ann\",\"start\":1.49,\"end\":2.08},{\"word\":\"mister whiting\",\"start\":4.13,\"end\":4.97}]"}
{"utterance_id": "3081-166546-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0031.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.36, 6.09, "detective's"]], "response_mode": "offline", "answer": "[{\"word\":\"detective's\",\"start\":5.36,\"end\":6.09}]"}
{"utterance_id": "2803-154320-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0009.flac", "task_type": "lexical_detection", "query": "vain", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"vain\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.04, 2.58, "vain"]], "response_mode": "offline", "answer": "[{\"word\":\"vain\",\"start\":2.04,\"end\":2.58}]"}
{"utterance_id": "3081-166546-0074", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0074.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.18, 8.4, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":8.18,\"end\":8.4}]"}
{"utterance_id": "6345-64257-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0020.flac", "task_type": "lexical_detection", "query": "intense", "prompt": "After listening to the whole recording, output every instance of the word \"intense\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.22, "intense"]], "response_mode": "offline", "answer": "[{\"word\":\"intense\",\"start\":1.63,\"end\":2.22}]"}
{"utterance_id": "3853-163249-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0054.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Once the audio has finished, list each occurrence of a body part as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.82, 2.26, "one"], [11.36, 11.81, "two"], [16.53, 16.84, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.82,\"end\":2.26},{\"word\":\"two\",\"start\":11.36,\"end\":11.81},{\"word\":\"three\",\"start\":16.53,\"end\":16.84}]"}
{"utterance_id": "3170-137482-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0020.flac", "task_type": "lexical_detection", "query": "insisted", "prompt": "After the audio ends, extract all occurrences of the word \"insisted\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.09, "insisted"]], "response_mode": "offline", "answer": "[{\"word\":\"insisted\",\"start\":1.55,\"end\":2.09}]"}
{"utterance_id": "1462-170145-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0001.flac", "task_type": "lexical_detection", "query": "fireplace", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"fireplace\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.71, "fireplace"]], "response_mode": "offline", "answer": "[{\"word\":\"fireplace\",\"start\":0.86,\"end\":1.71}]"}
{"utterance_id": "2803-154328-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.36, 8.84, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":8.36,\"end\":8.84}]"}
{"utterance_id": "7850-73752-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0007.flac", "task_type": "lexical_detection", "query": "bitter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"bitter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.37, 7.76, "bitter"]], "response_mode": "offline", "answer": "[{\"word\":\"bitter\",\"start\":7.37,\"end\":7.76}]"}
{"utterance_id": "5338-24615-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0001.flac", "task_type": "lexical_detection", "query": "extreme", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"extreme\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.39, 3.06, "extreme"]], "response_mode": "offline", "answer": "[{\"word\":\"extreme\",\"start\":2.39,\"end\":3.06}]"}
{"utterance_id": "174-50561-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0008.flac", "task_type": "lexical_detection", "query": "pledged", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"pledged\" with fields word, start, end (seconds). If none, reply [].", "labels": [[8.58, 8.99, "pledged"]], "response_mode": "offline", "answer": "[{\"word\":\"pledged\",\"start\":8.58,\"end\":8.99}]"}
{"utterance_id": "6345-93302-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0005.flac", "task_type": "semantic_counting", "query": "date", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a date with word/start/end. If none, reply [].", "labels": [[1.15, 1.65, "one"], [2.91, 3.81, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.15,\"end\":1.65},{\"word\":\"two\",\"start\":2.91,\"end\":3.81}]"}
{"utterance_id": "2035-147960-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.86, "antonia"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia\",\"start\":1.26,\"end\":1.86}]"}
{"utterance_id": "7850-281318-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0022.flac", "task_type": "lexical_detection", "query": "worst", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"worst\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.47, "worst"]], "response_mode": "offline", "answer": "[{\"word\":\"worst\",\"start\":1.96,\"end\":2.47}]"}
{"utterance_id": "3752-4944-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.82, 3.43, "burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"burgess\",\"start\":2.82,\"end\":3.43}]"}
{"utterance_id": "2803-154328-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0017.flac", "task_type": "lexical_detection", "query": "undertaking", "prompt": "Process the full audio offline. Find every instance of the word \"undertaking\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.24, 6.92, "undertaking"]], "response_mode": "offline", "answer": "[{\"word\":\"undertaking\",\"start\":6.24,\"end\":6.92}]"}
{"utterance_id": "1988-147956-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 1.04, "antonia"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia\",\"start\":0.49,\"end\":1.04}]"}
{"utterance_id": "2428-83699-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0029.flac", "task_type": "lexical_detection", "query": "twisted", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"twisted\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.0, 3.49, "twisted"]], "response_mode": "offline", "answer": "[{\"word\":\"twisted\",\"start\":3.0,\"end\":3.49}]"}
{"utterance_id": "6319-57405-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0000.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "When the utterance is complete, report how a person's name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.66, 4.37, "one"], [5.99, 6.87, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.66,\"end\":4.37},{\"word\":\"two\",\"start\":5.99,\"end\":6.87}]"}
{"utterance_id": "652-130726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0018.flac", "task_type": "lexical_detection", "query": "reasonable", "prompt": "Process the full audio offline. Find every instance of the word \"reasonable\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.5, "reasonable"]], "response_mode": "offline", "answer": "[{\"word\":\"reasonable\",\"start\":2.07,\"end\":2.5}]"}
{"utterance_id": "3576-138058-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0017.flac", "task_type": "lexical_detection", "query": "impulses", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"impulses\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[16.22, 16.89, "impulses"]], "response_mode": "offline", "answer": "[{\"word\":\"impulses\",\"start\":16.22,\"end\":16.89}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.45, 4.83, "apple"], [6.32, 6.78, "fruit"], [14.72, 15.23, "honey"]], "response_mode": "offline", "answer": "[{\"word\":\"apple\",\"start\":4.45,\"end\":4.83},{\"word\":\"fruit\",\"start\":6.32,\"end\":6.78},{\"word\":\"honey\",\"start\":14.72,\"end\":15.23}]"}
{"utterance_id": "251-137823-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0020.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 0.75, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.53,\"end\":0.75}]"}
{"utterance_id": "3536-8226-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0018.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.27, 7.83, "minister's"]], "response_mode": "offline", "answer": "[{\"word\":\"minister's\",\"start\":7.27,\"end\":7.83}]"}
{"utterance_id": "652-129742-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0000.flac", "task_type": "semantic_counting", "query": "food", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a food name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.51, 1.62, "one"], [2.89, 3.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.62},{\"word\":\"two\",\"start\":2.89,\"end\":3.59}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.27, 2.9, "beasts"]], "response_mode": "offline", "answer": "[{\"word\":\"beasts\",\"start\":2.27,\"end\":2.9}]"}
{"utterance_id": "7850-286674-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0007.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.88, 3.32, "nymph"]], "response_mode": "offline", "answer": "[{\"word\":\"nymph\",\"start\":2.88,\"end\":3.32}]"}
{"utterance_id": "2035-147961-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0028.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 1.61, "twenty thirty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty thirty\",\"start\":0.43,\"end\":1.61}]"}
{"utterance_id": "1919-142785-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0037.flac", "task_type": "lexical_detection", "query": "pies", "prompt": "Once the audio has finished, identify every instance of the word \"pies\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.18, 2.75, "pies"]], "response_mode": "offline", "answer": "[{\"word\":\"pies\",\"start\":2.18,\"end\":2.75}]"}
{"utterance_id": "1272-128104-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0003.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.86, 9.65, "ithaca"]], "response_mode": "offline", "answer": "[{\"word\":\"ithaca\",\"start\":8.86,\"end\":9.65}]"}
{"utterance_id": "7850-73752-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0018.flac", "task_type": "semantic_detection", "query": "country", "prompt": "You will hear a complete utterance. When it ends, list every instance of a country name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[22.52, 23.22, "lebanon"]], "response_mode": "offline", "answer": "[{\"word\":\"lebanon\",\"start\":22.52,\"end\":23.22}]"}
{"utterance_id": "1272-141231-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.68, 1.43, "twenties"]], "response_mode": "offline", "answer": "[{\"word\":\"twenties\",\"start\":0.68,\"end\":1.43}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a body part (word, start, end). If none, reply [].", "labels": [[4.45, 4.95, "one"], [6.23, 6.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.45,\"end\":4.95},{\"word\":\"two\",\"start\":6.23,\"end\":6.65}]"}
{"utterance_id": "652-130726-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0024.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "After the audio ends, tally every instance of a number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.54, 1.68, "one"], [3.77, 3.96, "two"], [5.39, 5.75, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.54,\"end\":1.68},{\"word\":\"two\",\"start\":3.77,\"end\":3.96},{\"word\":\"three\",\"start\":5.39,\"end\":5.75}]"}
{"utterance_id": "6313-66129-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0034.flac", "task_type": "lexical_detection", "query": "grasp", "prompt": "Once the audio has finished, identify every instance of the word \"grasp\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.77, 7.16, "grasp"]], "response_mode": "offline", "answer": "[{\"word\":\"grasp\",\"start\":6.77,\"end\":7.16}]"}
{"utterance_id": "3081-166546-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.35, 2.08, "detective"]], "response_mode": "offline", "answer": "[{\"word\":\"detective\",\"start\":1.35,\"end\":2.08}]"}
{"utterance_id": "2803-154328-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0020.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.33, 9.88, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":9.33,\"end\":9.88}]"}
{"utterance_id": "7976-105575-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.52, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":3.11,\"end\":3.52}]"}
{"utterance_id": "3081-166546-0083", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0083.flac", "task_type": "lexical_detection", "query": "further", "prompt": "Once the audio has finished, identify every instance of the word \"further\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.85, 2.37, "further"]], "response_mode": "offline", "answer": "[{\"word\":\"further\",\"start\":1.85,\"end\":2.37}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 0.91, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.58,\"end\":0.91}]"}
{"utterance_id": "2078-142845-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0033.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.69, 6.15, "butter"]], "response_mode": "offline", "answer": "[{\"word\":\"butter\",\"start\":5.69,\"end\":6.15}]"}
{"utterance_id": "5694-64025-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.15, 0.7, "officers"]], "response_mode": "offline", "answer": "[{\"word\":\"officers\",\"start\":0.15,\"end\":0.7}]"}
{"utterance_id": "5694-64025-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0006.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "After listening to the whole recording, output every instance of a state or province name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 2.11, "alabama"]], "response_mode": "offline", "answer": "[{\"word\":\"alabama\",\"start\":1.51,\"end\":2.11}]"}
{"utterance_id": "1462-170138-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0015.flac", "task_type": "lexical_detection", "query": "dare", "prompt": "Process the full audio offline. Find every instance of the word \"dare\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.51, "dare"]], "response_mode": "offline", "answer": "[{\"word\":\"dare\",\"start\":0.3,\"end\":0.51}]"}
{"utterance_id": "2803-154328-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0013.flac", "task_type": "lexical_detection", "query": "stoop", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"stoop\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.0, 3.74, "stoop"]], "response_mode": "offline", "answer": "[{\"word\":\"stoop\",\"start\":3.0,\"end\":3.74}]"}
{"utterance_id": "3081-166546-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.53, 1.86, "mother"], [3.42, 3.92, "aunt"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":1.53,\"end\":1.86},{\"word\":\"aunt\",\"start\":3.42,\"end\":3.92}]"}
{"utterance_id": "2078-142845-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0014.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.71, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":1.47,\"end\":1.71}]"}
{"utterance_id": "2412-153947-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0009.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Listen to the full audio. After it finishes, find every instance of a disease name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.49, 8.23, "consumption"]], "response_mode": "offline", "answer": "[{\"word\":\"consumption\",\"start\":7.49,\"end\":8.23}]"}
{"utterance_id": "6313-66129-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0027.flac", "task_type": "lexical_detection", "query": "save", "prompt": "Once the audio has finished, identify every instance of the word \"save\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.81, 4.11, "save"]], "response_mode": "offline", "answer": "[{\"word\":\"save\",\"start\":3.81,\"end\":4.11}]"}
{"utterance_id": "652-130726-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0027.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.36, 4.78, "brine"]], "response_mode": "offline", "answer": "[{\"word\":\"brine\",\"start\":4.36,\"end\":4.78}]"}
{"utterance_id": "1988-24833-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.95, 1.2, "hilda"], [2.47, 2.72, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":0.95,\"end\":1.2},{\"word\":\"tom\",\"start\":2.47,\"end\":2.72}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "lexical_detection", "query": "seeds", "prompt": "Once the audio has finished, identify every instance of the word \"seeds\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.13, 5.64, "seeds"]], "response_mode": "offline", "answer": "[{\"word\":\"seeds\",\"start\":5.13,\"end\":5.64}]"}
{"utterance_id": "422-122949-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0015.flac", "task_type": "lexical_counting", "query": "danger", "prompt": "Once the audio has finished, list each occurrence of the word \"danger\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.52, 1.2, "one"], [4.09, 4.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":1.2},{\"word\":\"two\",\"start\":4.09,\"end\":4.68}]"}
{"utterance_id": "7976-110523-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.91, "grethel"]], "response_mode": "offline", "answer": "[{\"word\":\"grethel\",\"start\":0.53,\"end\":0.91}]"}
{"utterance_id": "3752-4944-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0066.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a number: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.26, 0.69, "one"], [3.35, 3.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.26,\"end\":0.69},{\"word\":\"two\",\"start\":3.35,\"end\":3.86}]"}
{"utterance_id": "84-121550-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.75, 6.22, "griffin's"]], "response_mode": "offline", "answer": "[{\"word\":\"griffin's\",\"start\":5.75,\"end\":6.22}]"}
{"utterance_id": "6345-93302-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0009.flac", "task_type": "lexical_detection", "query": "shops", "prompt": "After the audio ends, extract all occurrences of the word \"shops\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.07, "shops"]], "response_mode": "offline", "answer": "[{\"word\":\"shops\",\"start\":3.58,\"end\":4.07}]"}
{"utterance_id": "1988-147956-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0012.flac", "task_type": "lexical_detection", "query": "distance", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"distance\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.87, 1.34, "distance"]], "response_mode": "offline", "answer": "[{\"word\":\"distance\",\"start\":0.87,\"end\":1.34}]"}
{"utterance_id": "2428-83699-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0020.flac", "task_type": "lexical_detection", "query": "uncommonly", "prompt": "Once the audio has finished, identify every instance of the word \"uncommonly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.51, 3.16, "uncommonly"]], "response_mode": "offline", "answer": "[{\"word\":\"uncommonly\",\"start\":2.51,\"end\":3.16}]"}
{"utterance_id": "2035-152373-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0006.flac", "task_type": "lexical_detection", "query": "sadly", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sadly\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.96, 5.48, "sadly"]], "response_mode": "offline", "answer": "[{\"word\":\"sadly\",\"start\":4.96,\"end\":5.48}]"}
{"utterance_id": "3536-8226-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0026.flac", "task_type": "lexical_counting", "query": "suspected", "prompt": "Listen to the entire clip, then report the count sequence for the word \"suspected\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.2, 6.8, "one"], [8.7, 9.39, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.2,\"end\":6.8},{\"word\":\"two\",\"start\":8.7,\"end\":9.39}]"}
{"utterance_id": "2428-83699-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.97, 1.62, "madge"]], "response_mode": "offline", "answer": "[{\"word\":\"madge\",\"start\":0.97,\"end\":1.62}]"}
{"utterance_id": "777-126732-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0058.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.35, "agent"]], "response_mode": "offline", "answer": "[{\"word\":\"agent\",\"start\":4.04,\"end\":4.35}]"}
{"utterance_id": "1673-143396-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0019.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.21, 11.6, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":11.21,\"end\":11.6}]"}
{"utterance_id": "7976-105575-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0026.flac", "task_type": "lexical_detection", "query": "protested", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"protested\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.12, 1.93, "protested"]], "response_mode": "offline", "answer": "[{\"word\":\"protested\",\"start\":1.12,\"end\":1.93}]"}
{"utterance_id": "422-122949-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0019.flac", "task_type": "lexical_detection", "query": "sought", "prompt": "After the audio ends, extract all occurrences of the word \"sought\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.52, 14.8, "sought"]], "response_mode": "offline", "answer": "[{\"word\":\"sought\",\"start\":14.52,\"end\":14.8}]"}
{"utterance_id": "3170-137482-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0002.flac", "task_type": "lexical_detection", "query": "succeeded", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"succeeded\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.53, 7.17, "succeeded"]], "response_mode": "offline", "answer": "[{\"word\":\"succeeded\",\"start\":6.53,\"end\":7.17}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Process the full audio offline. Find every instance of a state or province name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.16, 8.76, "new jersey"]], "response_mode": "offline", "answer": "[{\"word\":\"new jersey\",\"start\":8.16,\"end\":8.76}]"}
{"utterance_id": "2086-149220-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0035.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 3.27, "holgrave"], [5.02, 5.45, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"holgrave\",\"start\":2.66,\"end\":3.27},{\"word\":\"phoebe\",\"start\":5.02,\"end\":5.45}]"}
{"utterance_id": "6319-64726-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0019.flac", "task_type": "lexical_detection", "query": "grand", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"grand\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.17, 8.65, "grand"]], "response_mode": "offline", "answer": "[{\"word\":\"grand\",\"start\":8.17,\"end\":8.65}]"}
{"utterance_id": "251-137823-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0024.flac", "task_type": "lexical_detection", "query": "tom", "prompt": "Process the full audio offline. Find every instance of the word \"tom\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.99, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.5,\"end\":0.99}]"}
{"utterance_id": "7976-110124-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0017.flac", "task_type": "lexical_detection", "query": "yet", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"yet\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.89, "yet"]], "response_mode": "offline", "answer": "[{\"word\":\"yet\",\"start\":1.62,\"end\":1.89}]"}
{"utterance_id": "6319-64726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0004.flac", "task_type": "lexical_detection", "query": "castles", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"castles\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.57, "castles"]], "response_mode": "offline", "answer": "[{\"word\":\"castles\",\"start\":5.9,\"end\":6.57}]"}
{"utterance_id": "1919-142785-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0018.flac", "task_type": "lexical_detection", "query": "arabians", "prompt": "After the audio ends, extract all occurrences of the word \"arabians\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.86, 6.58, "arabians"]], "response_mode": "offline", "answer": "[{\"word\":\"arabians\",\"start\":5.86,\"end\":6.58}]"}
{"utterance_id": "1673-143396-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0015.flac", "task_type": "lexical_detection", "query": "silenced", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"silenced\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.44, "silenced"]], "response_mode": "offline", "answer": "[{\"word\":\"silenced\",\"start\":1.88,\"end\":2.44}]"}
{"utterance_id": "3170-137482-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0017.flac", "task_type": "lexical_detection", "query": "swords", "prompt": "After the audio ends, extract all occurrences of the word \"swords\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.96, 11.62, "swords"]], "response_mode": "offline", "answer": "[{\"word\":\"swords\",\"start\":10.96,\"end\":11.62}]"}
{"utterance_id": "3081-166546-0087", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0087.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Do not answer until the audio ends. Then list each instance of a disease name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.22, 2.75, "wound"]], "response_mode": "offline", "answer": "[{\"word\":\"wound\",\"start\":2.22,\"end\":2.75}]"}
{"utterance_id": "5694-64029-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.65, 1.29, "federal army"]], "response_mode": "offline", "answer": "[{\"word\":\"federal army\",\"start\":0.65,\"end\":1.29}]"}
{"utterance_id": "5895-34622-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0021.flac", "task_type": "lexical_detection", "query": "like", "prompt": "Once the audio has finished, identify every instance of the word \"like\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.47, 1.68, "like"]], "response_mode": "offline", "answer": "[{\"word\":\"like\",\"start\":1.47,\"end\":1.68}]"}
{"utterance_id": "2428-83699-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0019.flac", "task_type": "lexical_detection", "query": "trap", "prompt": "When the utterance is complete, report all instances of the word \"trap\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.43, "trap"]], "response_mode": "offline", "answer": "[{\"word\":\"trap\",\"start\":0.95,\"end\":1.43}]"}
{"utterance_id": "8297-275154-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0012.flac", "task_type": "lexical_detection", "query": "sydney", "prompt": "Process the full audio offline. Find every instance of the word \"sydney\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.62, 4.09, "sydney"]], "response_mode": "offline", "answer": "[{\"word\":\"sydney\",\"start\":3.62,\"end\":4.09}]"}
{"utterance_id": "2428-83705-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0004.flac", "task_type": "lexical_detection", "query": "play", "prompt": "Process the full audio offline. Find every instance of the word \"play\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.38, 6.64, "play"]], "response_mode": "offline", "answer": "[{\"word\":\"play\",\"start\":6.38,\"end\":6.64}]"}
{"utterance_id": "5895-34615-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.03, 4.33, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":4.03,\"end\":4.33}]"}
{"utterance_id": "5536-43359-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0014.flac", "task_type": "lexical_detection", "query": "hospitality", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"hospitality\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.68, 1.41, "hospitality"]], "response_mode": "offline", "answer": "[{\"word\":\"hospitality\",\"start\":0.68,\"end\":1.41}]"}
{"utterance_id": "1919-142785-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0054.flac", "task_type": "lexical_detection", "query": "firm", "prompt": "Once the audio has finished, identify every instance of the word \"firm\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.45, 1.75, "firm"]], "response_mode": "offline", "answer": "[{\"word\":\"firm\",\"start\":1.45,\"end\":1.75}]"}
{"utterance_id": "2078-142845-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0013.flac", "task_type": "lexical_detection", "query": "variety", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"variety\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.75, 1.48, "variety"]], "response_mode": "offline", "answer": "[{\"word\":\"variety\",\"start\":0.75,\"end\":1.48}]"}
{"utterance_id": "251-118436-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 3.1, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"mouth\",\"start\":2.66,\"end\":3.1}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.88, 3.22, "hunter"]], "response_mode": "offline", "answer": "[{\"word\":\"hunter\",\"start\":2.88,\"end\":3.22}]"}
{"utterance_id": "2428-83699-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0024.flac", "task_type": "lexical_detection", "query": "wet", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"wet\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.78, 2.2, "wet"]], "response_mode": "offline", "answer": "[{\"word\":\"wet\",\"start\":1.78,\"end\":2.2}]"}
{"utterance_id": "3853-163249-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0049.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.84, 12.15, "gray"]], "response_mode": "offline", "answer": "[{\"word\":\"gray\",\"start\":11.84,\"end\":12.15}]"}
{"utterance_id": "8297-275155-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0016.flac", "task_type": "lexical_counting", "query": "missus", "prompt": "Listen to the entire clip, then report the count sequence for the word \"missus\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.13, "one"], [3.58, 4.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.72,\"end\":2.13},{\"word\":\"two\",\"start\":3.58,\"end\":4.0}]"}
{"utterance_id": "3853-163249-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.19, 2.0, "missus wilkins"]], "response_mode": "offline", "answer": "[{\"word\":\"missus wilkins\",\"start\":1.19,\"end\":2.0}]"}
{"utterance_id": "8842-302196-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.63, 4.83, "ear"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":4.63,\"end\":4.83}]"}
{"utterance_id": "84-121550-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0035.flac", "task_type": "lexical_detection", "query": "forehead", "prompt": "When the utterance is complete, report all instances of the word \"forehead\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.43, "forehead"]], "response_mode": "offline", "answer": "[{\"word\":\"forehead\",\"start\":2.98,\"end\":3.43}]"}
{"utterance_id": "777-126732-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0040.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.44, "flesh"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":2.05,\"end\":2.44}]"}
{"utterance_id": "251-118436-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0010.flac", "task_type": "lexical_detection", "query": "point", "prompt": "After the audio ends, extract all occurrences of the word \"point\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.83, "point"]], "response_mode": "offline", "answer": "[{\"word\":\"point\",\"start\":0.53,\"end\":0.83}]"}
{"utterance_id": "5895-34622-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0019.flac", "task_type": "lexical_detection", "query": "arch", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"arch\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 1.8, "arch"]], "response_mode": "offline", "answer": "[{\"word\":\"arch\",\"start\":1.5,\"end\":1.8}]"}
{"utterance_id": "2277-149874-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0021.flac", "task_type": "lexical_detection", "query": "buildings", "prompt": "Process the full audio offline. Find every instance of the word \"buildings\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.79, "buildings"]], "response_mode": "offline", "answer": "[{\"word\":\"buildings\",\"start\":1.19,\"end\":1.79}]"}
{"utterance_id": "2277-149874-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0007.flac", "task_type": "lexical_detection", "query": "carpet", "prompt": "After the audio ends, extract all occurrences of the word \"carpet\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.54, 4.1, "carpet"]], "response_mode": "offline", "answer": "[{\"word\":\"carpet\",\"start\":3.54,\"end\":4.1}]"}
{"utterance_id": "7850-73752-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0007.flac", "task_type": "lexical_counting", "query": "how", "prompt": "When the utterance is complete, report how the word \"how\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.64, "one"], [1.83, 1.98, "two"], [3.35, 3.5, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.49,\"end\":0.64},{\"word\":\"two\",\"start\":1.83,\"end\":1.98},{\"word\":\"three\",\"start\":3.35,\"end\":3.5}]"}
{"utterance_id": "1272-135031-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0008.flac", "task_type": "lexical_detection", "query": "shaggy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"shaggy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.16, "shaggy"]], "response_mode": "offline", "answer": "[{\"word\":\"shaggy\",\"start\":2.74,\"end\":3.16}]"}
{"utterance_id": "6241-61946-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0023.flac", "task_type": "lexical_detection", "query": "boats", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"boats\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.15, 2.38, "boats"]], "response_mode": "offline", "answer": "[{\"word\":\"boats\",\"start\":2.15,\"end\":2.38}]"}
{"utterance_id": "6313-76958-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0028.flac", "task_type": "lexical_detection", "query": "except", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"except\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.6, 2.97, "except"]], "response_mode": "offline", "answer": "[{\"word\":\"except\",\"start\":2.6,\"end\":2.97}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "lexical_counting", "query": "cup", "prompt": "Listen to the entire clip, then report the count sequence for the word \"cup\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.17, 5.4, "one"], [7.78, 8.03, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.17,\"end\":5.4},{\"word\":\"two\",\"start\":7.78,\"end\":8.03}]"}
{"utterance_id": "2428-83705-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0012.flac", "task_type": "lexical_detection", "query": "persuaded", "prompt": "After the audio ends, extract all occurrences of the word \"persuaded\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.3, "persuaded"]], "response_mode": "offline", "answer": "[{\"word\":\"persuaded\",\"start\":0.66,\"end\":1.3}]"}
{"utterance_id": "6319-57405-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0009.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[6.82, 7.29, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":6.82,\"end\":7.29}]"}
{"utterance_id": "84-121550-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0009.flac", "task_type": "lexical_detection", "query": "equally", "prompt": "After the audio ends, extract all occurrences of the word \"equally\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.86, "equally"]], "response_mode": "offline", "answer": "[{\"word\":\"equally\",\"start\":3.4,\"end\":3.86}]"}
{"utterance_id": "5895-34629-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0020.flac", "task_type": "lexical_detection", "query": "talents", "prompt": "Once the audio has finished, identify every instance of the word \"talents\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.43, 2.05, "talents"]], "response_mode": "offline", "answer": "[{\"word\":\"talents\",\"start\":1.43,\"end\":2.05}]"}
{"utterance_id": "2412-153954-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0005.flac", "task_type": "lexical_detection", "query": "feature", "prompt": "Process the full audio offline. Find every instance of the word \"feature\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.04, "feature"]], "response_mode": "offline", "answer": "[{\"word\":\"feature\",\"start\":0.72,\"end\":1.04}]"}
{"utterance_id": "2277-149896-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0022.flac", "task_type": "lexical_detection", "query": "discretion", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"discretion\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.7, 2.21, "discretion"]], "response_mode": "offline", "answer": "[{\"word\":\"discretion\",\"start\":1.7,\"end\":2.21}]"}
{"utterance_id": "5895-34629-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0003.flac", "task_type": "lexical_detection", "query": "heaven", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"heaven\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.75, 1.16, "heaven"]], "response_mode": "offline", "answer": "[{\"word\":\"heaven\",\"start\":0.75,\"end\":1.16}]"}
{"utterance_id": "1462-170142-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0019.flac", "task_type": "lexical_detection", "query": "contempt", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"contempt\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.48, "contempt"]], "response_mode": "offline", "answer": "[{\"word\":\"contempt\",\"start\":3.83,\"end\":4.48}]"}
{"utterance_id": "2902-9006-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0011.flac", "task_type": "lexical_detection", "query": "story", "prompt": "Once the audio has finished, identify every instance of the word \"story\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.64, 3.17, "story"]], "response_mode": "offline", "answer": "[{\"word\":\"story\",\"start\":2.64,\"end\":3.17}]"}
{"utterance_id": "3170-137482-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0037.flac", "task_type": "lexical_detection", "query": "calculus", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"calculus\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.61, 5.44, "calculus"]], "response_mode": "offline", "answer": "[{\"word\":\"calculus\",\"start\":4.61,\"end\":5.44}]"}
{"utterance_id": "3170-137482-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0031.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[7.8, 7.98, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":7.8,\"end\":7.98}]"}
{"utterance_id": "2086-149220-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0004.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.03, 4.61, "vegetables"]], "response_mode": "offline", "answer": "[{\"word\":\"vegetables\",\"start\":4.03,\"end\":4.61}]"}
{"utterance_id": "6345-93306-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0001.flac", "task_type": "lexical_detection", "query": "writing", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"writing\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.23, 6.55, "writing"]], "response_mode": "offline", "answer": "[{\"word\":\"writing\",\"start\":6.23,\"end\":6.55}]"}
{"utterance_id": "652-130726-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0026.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.42, 5.91, "flour"]], "response_mode": "offline", "answer": "[{\"word\":\"flour\",\"start\":5.42,\"end\":5.91}]"}
{"utterance_id": "6295-244435-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0037.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.38, 5.01, "collins"]], "response_mode": "offline", "answer": "[{\"word\":\"collins\",\"start\":4.38,\"end\":5.01}]"}
{"utterance_id": "84-121550-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0026.flac", "task_type": "lexical_detection", "query": "direct", "prompt": "Once the audio has finished, identify every instance of the word \"direct\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.79, 6.23, "direct"]], "response_mode": "offline", "answer": "[{\"word\":\"direct\",\"start\":5.79,\"end\":6.23}]"}
{"utterance_id": "6319-275224-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0003.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.27, 9.22, "wire worms"]], "response_mode": "offline", "answer": "[{\"word\":\"wire worms\",\"start\":8.27,\"end\":9.22}]"}
{"utterance_id": "3752-4943-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.68, "clerk"]], "response_mode": "offline", "answer": "[{\"word\":\"clerk\",\"start\":2.26,\"end\":2.68}]"}
{"utterance_id": "3752-4943-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0011.flac", "task_type": "lexical_detection", "query": "wolves", "prompt": "When the utterance is complete, report all instances of the word \"wolves\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.34, 3.13, "wolves"]], "response_mode": "offline", "answer": "[{\"word\":\"wolves\",\"start\":2.34,\"end\":3.13}]"}
{"utterance_id": "251-118436-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0008.flac", "task_type": "lexical_detection", "query": "stained", "prompt": "When the utterance is complete, report all instances of the word \"stained\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.82, "stained"]], "response_mode": "offline", "answer": "[{\"word\":\"stained\",\"start\":1.29,\"end\":1.82}]"}
{"utterance_id": "6345-93306-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0015.flac", "task_type": "lexical_detection", "query": "wicked", "prompt": "When the utterance is complete, report all instances of the word \"wicked\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.45, "wicked"]], "response_mode": "offline", "answer": "[{\"word\":\"wicked\",\"start\":3.15,\"end\":3.45}]"}
{"utterance_id": "6345-93306-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0013.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.08, 9.41, "money"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":9.08,\"end\":9.41}]"}
{"utterance_id": "7976-105575-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.06, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":1.77,\"end\":2.06}]"}
{"utterance_id": "6241-61946-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0009.flac", "task_type": "lexical_detection", "query": "violent", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"violent\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.75, 1.13, "violent"]], "response_mode": "offline", "answer": "[{\"word\":\"violent\",\"start\":0.75,\"end\":1.13}]"}
{"utterance_id": "1988-24833-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0012.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.26, "bone"]], "response_mode": "offline", "answer": "[{\"word\":\"bone\",\"start\":1.01,\"end\":1.26}]"}
{"utterance_id": "652-129742-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0015.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.39, 10.62, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":10.39,\"end\":10.62}]"}
{"utterance_id": "652-130726-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0034.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.56, 9.98, "chowder"]], "response_mode": "offline", "answer": "[{\"word\":\"chowder\",\"start\":9.56,\"end\":9.98}]"}
{"utterance_id": "5338-284437-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0008.flac", "task_type": "lexical_detection", "query": "pink", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pink\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.75, 8.07, "pink"]], "response_mode": "offline", "answer": "[{\"word\":\"pink\",\"start\":7.75,\"end\":8.07}]"}
{"utterance_id": "1993-147149-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0027.flac", "task_type": "lexical_detection", "query": "practical", "prompt": "Process the full audio offline. Find every instance of the word \"practical\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.1, 2.68, "practical"]], "response_mode": "offline", "answer": "[{\"word\":\"practical\",\"start\":2.1,\"end\":2.68}]"}
{"utterance_id": "2902-9008-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0014.flac", "task_type": "lexical_detection", "query": "bound", "prompt": "Once the audio has finished, identify every instance of the word \"bound\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.36, 7.7, "bound"]], "response_mode": "offline", "answer": "[{\"word\":\"bound\",\"start\":7.36,\"end\":7.7}]"}
{"utterance_id": "3081-166546-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0079.flac", "task_type": "lexical_detection", "query": "importance", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"importance\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.25, 6.94, "importance"]], "response_mode": "offline", "answer": "[{\"word\":\"importance\",\"start\":6.25,\"end\":6.94}]"}
{"utterance_id": "652-130726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0000.flac", "task_type": "lexical_detection", "query": "nineteen", "prompt": "Once the audio has finished, identify every instance of the word \"nineteen\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.42, 5.81, "nineteen"]], "response_mode": "offline", "answer": "[{\"word\":\"nineteen\",\"start\":5.42,\"end\":5.81}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Once the audio has finished, identify every instance of a plant name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.69, 6.27, "trees"]], "response_mode": "offline", "answer": "[{\"word\":\"trees\",\"start\":5.69,\"end\":6.27}]"}
{"utterance_id": "777-126732-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0050.flac", "task_type": "lexical_detection", "query": "downstairs", "prompt": "Process the full audio offline. Find every instance of the word \"downstairs\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 6.33, "downstairs"]], "response_mode": "offline", "answer": "[{\"word\":\"downstairs\",\"start\":5.52,\"end\":6.33}]"}
{"utterance_id": "1462-170142-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0017.flac", "task_type": "lexical_detection", "query": "misery", "prompt": "After the audio ends, extract all occurrences of the word \"misery\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.09, 1.54, "misery"]], "response_mode": "offline", "answer": "[{\"word\":\"misery\",\"start\":1.09,\"end\":1.54}]"}
{"utterance_id": "7850-286674-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0017.flac", "task_type": "lexical_detection", "query": "gotten", "prompt": "When the utterance is complete, report all instances of the word \"gotten\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 1.77, "gotten"]], "response_mode": "offline", "answer": "[{\"word\":\"gotten\",\"start\":1.49,\"end\":1.77}]"}
{"utterance_id": "1673-143397-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.96, "orestes"], [4.69, 5.37, "theodosius"]], "response_mode": "offline", "answer": "[{\"word\":\"orestes\",\"start\":0.45,\"end\":0.96},{\"word\":\"theodosius\",\"start\":4.69,\"end\":5.37}]"}
{"utterance_id": "84-121550-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0035.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.98, 3.43, "forehead"], [6.31, 7.01, "footsteps"]], "response_mode": "offline", "answer": "[{\"word\":\"forehead\",\"start\":2.98,\"end\":3.43},{\"word\":\"footsteps\",\"start\":6.31,\"end\":7.01}]"}
{"utterance_id": "251-118436-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.26, 2.7, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":2.26,\"end\":2.7}]"}
{"utterance_id": "1993-147149-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.64, "jem's"]], "response_mode": "offline", "answer": "[{\"word\":\"jem's\",\"start\":1.36,\"end\":1.64}]"}
{"utterance_id": "5694-64029-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0012.flac", "task_type": "lexical_detection", "query": "falling", "prompt": "After the audio ends, extract all occurrences of the word \"falling\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.19, "falling"]], "response_mode": "offline", "answer": "[{\"word\":\"falling\",\"start\":0.79,\"end\":1.19}]"}
{"utterance_id": "3536-23268-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0011.flac", "task_type": "lexical_detection", "query": "beseeching", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"beseeching\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.39, 6.88, "beseeching"]], "response_mode": "offline", "answer": "[{\"word\":\"beseeching\",\"start\":6.39,\"end\":6.88}]"}
{"utterance_id": "2412-153954-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0015.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.33, 8.65, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":8.33,\"end\":8.65}]"}
{"utterance_id": "1272-141231-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0007.flac", "task_type": "lexical_detection", "query": "lungs", "prompt": "After listening to the whole recording, output every instance of the word \"lungs\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 1.94, "lungs"]], "response_mode": "offline", "answer": "[{\"word\":\"lungs\",\"start\":1.49,\"end\":1.94}]"}
{"utterance_id": "2078-142845-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0042.flac", "task_type": "lexical_detection", "query": "two", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"two\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.3, 3.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.3,\"end\":3.83}]"}
{"utterance_id": "5895-34629-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0027.flac", "task_type": "lexical_detection", "query": "wept", "prompt": "After the audio ends, extract all occurrences of the word \"wept\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.31, 4.79, "wept"]], "response_mode": "offline", "answer": "[{\"word\":\"wept\",\"start\":4.31,\"end\":4.79}]"}
{"utterance_id": "3000-15664-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0041.flac", "task_type": "lexical_counting", "query": "river", "prompt": "Process the full audio offline. For every occurrence of the word \"river\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.49, 5.98, "one"], [8.54, 8.9, "two"], [18.3, 18.76, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.49,\"end\":5.98},{\"word\":\"two\",\"start\":8.54,\"end\":8.9},{\"word\":\"three\",\"start\":18.3,\"end\":18.76}]"}
{"utterance_id": "3752-4943-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0013.flac", "task_type": "lexical_detection", "query": "lie", "prompt": "Process the full audio offline. Find every instance of the word \"lie\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.99, 4.53, "lie"]], "response_mode": "offline", "answer": "[{\"word\":\"lie\",\"start\":3.99,\"end\":4.53}]"}
{"utterance_id": "1919-142785-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0011.flac", "task_type": "lexical_detection", "query": "brine", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"brine\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.96, 5.33, "brine"]], "response_mode": "offline", "answer": "[{\"word\":\"brine\",\"start\":4.96,\"end\":5.33}]"}
{"utterance_id": "6295-244435-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0038.flac", "task_type": "lexical_detection", "query": "thanked", "prompt": "After listening to the whole recording, output every instance of the word \"thanked\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.87, "thanked"]], "response_mode": "offline", "answer": "[{\"word\":\"thanked\",\"start\":0.51,\"end\":0.87}]"}
{"utterance_id": "2803-154320-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[0.88, 1.33, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":0.88,\"end\":1.33}]"}
{"utterance_id": "3853-163249-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0034.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.58, 1.19, "cynthy"]], "response_mode": "offline", "answer": "[{\"word\":\"cynthy\",\"start\":0.58,\"end\":1.19}]"}
{"utterance_id": "174-168635-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0018.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[13.52, 14.01, "bishop"]], "response_mode": "offline", "answer": "[{\"word\":\"bishop\",\"start\":13.52,\"end\":14.01}]"}
{"utterance_id": "8297-275154-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0016.flac", "task_type": "lexical_detection", "query": "haven't", "prompt": "Once the audio has finished, identify every instance of the word \"haven't\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.38, 0.69, "haven't"]], "response_mode": "offline", "answer": "[{\"word\":\"haven't\",\"start\":0.38,\"end\":0.69}]"}
{"utterance_id": "2086-149220-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0042.flac", "task_type": "lexical_detection", "query": "passage", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"passage\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.01, 2.64, "passage"]], "response_mode": "offline", "answer": "[{\"word\":\"passage\",\"start\":2.01,\"end\":2.64}]"}
{"utterance_id": "251-118436-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0003.flac", "task_type": "lexical_detection", "query": "sole", "prompt": "Process the full audio offline. Find every instance of the word \"sole\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.11, "sole"]], "response_mode": "offline", "answer": "[{\"word\":\"sole\",\"start\":3.79,\"end\":4.11}]"}
{"utterance_id": "3853-163249-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0029.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 1.99, "vittles"]], "response_mode": "offline", "answer": "[{\"word\":\"vittles\",\"start\":1.64,\"end\":1.99}]"}
{"utterance_id": "3081-166546-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0031.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.75, 7.15, "legs"]], "response_mode": "offline", "answer": "[{\"word\":\"legs\",\"start\":6.75,\"end\":7.15}]"}
{"utterance_id": "8297-275156-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0012.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.54, "sydney's"]], "response_mode": "offline", "answer": "[{\"word\":\"sydney's\",\"start\":1.06,\"end\":1.54}]"}
{"utterance_id": "1919-142785-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0023.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.66, "eggs"]], "response_mode": "offline", "answer": "[{\"word\":\"eggs\",\"start\":1.33,\"end\":1.66}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.64, 8.03, "ear"], [12.85, 13.38, "stem"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":7.64,\"end\":8.03},{\"word\":\"stem\",\"start\":12.85,\"end\":13.38}]"}
{"utterance_id": "6295-64301-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0001.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.92, 1.43, "afternoon"]], "response_mode": "offline", "answer": "[{\"word\":\"afternoon\",\"start\":0.92,\"end\":1.43}]"}
{"utterance_id": "422-122949-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.79, 12.19, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":11.79,\"end\":12.19}]"}
{"utterance_id": "2078-142845-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0014.flac", "task_type": "lexical_detection", "query": "attacks", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"attacks\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.8, 6.35, "attacks"]], "response_mode": "offline", "answer": "[{\"word\":\"attacks\",\"start\":5.8,\"end\":6.35}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "lexical_counting", "query": "look", "prompt": "Once the audio has finished, list each occurrence of the word \"look\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.76, 1.92, "one"], [5.94, 6.11, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.76,\"end\":1.92},{\"word\":\"two\",\"start\":5.94,\"end\":6.11}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[6.97, 7.31, "butter"], [9.57, 10.02, "rolls"], [15.12, 15.67, "butter"], [19.24, 19.62, "butter"], [21.69, 22.11, "rolls"]], "response_mode": "offline", "answer": "[{\"word\":\"butter\",\"start\":6.97,\"end\":7.31},{\"word\":\"rolls\",\"start\":9.57,\"end\":10.02},{\"word\":\"butter\",\"start\":15.12,\"end\":15.67},{\"word\":\"butter\",\"start\":19.24,\"end\":19.62},{\"word\":\"rolls\",\"start\":21.69,\"end\":22.11}]"}
{"utterance_id": "3752-4944-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0052.flac", "task_type": "lexical_detection", "query": "violence", "prompt": "Process the full audio offline. Find every instance of the word \"violence\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.72, 4.33, "violence"]], "response_mode": "offline", "answer": "[{\"word\":\"violence\",\"start\":3.72,\"end\":4.33}]"}
{"utterance_id": "8842-302201-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0008.flac", "task_type": "lexical_detection", "query": "shalt", "prompt": "After listening to the whole recording, output every instance of the word \"shalt\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.56, "shalt"]], "response_mode": "offline", "answer": "[{\"word\":\"shalt\",\"start\":1.23,\"end\":1.56}]"}
{"utterance_id": "2086-149220-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0025.flac", "task_type": "lexical_detection", "query": "brings", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"brings\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.29, 4.57, "brings"]], "response_mode": "offline", "answer": "[{\"word\":\"brings\",\"start\":4.29,\"end\":4.57}]"}
{"utterance_id": "2902-9006-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0001.flac", "task_type": "lexical_detection", "query": "disadvantage", "prompt": "After the audio ends, extract all occurrences of the word \"disadvantage\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 2.23, "disadvantage"]], "response_mode": "offline", "answer": "[{\"word\":\"disadvantage\",\"start\":1.51,\"end\":2.23}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a number with word/start/end. If none, reply [].", "labels": [[6.23, 6.82, "one"], [13.3, 13.52, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.23,\"end\":6.82},{\"word\":\"two\",\"start\":13.3,\"end\":13.52}]"}
{"utterance_id": "652-129742-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0006.flac", "task_type": "lexical_detection", "query": "cauliflower", "prompt": "When the utterance is complete, report all instances of the word \"cauliflower\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.2, "cauliflower"], [4.0, 4.73, "cauliflower"]], "response_mode": "offline", "answer": "[{\"word\":\"cauliflower\",\"start\":0.51,\"end\":1.2},{\"word\":\"cauliflower\",\"start\":4.0,\"end\":4.73}]"}
{"utterance_id": "1272-128104-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.04,\"end\":2.35}]"}
{"utterance_id": "8842-302196-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.43, "fifth"]], "response_mode": "offline", "answer": "[{\"word\":\"fifth\",\"start\":1.14,\"end\":1.43}]"}
{"utterance_id": "2086-149220-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0008.flac", "task_type": "lexical_detection", "query": "consequence", "prompt": "After the audio ends, extract all occurrences of the word \"consequence\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.53, 6.11, "consequence"]], "response_mode": "offline", "answer": "[{\"word\":\"consequence\",\"start\":5.53,\"end\":6.11}]"}
{"utterance_id": "251-136532-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.74, 7.24, "officer"]], "response_mode": "offline", "answer": "[{\"word\":\"officer\",\"start\":6.74,\"end\":7.24}]"}
{"utterance_id": "1919-142785-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0020.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After listening to the whole recording, output timed count words for a food name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.65, 1.2, "one"], [8.57, 8.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.65,\"end\":1.2},{\"word\":\"two\",\"start\":8.57,\"end\":8.9}]"}
{"utterance_id": "6319-275224-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 1.04, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":0.52,\"end\":1.04}]"}
{"utterance_id": "2803-161169-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.15, 6.52, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"black\",\"start\":6.15,\"end\":6.52}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.11, "eldest"]], "response_mode": "offline", "answer": "[{\"word\":\"eldest\",\"start\":3.7,\"end\":4.11}]"}
{"utterance_id": "1272-141231-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0008.flac", "task_type": "lexical_detection", "query": "borders", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"borders\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.7, 3.17, "borders"]], "response_mode": "offline", "answer": "[{\"word\":\"borders\",\"start\":2.7,\"end\":3.17}]"}
{"utterance_id": "84-121550-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0020.flac", "task_type": "lexical_detection", "query": "opposite", "prompt": "Process the full audio offline. Find every instance of the word \"opposite\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.99, "opposite"]], "response_mode": "offline", "answer": "[{\"word\":\"opposite\",\"start\":1.52,\"end\":1.99}]"}
{"utterance_id": "2078-142845-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0032.flac", "task_type": "lexical_detection", "query": "piled", "prompt": "After the audio ends, extract all occurrences of the word \"piled\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.43, 8.95, "piled"]], "response_mode": "offline", "answer": "[{\"word\":\"piled\",\"start\":8.43,\"end\":8.95}]"}
{"utterance_id": "3000-15664-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0012.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "After the audio ends, tally every instance of a plant name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.62, 2.01, "one"], [10.91, 11.64, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.62,\"end\":2.01},{\"word\":\"two\",\"start\":10.91,\"end\":11.64}]"}
{"utterance_id": "1919-142785-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0052.flac", "task_type": "lexical_detection", "query": "mortar", "prompt": "After the audio ends, extract all occurrences of the word \"mortar\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.28, 2.61, "mortar"]], "response_mode": "offline", "answer": "[{\"word\":\"mortar\",\"start\":2.28,\"end\":2.61}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "lexical_counting", "query": "pass", "prompt": "Once the audio has finished, list each occurrence of the word \"pass\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[5.73, 6.09, "one"], [12.01, 12.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.73,\"end\":6.09},{\"word\":\"two\",\"start\":12.01,\"end\":12.54}]"}
{"utterance_id": "251-118436-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0016.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.82, 3.4, "one"], [4.91, 5.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.82,\"end\":3.4},{\"word\":\"two\",\"start\":4.91,\"end\":5.43}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After listening to the whole recording, output timed count words for a person's name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.98, 2.35, "one"], [5.14, 5.87, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.98,\"end\":2.35},{\"word\":\"two\",\"start\":5.14,\"end\":5.87}]"}
{"utterance_id": "1919-142785-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0024.flac", "task_type": "lexical_detection", "query": "ginger", "prompt": "Process the full audio offline. Find every instance of the word \"ginger\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.39, "ginger"]], "response_mode": "offline", "answer": "[{\"word\":\"ginger\",\"start\":1.87,\"end\":2.39}]"}
{"utterance_id": "2277-149874-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0003.flac", "task_type": "lexical_detection", "query": "disposition", "prompt": "After listening to the whole recording, output every instance of the word \"disposition\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.61, "disposition"]], "response_mode": "offline", "answer": "[{\"word\":\"disposition\",\"start\":1.81,\"end\":2.61}]"}
{"utterance_id": "2086-149220-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0003.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.44, "golden"]], "response_mode": "offline", "answer": "[{\"word\":\"golden\",\"start\":2.06,\"end\":2.44}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "After the audio ends, tally every instance of a number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.48, 5.69, "one"], [7.3, 7.49, "two"], [11.21, 11.57, "three"], [15.56, 15.9, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.48,\"end\":5.69},{\"word\":\"two\",\"start\":7.3,\"end\":7.49},{\"word\":\"three\",\"start\":11.21,\"end\":11.57},{\"word\":\"four\",\"start\":15.56,\"end\":15.9}]"}
{"utterance_id": "1272-141231-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0030.flac", "task_type": "lexical_detection", "query": "recognized", "prompt": "Once the audio has finished, identify every instance of the word \"recognized\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.25, 5.95, "recognized"]], "response_mode": "offline", "answer": "[{\"word\":\"recognized\",\"start\":5.25,\"end\":5.95}]"}
{"utterance_id": "777-126732-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0041.flac", "task_type": "lexical_detection", "query": "terrifying", "prompt": "After the audio ends, extract all occurrences of the word \"terrifying\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.89, "terrifying"]], "response_mode": "offline", "answer": "[{\"word\":\"terrifying\",\"start\":1.29,\"end\":1.89}]"}
{"utterance_id": "3081-166546-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0024.flac", "task_type": "lexical_detection", "query": "trail", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"trail\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.8, "trail"]], "response_mode": "offline", "answer": "[{\"word\":\"trail\",\"start\":0.41,\"end\":0.8}]"}
{"utterance_id": "1993-147149-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0002.flac", "task_type": "lexical_detection", "query": "prey", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"prey\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.91, 5.35, "prey"]], "response_mode": "offline", "answer": "[{\"word\":\"prey\",\"start\":4.91,\"end\":5.35}]"}
{"utterance_id": "1988-148538-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0004.flac", "task_type": "lexical_detection", "query": "classes", "prompt": "Process the full audio offline. Find every instance of the word \"classes\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 2.36, "classes"]], "response_mode": "offline", "answer": "[{\"word\":\"classes\",\"start\":1.71,\"end\":2.36}]"}
{"utterance_id": "7850-281318-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0010.flac", "task_type": "lexical_detection", "query": "nests", "prompt": "Process the full audio offline. Find every instance of the word \"nests\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.57, 7.1, "nests"]], "response_mode": "offline", "answer": "[{\"word\":\"nests\",\"start\":6.57,\"end\":7.1}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "When the utterance is complete, report all instances of a weather term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.08, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":3.58,\"end\":4.08}]"}
{"utterance_id": "2428-83705-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0043.flac", "task_type": "lexical_detection", "query": "sell", "prompt": "After listening to the whole recording, output every instance of the word \"sell\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.54, "sell"]], "response_mode": "offline", "answer": "[{\"word\":\"sell\",\"start\":2.14,\"end\":2.54}]"}
{"utterance_id": "8842-302196-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0007.flac", "task_type": "lexical_detection", "query": "thanks", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"thanks\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.02, 3.43, "thanks"]], "response_mode": "offline", "answer": "[{\"word\":\"thanks\",\"start\":3.02,\"end\":3.43}]"}
{"utterance_id": "3081-166546-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0055.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the full audio. After it finishes, find every instance of a country name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.33, "germans"]], "response_mode": "offline", "answer": "[{\"word\":\"germans\",\"start\":0.78,\"end\":1.33}]"}
{"utterance_id": "2086-149220-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0003.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.36, "squashes"], [2.44, 4.26, "blossom cucumbers"], [10.4, 11.13, "string beans"], [15.01, 15.7, "tomatoes"], [18.32, 18.71, "plants"]], "response_mode": "offline", "answer": "[{\"word\":\"squashes\",\"start\":0.71,\"end\":1.36},{\"word\":\"blossom cucumbers\",\"start\":2.44,\"end\":4.26},{\"word\":\"string beans\",\"start\":10.4,\"end\":11.13},{\"word\":\"tomatoes\",\"start\":15.01,\"end\":15.7},{\"word\":\"plants\",\"start\":18.32,\"end\":18.71}]"}
{"utterance_id": "1462-170138-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.12, "mac connell"], [2.24, 3.63, "mister bartley alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"mac connell\",\"start\":0.5,\"end\":1.12},{\"word\":\"mister bartley alexander\",\"start\":2.24,\"end\":3.63}]"}
{"utterance_id": "174-50561-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0005.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.26, 2.51, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":2.26,\"end\":2.51}]"}
{"utterance_id": "1993-147966-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.15, "jim"]], "response_mode": "offline", "answer": "[{\"word\":\"jim\",\"start\":1.93,\"end\":2.15}]"}
{"utterance_id": "1919-142785-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0049.flac", "task_type": "lexical_counting", "query": "panada", "prompt": "Listen to the full audio. Count every occurrence of the word \"panada\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.05, 10.64, "one"], [14.8, 15.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.05,\"end\":10.64},{\"word\":\"two\",\"start\":14.8,\"end\":15.36}]"}
{"utterance_id": "1919-142785-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0036.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[7.15, 7.67, "one"], [11.62, 12.49, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.15,\"end\":7.67},{\"word\":\"two\",\"start\":11.62,\"end\":12.49}]"}
{"utterance_id": "6295-244435-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0002.flac", "task_type": "lexical_detection", "query": "comrades", "prompt": "Process the full audio offline. Find every instance of the word \"comrades\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.36, "comrades"]], "response_mode": "offline", "answer": "[{\"word\":\"comrades\",\"start\":1.75,\"end\":2.36}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "lexical_detection", "query": "horseback", "prompt": "Process the full audio offline. Find every instance of the word \"horseback\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.32, 7.0, "horseback"], [13.1, 13.77, "horseback"]], "response_mode": "offline", "answer": "[{\"word\":\"horseback\",\"start\":6.32,\"end\":7.0},{\"word\":\"horseback\",\"start\":13.1,\"end\":13.77}]"}
{"utterance_id": "3752-4944-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0018.flac", "task_type": "lexical_detection", "query": "sounds", "prompt": "Once the audio has finished, identify every instance of the word \"sounds\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.96, 5.59, "sounds"]], "response_mode": "offline", "answer": "[{\"word\":\"sounds\",\"start\":4.96,\"end\":5.59}]"}
{"utterance_id": "3752-4944-0060", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0060.flac", "task_type": "lexical_detection", "query": "aid", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"aid\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.23, 2.57, "aid"]], "response_mode": "offline", "answer": "[{\"word\":\"aid\",\"start\":2.23,\"end\":2.57}]"}
{"utterance_id": "7850-286674-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.93, 1.48, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":0.93,\"end\":1.48}]"}
{"utterance_id": "7850-111771-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.43, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":1.15,\"end\":1.43}]"}
{"utterance_id": "1988-147956-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0028.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.76, 2.17, "shoulder"], [4.17, 4.52, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulder\",\"start\":1.76,\"end\":2.17},{\"word\":\"face\",\"start\":4.17,\"end\":4.52}]"}
{"utterance_id": "2078-142845-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[9.8, 10.4, "loaves"], [15.25, 15.89, "loaves"]], "response_mode": "offline", "answer": "[{\"word\":\"loaves\",\"start\":9.8,\"end\":10.4},{\"word\":\"loaves\",\"start\":15.25,\"end\":15.89}]"}
{"utterance_id": "3000-15664-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0015.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.25, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":2.9,\"end\":3.25}]"}
{"utterance_id": "3752-4944-0068", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0068.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.08, 12.54, "minister"]], "response_mode": "offline", "answer": "[{\"word\":\"minister\",\"start\":12.08,\"end\":12.54}]"}
{"utterance_id": "1919-142785-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0049.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.3, 7.66, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":7.3,\"end\":7.66}]"}
{"utterance_id": "3536-23268-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0013.flac", "task_type": "lexical_detection", "query": "eagerness", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"eagerness\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.46, 3.94, "eagerness"]], "response_mode": "offline", "answer": "[{\"word\":\"eagerness\",\"start\":3.46,\"end\":3.94}]"}
{"utterance_id": "1919-142785-0060", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0060.flac", "task_type": "lexical_detection", "query": "border", "prompt": "When the utterance is complete, report all instances of the word \"border\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.49, "border"]], "response_mode": "offline", "answer": "[{\"word\":\"border\",\"start\":2.14,\"end\":2.49}]"}
{"utterance_id": "251-136532-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.35, 8.79, "dozen"]], "response_mode": "offline", "answer": "[{\"word\":\"dozen\",\"start\":8.35,\"end\":8.79}]"}
{"utterance_id": "2078-142845-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0006.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a food name with word/start/end. If none, reply [].", "labels": [[6.03, 6.43, "one"], [8.11, 8.64, "two"], [13.84, 14.35, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.03,\"end\":6.43},{\"word\":\"two\",\"start\":8.11,\"end\":8.64},{\"word\":\"three\",\"start\":13.84,\"end\":14.35}]"}
{"utterance_id": "6319-275224-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0002.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After listening to the whole recording, output every instance of a plant name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.99, 2.87, "convolvulus"]], "response_mode": "offline", "answer": "[{\"word\":\"convolvulus\",\"start\":1.99,\"end\":2.87}]"}
{"utterance_id": "1919-142785-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0026.flac", "task_type": "lexical_detection", "query": "august", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"august\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.1, 2.55, "august"]], "response_mode": "offline", "answer": "[{\"word\":\"august\",\"start\":2.1,\"end\":2.55}]"}
{"utterance_id": "1462-170142-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.27, 2.64, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":2.27,\"end\":2.64}]"}
{"utterance_id": "5694-64029-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0021.flac", "task_type": "lexical_detection", "query": "fourth", "prompt": "After listening to the whole recording, output every instance of the word \"fourth\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.06, "fourth"]], "response_mode": "offline", "answer": "[{\"word\":\"fourth\",\"start\":1.74,\"end\":2.06}]"}
{"utterance_id": "8842-304647-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0012.flac", "task_type": "lexical_detection", "query": "due", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"due\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.49, 0.79, "due"]], "response_mode": "offline", "answer": "[{\"word\":\"due\",\"start\":0.49,\"end\":0.79}]"}
{"utterance_id": "2078-142845-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0013.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.75, "yellow"], [1.98, 2.46, "golden"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":0.46,\"end\":0.75},{\"word\":\"golden\",\"start\":1.98,\"end\":2.46}]"}
{"utterance_id": "2277-149897-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0021.flac", "task_type": "lexical_detection", "query": "busy", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"busy\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.01, "busy"]], "response_mode": "offline", "answer": "[{\"word\":\"busy\",\"start\":0.69,\"end\":1.01}]"}
{"utterance_id": "2803-154328-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0003.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[5.7, 6.45, "one"], [8.28, 8.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.7,\"end\":6.45},{\"word\":\"two\",\"start\":8.28,\"end\":8.8}]"}
{"utterance_id": "1993-147964-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0003.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.68, "cotton"]], "response_mode": "offline", "answer": "[{\"word\":\"cotton\",\"start\":1.34,\"end\":1.68}]"}
{"utterance_id": "7850-286674-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0011.flac", "task_type": "lexical_detection", "query": "pincers", "prompt": "After the audio ends, extract all occurrences of the word \"pincers\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.98, 5.51, "pincers"]], "response_mode": "offline", "answer": "[{\"word\":\"pincers\",\"start\":4.98,\"end\":5.51}]"}
{"utterance_id": "5338-284437-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0028.flac", "task_type": "lexical_detection", "query": "desirable", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"desirable\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.72, "desirable"]], "response_mode": "offline", "answer": "[{\"word\":\"desirable\",\"start\":1.12,\"end\":1.72}]"}
{"utterance_id": "1988-147956-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0004.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.16, 7.66, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":7.16,\"end\":7.66}]"}
{"utterance_id": "2078-142845-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0040.flac", "task_type": "lexical_counting", "query": "buns", "prompt": "After the audio ends, tally every instance of the word \"buns\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[2.05, 2.75, "one"], [6.07, 6.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.05,\"end\":2.75},{\"word\":\"two\",\"start\":6.07,\"end\":6.79}]"}
{"utterance_id": "6295-64301-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0022.flac", "task_type": "lexical_counting", "query": "love", "prompt": "Listen to the entire clip, then report the count sequence for the word \"love\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 2.79, "one"], [4.24, 4.81, "two"], [9.09, 9.35, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.5,\"end\":2.79},{\"word\":\"two\",\"start\":4.24,\"end\":4.81},{\"word\":\"three\",\"start\":9.09,\"end\":9.35}]"}
{"utterance_id": "6295-244435-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0040.flac", "task_type": "lexical_detection", "query": "signal", "prompt": "Once the audio has finished, identify every instance of the word \"signal\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.44, 2.82, "signal"]], "response_mode": "offline", "answer": "[{\"word\":\"signal\",\"start\":2.44,\"end\":2.82}]"}
{"utterance_id": "6241-61946-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0007.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Process the full audio offline. Find every instance of a weather term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 1.85, "snow tempest"]], "response_mode": "offline", "answer": "[{\"word\":\"snow tempest\",\"start\":0.4,\"end\":1.85}]"}
{"utterance_id": "3081-166546-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0036.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.32, 1.88, "thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"thousand\",\"start\":1.32,\"end\":1.88}]"}
{"utterance_id": "2078-142845-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0036.flac", "task_type": "lexical_detection", "query": "nine", "prompt": "When the utterance is complete, report all instances of the word \"nine\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.4, "nine"]], "response_mode": "offline", "answer": "[{\"word\":\"nine\",\"start\":2.87,\"end\":3.4}]"}
{"utterance_id": "6241-61946-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0021.flac", "task_type": "lexical_detection", "query": "tide", "prompt": "After the audio ends, extract all occurrences of the word \"tide\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.25, "tide"]], "response_mode": "offline", "answer": "[{\"word\":\"tide\",\"start\":0.78,\"end\":1.25}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.07, 1.78, "officers"]], "response_mode": "offline", "answer": "[{\"word\":\"officers\",\"start\":1.07,\"end\":1.78}]"}
{"utterance_id": "3000-15664-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0036.flac", "task_type": "lexical_detection", "query": "rushing", "prompt": "After listening to the whole recording, output every instance of the word \"rushing\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.03, 4.49, "rushing"]], "response_mode": "offline", "answer": "[{\"word\":\"rushing\",\"start\":4.03,\"end\":4.49}]"}
{"utterance_id": "6345-93302-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0024.flac", "task_type": "lexical_detection", "query": "suddenly", "prompt": "After listening to the whole recording, output every instance of the word \"suddenly\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.45, "suddenly"]], "response_mode": "offline", "answer": "[{\"word\":\"suddenly\",\"start\":0.91,\"end\":1.45}]"}
{"utterance_id": "3000-15664-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0005.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.93, 6.18, "wood"]], "response_mode": "offline", "answer": "[{\"word\":\"wood\",\"start\":5.93,\"end\":6.18}]"}
{"utterance_id": "3752-4944-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.78, "kirkland"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkland\",\"start\":2.23,\"end\":2.78}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.24, 1.84, "squires"], [11.57, 12.05, "servants"], [16.03, 17.05, "muleteers"]], "response_mode": "offline", "answer": "[{\"word\":\"squires\",\"start\":1.24,\"end\":1.84},{\"word\":\"servants\",\"start\":11.57,\"end\":12.05},{\"word\":\"muleteers\",\"start\":16.03,\"end\":17.05}]"}
{"utterance_id": "3170-137482-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0007.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.91, 14.6, "churchwardens"]], "response_mode": "offline", "answer": "[{\"word\":\"churchwardens\",\"start\":13.91,\"end\":14.6}]"}
{"utterance_id": "6241-66616-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.42, 1.67, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.42,\"end\":1.67}]"}
{"utterance_id": "2086-149220-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 5.93, "fowls"]], "response_mode": "offline", "answer": "[{\"word\":\"fowls\",\"start\":5.52,\"end\":5.93}]"}
{"utterance_id": "7976-110523-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0000.flac", "task_type": "lexical_detection", "query": "land", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"land\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.65, 5.07, "land"]], "response_mode": "offline", "answer": "[{\"word\":\"land\",\"start\":4.65,\"end\":5.07}]"}
{"utterance_id": "8842-304647-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0006.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Process the full audio offline. For every occurrence of an animal name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.54, "one"], [13.93, 14.24, "two"], [21.53, 22.07, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.14,\"end\":1.54},{\"word\":\"two\",\"start\":13.93,\"end\":14.24},{\"word\":\"three\",\"start\":21.53,\"end\":22.07}]"}
{"utterance_id": "3576-138058-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.27, 5.76, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":5.27,\"end\":5.76}]"}
{"utterance_id": "2277-149874-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0001.flac", "task_type": "lexical_detection", "query": "cars", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"cars\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.66, 3.09, "cars"]], "response_mode": "offline", "answer": "[{\"word\":\"cars\",\"start\":2.66,\"end\":3.09}]"}
{"utterance_id": "3081-166546-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0066.flac", "task_type": "lexical_detection", "query": "pleases", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"pleases\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 2.11, "pleases"]], "response_mode": "offline", "answer": "[{\"word\":\"pleases\",\"start\":1.39,\"end\":2.11}]"}
{"utterance_id": "6295-64301-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0015.flac", "task_type": "lexical_detection", "query": "swell", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"swell\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.45, 2.13, "swell"]], "response_mode": "offline", "answer": "[{\"word\":\"swell\",\"start\":1.45,\"end\":2.13}]"}
{"utterance_id": "6241-66616-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0025.flac", "task_type": "lexical_detection", "query": "rod", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"rod\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.38, 6.62, "rod"]], "response_mode": "offline", "answer": "[{\"word\":\"rod\",\"start\":6.38,\"end\":6.62}]"}
{"utterance_id": "6345-93306-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0022.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.09, 2.36, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"third\",\"start\":2.09,\"end\":2.36}]"}
{"utterance_id": "2902-9008-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0002.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen carefully to the entire clip, then report every instance of a time of day as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.56, 4.01, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":3.56,\"end\":4.01}]"}
{"utterance_id": "5694-64025-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0023.flac", "task_type": "lexical_counting", "query": "smart", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"smart\" (word, start, end). If none, reply [].", "labels": [[5.5, 5.86, "one"], [7.71, 8.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.5,\"end\":5.86},{\"word\":\"two\",\"start\":7.71,\"end\":8.16}]"}
{"utterance_id": "1993-147965-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0000.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.25, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.92,\"end\":2.25}]"}
{"utterance_id": "5536-43359-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0007.flac", "task_type": "lexical_detection", "query": "cousins", "prompt": "Once the audio has finished, identify every instance of the word \"cousins\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.6, 5.98, "cousins"], [13.14, 13.6, "cousins"]], "response_mode": "offline", "answer": "[{\"word\":\"cousins\",\"start\":5.6,\"end\":5.98},{\"word\":\"cousins\",\"start\":13.14,\"end\":13.6}]"}
{"utterance_id": "5895-34615-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0018.flac", "task_type": "lexical_detection", "query": "yellow", "prompt": "After the audio ends, extract all occurrences of the word \"yellow\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.6, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":1.26,\"end\":1.6}]"}
{"utterance_id": "1272-141231-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.51, 9.13, "thousands"]], "response_mode": "offline", "answer": "[{\"word\":\"thousands\",\"start\":8.51,\"end\":9.13}]"}
{"utterance_id": "1462-170142-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0036.flac", "task_type": "lexical_detection", "query": "stay", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"stay\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.92, 1.17, "stay"]], "response_mode": "offline", "answer": "[{\"word\":\"stay\",\"start\":0.92,\"end\":1.17}]"}
{"utterance_id": "1988-24833-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.55,\"end\":0.77}]"}
{"utterance_id": "8297-275154-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 1.09, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":0.42,\"end\":1.09}]"}
{"utterance_id": "1272-141231-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0001.flac", "task_type": "lexical_detection", "query": "tight", "prompt": "After listening to the whole recording, output every instance of the word \"tight\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.53, 3.82, "tight"]], "response_mode": "offline", "answer": "[{\"word\":\"tight\",\"start\":3.53,\"end\":3.82}]"}
{"utterance_id": "3576-138058-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[14.68, 15.61, "two thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"two thousand\",\"start\":14.68,\"end\":15.61}]"}
{"utterance_id": "6319-57405-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0000.flac", "task_type": "lexical_detection", "query": "prometheus", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"prometheus\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.99, 6.87, "prometheus"]], "response_mode": "offline", "answer": "[{\"word\":\"prometheus\",\"start\":5.99,\"end\":6.87}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.87, "camel"]], "response_mode": "offline", "answer": "[{\"word\":\"camel\",\"start\":2.44,\"end\":2.87}]"}
{"utterance_id": "2428-83699-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0037.flac", "task_type": "lexical_detection", "query": "retreating", "prompt": "Process the full audio offline. Find every instance of the word \"retreating\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.61, "retreating"]], "response_mode": "offline", "answer": "[{\"word\":\"retreating\",\"start\":1.16,\"end\":1.61}]"}
{"utterance_id": "8297-275156-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0012.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.82, 5.26, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":4.82,\"end\":5.26}]"}
{"utterance_id": "251-137823-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0023.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.83, "scientist"]], "response_mode": "offline", "answer": "[{\"word\":\"scientist\",\"start\":1.03,\"end\":1.83}]"}
{"utterance_id": "174-84280-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0013.flac", "task_type": "lexical_counting", "query": "found", "prompt": "Process the full audio offline. For every occurrence of the word \"found\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.19, "one"], [5.88, 6.27, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.72,\"end\":3.19},{\"word\":\"two\",\"start\":5.88,\"end\":6.27}]"}
{"utterance_id": "3000-15664-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0025.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.49, "hunters"]], "response_mode": "offline", "answer": "[{\"word\":\"hunters\",\"start\":1.86,\"end\":2.49}]"}
{"utterance_id": "3752-4943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0025.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.45, 4.99, "chaplain"]], "response_mode": "offline", "answer": "[{\"word\":\"chaplain\",\"start\":4.45,\"end\":4.99}]"}
{"utterance_id": "777-126732-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0042.flac", "task_type": "lexical_detection", "query": "backs", "prompt": "Process the full audio offline. Find every instance of the word \"backs\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.17, "backs"]], "response_mode": "offline", "answer": "[{\"word\":\"backs\",\"start\":1.79,\"end\":2.17}]"}
{"utterance_id": "2078-142845-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0020.flac", "task_type": "lexical_detection", "query": "acid", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"acid\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.41, 3.0, "acid"]], "response_mode": "offline", "answer": "[{\"word\":\"acid\",\"start\":2.41,\"end\":3.0}]"}
{"utterance_id": "777-126732-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0066.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.8, 4.28, "stephen"]], "response_mode": "offline", "answer": "[{\"word\":\"stephen\",\"start\":3.8,\"end\":4.28}]"}
{"utterance_id": "6295-244435-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.53, 4.98, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":4.53,\"end\":4.98}]"}
{"utterance_id": "2902-9008-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0003.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the entire clip, then report the count sequence for a body part as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.65, "one"], [8.66, 9.11, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.44,\"end\":1.65},{\"word\":\"two\",\"start\":8.66,\"end\":9.11}]"}
{"utterance_id": "1993-147149-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0000.flac", "task_type": "lexical_detection", "query": "infinite", "prompt": "Process the full audio offline. Find every instance of the word \"infinite\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.27, "infinite"]], "response_mode": "offline", "answer": "[{\"word\":\"infinite\",\"start\":0.74,\"end\":1.27}]"}
{"utterance_id": "5895-34622-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0004.flac", "task_type": "lexical_detection", "query": "wagon", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wagon\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.13, 3.5, "wagon"]], "response_mode": "offline", "answer": "[{\"word\":\"wagon\",\"start\":3.13,\"end\":3.5}]"}
{"utterance_id": "1919-142785-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0031.flac", "task_type": "lexical_detection", "query": "autumn", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"autumn\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.23, "autumn"]], "response_mode": "offline", "answer": "[{\"word\":\"autumn\",\"start\":1.82,\"end\":2.23}]"}
{"utterance_id": "2803-161169-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.18, 1.75, "father"], [3.42, 4.13, "children"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.18,\"end\":1.75},{\"word\":\"children\",\"start\":3.42,\"end\":4.13}]"}
{"utterance_id": "6319-57405-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0008.flac", "task_type": "lexical_detection", "query": "neighbors", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"neighbors\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.04, 8.47, "neighbors"]], "response_mode": "offline", "answer": "[{\"word\":\"neighbors\",\"start\":8.04,\"end\":8.47}]"}
{"utterance_id": "652-130726-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0013.flac", "task_type": "lexical_detection", "query": "commercial", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"commercial\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 2.96, "commercial"]], "response_mode": "offline", "answer": "[{\"word\":\"commercial\",\"start\":2.55,\"end\":2.96}]"}
{"utterance_id": "777-126732-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0048.flac", "task_type": "lexical_detection", "query": "shop", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"shop\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.45, 3.73, "shop"]], "response_mode": "offline", "answer": "[{\"word\":\"shop\",\"start\":3.45,\"end\":3.73}]"}
{"utterance_id": "2035-152373-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0011.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 1.79, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.5,\"end\":1.79}]"}
{"utterance_id": "1919-142785-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0057.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 0.88, "fat"], [2.91, 3.31, "crumbs"]], "response_mode": "offline", "answer": "[{\"word\":\"fat\",\"start\":0.56,\"end\":0.88},{\"word\":\"crumbs\",\"start\":2.91,\"end\":3.31}]"}
{"utterance_id": "174-168635-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0010.flac", "task_type": "lexical_counting", "query": "each", "prompt": "Listen to the entire clip, then report the count sequence for the word \"each\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 2.78, "one"], [4.54, 4.75, "two"], [6.57, 6.77, "three"], [8.49, 8.69, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.55,\"end\":2.78},{\"word\":\"two\",\"start\":4.54,\"end\":4.75},{\"word\":\"three\",\"start\":6.57,\"end\":6.77},{\"word\":\"four\",\"start\":8.49,\"end\":8.69}]"}
{"utterance_id": "7976-105575-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.86, 1.42, "regiment"]], "response_mode": "offline", "answer": "[{\"word\":\"regiment\",\"start\":0.86,\"end\":1.42}]"}
{"utterance_id": "2078-142845-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0013.flac", "task_type": "semantic_counting", "query": "color", "prompt": "After the audio ends, tally every instance of a color name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.46, 0.75, "one"], [1.98, 2.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.46,\"end\":0.75},{\"word\":\"two\",\"start\":1.98,\"end\":2.46}]"}
{"utterance_id": "1988-147956-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0022.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.83, "gold"]], "response_mode": "offline", "answer": "[{\"word\":\"gold\",\"start\":1.46,\"end\":1.83}]"}
{"utterance_id": "777-126732-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0020.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.1, 1.76, "ass"]], "response_mode": "offline", "answer": "[{\"word\":\"ass\",\"start\":1.1,\"end\":1.76}]"}
{"utterance_id": "2412-153948-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0006.flac", "task_type": "lexical_detection", "query": "ewes", "prompt": "When the utterance is complete, report all instances of the word \"ewes\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.12, 14.65, "ewes"]], "response_mode": "offline", "answer": "[{\"word\":\"ewes\",\"start\":14.12,\"end\":14.65}]"}
{"utterance_id": "1673-143396-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0007.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Process the full audio offline. Find every instance of an organization name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.62, 6.01, "churches"]], "response_mode": "offline", "answer": "[{\"word\":\"churches\",\"start\":5.62,\"end\":6.01}]"}
{"utterance_id": "1673-143396-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.85, "apollinaris"]], "response_mode": "offline", "answer": "[{\"word\":\"apollinaris\",\"start\":4.13,\"end\":4.85}]"}
{"utterance_id": "251-136532-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.65, "lattimer"]], "response_mode": "offline", "answer": "[{\"word\":\"lattimer\",\"start\":3.15,\"end\":3.65}]"}
{"utterance_id": "5895-34622-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0023.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "After listening to the whole recording, output timed count words for an organization name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[3.09, 3.72, "one"], [4.83, 5.44, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.09,\"end\":3.72},{\"word\":\"two\",\"start\":4.83,\"end\":5.44}]"}
{"utterance_id": "6319-64726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0004.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Process the full audio offline. For every occurrence of a number, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.43, "one"], [5.51, 5.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.79,\"end\":1.43},{\"word\":\"two\",\"start\":5.51,\"end\":5.67}]"}
{"utterance_id": "1988-148538-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0014.flac", "task_type": "lexical_detection", "query": "carries", "prompt": "Once the audio has finished, identify every instance of the word \"carries\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.3, 1.66, "carries"]], "response_mode": "offline", "answer": "[{\"word\":\"carries\",\"start\":1.3,\"end\":1.66}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, produce a running count of an ordinal number: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.75, 1.07, "one"], [14.06, 14.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.75,\"end\":1.07},{\"word\":\"two\",\"start\":14.06,\"end\":14.45}]"}
{"utterance_id": "6319-275224-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.82, "mamma"]], "response_mode": "offline", "answer": "[{\"word\":\"mamma\",\"start\":2.44,\"end\":2.82}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.11, 2.43, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":2.11,\"end\":2.43}]"}
{"utterance_id": "6295-244435-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0031.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen to the full audio. After it finishes, find every instance of a state or province name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.85, "south carolina"]], "response_mode": "offline", "answer": "[{\"word\":\"south carolina\",\"start\":1.79,\"end\":2.85}]"}
{"utterance_id": "3752-4943-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0005.flac", "task_type": "lexical_detection", "query": "reverence", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"reverence\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.57, 2.01, "reverence"]], "response_mode": "offline", "answer": "[{\"word\":\"reverence\",\"start\":1.57,\"end\":2.01}]"}
{"utterance_id": "5536-43359-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0001.flac", "task_type": "lexical_detection", "query": "creation", "prompt": "Once the audio has finished, identify every instance of the word \"creation\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.36, 9.9, "creation"]], "response_mode": "offline", "answer": "[{\"word\":\"creation\",\"start\":9.36,\"end\":9.9}]"}
{"utterance_id": "7976-105575-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0014.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.41, 4.01, "division"]], "response_mode": "offline", "answer": "[{\"word\":\"division\",\"start\":3.41,\"end\":4.01}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "lexical_detection", "query": "restraint", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"restraint\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.69, 10.16, "restraint"]], "response_mode": "offline", "answer": "[{\"word\":\"restraint\",\"start\":9.69,\"end\":10.16}]"}
{"utterance_id": "251-118436-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0023.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.98, 2.6, "asura"]], "response_mode": "offline", "answer": "[{\"word\":\"asura\",\"start\":1.98,\"end\":2.6}]"}
{"utterance_id": "2078-142845-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0019.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.28, 2.83, "flour"], [6.95, 7.38, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"flour\",\"start\":2.28,\"end\":2.83},{\"word\":\"food\",\"start\":6.95,\"end\":7.38}]"}
{"utterance_id": "6241-61943-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.89, 7.44, "hamlet"]], "response_mode": "offline", "answer": "[{\"word\":\"hamlet\",\"start\":6.89,\"end\":7.44}]"}
{"utterance_id": "84-121123-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0016.flac", "task_type": "lexical_detection", "query": "remain", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"remain\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.87, 5.19, "remain"]], "response_mode": "offline", "answer": "[{\"word\":\"remain\",\"start\":4.87,\"end\":5.19}]"}
{"utterance_id": "5694-64025-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0010.flac", "task_type": "lexical_counting", "query": "battlefields", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"battlefields\" with word/start/end. If none, reply [].", "labels": [[1.72, 2.62, "one"], [3.66, 4.52, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.72,\"end\":2.62},{\"word\":\"two\",\"start\":3.66,\"end\":4.52}]"}
{"utterance_id": "777-126732-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0038.flac", "task_type": "lexical_detection", "query": "paused", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"paused\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.11, "paused"]], "response_mode": "offline", "answer": "[{\"word\":\"paused\",\"start\":0.59,\"end\":1.11}]"}
{"utterance_id": "3000-15664-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0010.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.33, 8.18, "vegetation"]], "response_mode": "offline", "answer": "[{\"word\":\"vegetation\",\"start\":7.33,\"end\":8.18}]"}
{"utterance_id": "3752-4944-0065", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0065.flac", "task_type": "lexical_detection", "query": "teaspoonful", "prompt": "Once the audio has finished, identify every instance of the word \"teaspoonful\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.11, 1.96, "teaspoonful"]], "response_mode": "offline", "answer": "[{\"word\":\"teaspoonful\",\"start\":1.11,\"end\":1.96}]"}
{"utterance_id": "251-137823-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0016.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.52, 5.39, "superintendent"]], "response_mode": "offline", "answer": "[{\"word\":\"superintendent\",\"start\":4.52,\"end\":5.39}]"}
{"utterance_id": "174-50561-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0013.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a plant name (word, start, end). If none, reply [].", "labels": [[1.69, 2.49, "one"], [8.97, 9.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.69,\"end\":2.49},{\"word\":\"two\",\"start\":8.97,\"end\":9.32}]"}
{"utterance_id": "5694-64025-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.76, "mule"]], "response_mode": "offline", "answer": "[{\"word\":\"mule\",\"start\":2.25,\"end\":2.76}]"}
{"utterance_id": "8842-302201-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0002.flac", "task_type": "lexical_counting", "query": "thou", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"thou\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[6.79, 6.96, "one"], [8.24, 8.4, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.79,\"end\":6.96},{\"word\":\"two\",\"start\":8.24,\"end\":8.4}]"}
{"utterance_id": "8842-302201-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0013.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.71, 3.11, "earth"]], "response_mode": "offline", "answer": "[{\"word\":\"earth\",\"start\":2.71,\"end\":3.11}]"}
{"utterance_id": "8842-302203-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0002.flac", "task_type": "lexical_detection", "query": "whereby", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"whereby\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.18, 0.58, "whereby"]], "response_mode": "offline", "answer": "[{\"word\":\"whereby\",\"start\":0.18,\"end\":0.58}]"}
{"utterance_id": "6295-244435-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0034.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "You will hear a complete utterance. When it ends, list every instance of a state or province name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 3.21, "kentucky"]], "response_mode": "offline", "answer": "[{\"word\":\"kentucky\",\"start\":2.5,\"end\":3.21}]"}
{"utterance_id": "2277-149874-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0008.flac", "task_type": "lexical_detection", "query": "disturbed", "prompt": "When the utterance is complete, report all instances of the word \"disturbed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.69, 3.13, "disturbed"]], "response_mode": "offline", "answer": "[{\"word\":\"disturbed\",\"start\":2.69,\"end\":3.13}]"}
{"utterance_id": "8842-302203-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0003.flac", "task_type": "lexical_detection", "query": "opened", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"opened\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.04, 1.34, "opened"]], "response_mode": "offline", "answer": "[{\"word\":\"opened\",\"start\":1.04,\"end\":1.34}]"}
{"utterance_id": "777-126732-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0031.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.31, "fingers"]], "response_mode": "offline", "answer": "[{\"word\":\"fingers\",\"start\":2.81,\"end\":3.31}]"}
{"utterance_id": "6345-93302-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0013.flac", "task_type": "lexical_detection", "query": "frightfully", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"frightfully\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.98, 1.5, "frightfully"]], "response_mode": "offline", "answer": "[{\"word\":\"frightfully\",\"start\":0.98,\"end\":1.5}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "Listen to the full audio. Count every occurrence of an ordinal number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.36, "one"], [2.97, 3.44, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.91,\"end\":1.36},{\"word\":\"two\",\"start\":2.97,\"end\":3.44}]"}
{"utterance_id": "2428-83705-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0001.flac", "task_type": "lexical_detection", "query": "snellings", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"snellings\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 6.35, "snellings"]], "response_mode": "offline", "answer": "[{\"word\":\"snellings\",\"start\":5.55,\"end\":6.35}]"}
{"utterance_id": "777-126732-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0014.flac", "task_type": "lexical_detection", "query": "throat", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"throat\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.6, 7.06, "throat"]], "response_mode": "offline", "answer": "[{\"word\":\"throat\",\"start\":6.6,\"end\":7.06}]"}
{"utterance_id": "2803-161169-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0011.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.06, 4.45, "wood"]], "response_mode": "offline", "answer": "[{\"word\":\"wood\",\"start\":4.06,\"end\":4.45}]"}
{"utterance_id": "2035-147961-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0036.flac", "task_type": "lexical_detection", "query": "head", "prompt": "After the audio ends, extract all occurrences of the word \"head\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.4, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":2.17,\"end\":2.4}]"}
{"utterance_id": "5895-34629-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0033.flac", "task_type": "lexical_detection", "query": "fair", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"fair\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.28, "fair"]], "response_mode": "offline", "answer": "[{\"word\":\"fair\",\"start\":2.06,\"end\":2.28}]"}
{"utterance_id": "2277-149896-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0003.flac", "task_type": "lexical_detection", "query": "position", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"position\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.66, "position"]], "response_mode": "offline", "answer": "[{\"word\":\"position\",\"start\":2.08,\"end\":2.66}]"}
{"utterance_id": "2086-149220-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0008.flac", "task_type": "lexical_counting", "query": "race", "prompt": "Once the audio has finished, list each occurrence of the word \"race\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.65, 1.91, "one"], [4.04, 4.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.65,\"end\":1.91},{\"word\":\"two\",\"start\":4.04,\"end\":4.29}]"}
{"utterance_id": "1993-147964-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen carefully to the entire clip, then report every instance of a date as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.97, 3.48, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":2.97,\"end\":3.48}]"}
{"utterance_id": "7976-105575-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0004.flac", "task_type": "lexical_detection", "query": "death", "prompt": "After the audio ends, extract all occurrences of the word \"death\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.32, 5.77, "death"]], "response_mode": "offline", "answer": "[{\"word\":\"death\",\"start\":5.32,\"end\":5.77}]"}
{"utterance_id": "3853-163249-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0053.flac", "task_type": "lexical_detection", "query": "adorned", "prompt": "After the audio ends, extract all occurrences of the word \"adorned\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.64, 5.0, "adorned"]], "response_mode": "offline", "answer": "[{\"word\":\"adorned\",\"start\":4.64,\"end\":5.0}]"}
{"utterance_id": "6295-244435-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0032.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[0.75, 1.1, "blood"]], "response_mode": "offline", "answer": "[{\"word\":\"blood\",\"start\":0.75,\"end\":1.1}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "lexical_counting", "query": "softly", "prompt": "After listening to the whole recording, output timed count words for the word \"softly\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.49, 2.13, "one"], [11.84, 12.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.49,\"end\":2.13},{\"word\":\"two\",\"start\":11.84,\"end\":12.35}]"}
{"utterance_id": "3752-4944-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0048.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.25, 1.67, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":1.25,\"end\":1.67}]"}
{"utterance_id": "5895-34615-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.01, "hair"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":1.6,\"end\":2.01}]"}
{"utterance_id": "3081-166546-0071", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0071.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.39, 4.13, "one"], [5.16, 6.02, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.39,\"end\":4.13},{\"word\":\"two\",\"start\":5.16,\"end\":6.02}]"}
{"utterance_id": "6313-66129-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0032.flac", "task_type": "lexical_detection", "query": "pony", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pony\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.74, "pony"]], "response_mode": "offline", "answer": "[{\"word\":\"pony\",\"start\":1.2,\"end\":1.74}]"}
{"utterance_id": "3853-163249-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0045.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.2, "hearts"], [6.2, 6.51, "tongue"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":3.86,\"end\":4.2},{\"word\":\"tongue\",\"start\":6.2,\"end\":6.51}]"}
{"utterance_id": "6313-66129-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0025.flac", "task_type": "lexical_detection", "query": "pommel", "prompt": "After listening to the whole recording, output every instance of the word \"pommel\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.14, "pommel"]], "response_mode": "offline", "answer": "[{\"word\":\"pommel\",\"start\":0.8,\"end\":1.14}]"}
{"utterance_id": "8842-302203-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0011.flac", "task_type": "lexical_detection", "query": "begins", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"begins\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.96, 1.32, "begins"]], "response_mode": "offline", "answer": "[{\"word\":\"begins\",\"start\":0.96,\"end\":1.32}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "lexical_counting", "query": "work", "prompt": "Listen to the entire clip, then report the count sequence for the word \"work\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 4.79, "one"], [15.06, 15.5, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.47,\"end\":4.79},{\"word\":\"two\",\"start\":15.06,\"end\":15.5}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "When the utterance is complete, report all instances of a disease name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.66, 4.18, "wounded"]], "response_mode": "offline", "answer": "[{\"word\":\"wounded\",\"start\":3.66,\"end\":4.18}]"}
{"utterance_id": "2035-147960-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.61, 0.82, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":0.61,\"end\":0.82}]"}
{"utterance_id": "8842-302203-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0011.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.43, 0.71, "second"], [3.91, 4.2, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.43,\"end\":0.71},{\"word\":\"third\",\"start\":3.91,\"end\":4.2}]"}
{"utterance_id": "1919-142785-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0010.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.21, 5.74, "chester"]], "response_mode": "offline", "answer": "[{\"word\":\"chester\",\"start\":5.21,\"end\":5.74}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen carefully to the entire clip, then report every instance of a country name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.9, "german"]], "response_mode": "offline", "answer": "[{\"word\":\"german\",\"start\":0.5,\"end\":0.9}]"}
{"utterance_id": "5338-24615-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.51, 10.98, "waverley"]], "response_mode": "offline", "answer": "[{\"word\":\"waverley\",\"start\":10.51,\"end\":10.98}]"}
{"utterance_id": "3752-4944-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0064.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After listening to the whole recording, output every instance of a disease name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.43, "disease"]], "response_mode": "offline", "answer": "[{\"word\":\"disease\",\"start\":2.72,\"end\":3.43}]"}
{"utterance_id": "3081-166546-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0019.flac", "task_type": "lexical_detection", "query": "returned", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"returned\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.01, 2.5, "returned"]], "response_mode": "offline", "answer": "[{\"word\":\"returned\",\"start\":2.01,\"end\":2.5}]"}
{"utterance_id": "6345-93306-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.31, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.1,\"end\":3.31}]"}
{"utterance_id": "3752-4944-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0051.flac", "task_type": "lexical_detection", "query": "mere", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"mere\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.28, 1.55, "mere"]], "response_mode": "offline", "answer": "[{\"word\":\"mere\",\"start\":1.28,\"end\":1.55}]"}
{"utterance_id": "3081-166546-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0014.flac", "task_type": "lexical_detection", "query": "why", "prompt": "When the utterance is complete, report all instances of the word \"why\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.59, 3.04, "why"]], "response_mode": "offline", "answer": "[{\"word\":\"why\",\"start\":2.59,\"end\":3.04}]"}
{"utterance_id": "7850-286674-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0016.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 1.87, "legs"], [10.36, 10.82, "tail"]], "response_mode": "offline", "answer": "[{\"word\":\"legs\",\"start\":1.43,\"end\":1.87},{\"word\":\"tail\",\"start\":10.36,\"end\":10.82}]"}
{"utterance_id": "1988-147956-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0013.flac", "task_type": "lexical_detection", "query": "smart", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"smart\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.72, "smart"]], "response_mode": "offline", "answer": "[{\"word\":\"smart\",\"start\":2.33,\"end\":2.72}]"}
{"utterance_id": "2277-149897-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0005.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.35, 6.72, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":6.35,\"end\":6.72}]"}
{"utterance_id": "652-130737-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0006.flac", "task_type": "lexical_counting", "query": "wines", "prompt": "After the audio ends, tally every instance of the word \"wines\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.9, 1.43, "one"], [4.15, 4.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.9,\"end\":1.43},{\"word\":\"two\",\"start\":4.15,\"end\":4.59}]"}
{"utterance_id": "1462-170138-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0011.flac", "task_type": "lexical_detection", "query": "singing", "prompt": "After listening to the whole recording, output every instance of the word \"singing\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.87, 7.28, "singing"]], "response_mode": "offline", "answer": "[{\"word\":\"singing\",\"start\":6.87,\"end\":7.28}]"}
{"utterance_id": "7850-286674-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0004.flac", "task_type": "lexical_detection", "query": "walk", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"walk\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.05, 5.28, "walk"]], "response_mode": "offline", "answer": "[{\"word\":\"walk\",\"start\":5.05,\"end\":5.28}]"}
{"utterance_id": "6313-66129-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0030.flac", "task_type": "lexical_detection", "query": "demanded", "prompt": "After the audio ends, extract all occurrences of the word \"demanded\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.99, "demanded"]], "response_mode": "offline", "answer": "[{\"word\":\"demanded\",\"start\":1.42,\"end\":1.99}]"}
{"utterance_id": "2277-149897-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0004.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.6, "cigar"]], "response_mode": "offline", "answer": "[{\"word\":\"cigar\",\"start\":2.14,\"end\":2.6}]"}
{"utterance_id": "2412-153947-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0010.flac", "task_type": "lexical_detection", "query": "hall", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hall\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.94, 4.26, "hall"]], "response_mode": "offline", "answer": "[{\"word\":\"hall\",\"start\":3.94,\"end\":4.26}]"}
{"utterance_id": "174-168635-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.07, "cosette"]], "response_mode": "offline", "answer": "[{\"word\":\"cosette\",\"start\":0.51,\"end\":1.07}]"}
{"utterance_id": "1462-170142-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.24, 1.83, "shoulders"], [3.09, 3.45, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulders\",\"start\":1.24,\"end\":1.83},{\"word\":\"head\",\"start\":3.09,\"end\":3.45}]"}
{"utterance_id": "5895-34629-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0009.flac", "task_type": "lexical_detection", "query": "prepared", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"prepared\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.52, "prepared"]], "response_mode": "offline", "answer": "[{\"word\":\"prepared\",\"start\":2.95,\"end\":3.52}]"}
{"utterance_id": "2277-149896-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0033.flac", "task_type": "lexical_detection", "query": "rang", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"rang\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.58, 0.81, "rang"]], "response_mode": "offline", "answer": "[{\"word\":\"rang\",\"start\":0.58,\"end\":0.81}]"}
{"utterance_id": "6313-76958-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0006.flac", "task_type": "lexical_detection", "query": "fit", "prompt": "When the utterance is complete, report all instances of the word \"fit\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.35, "fit"]], "response_mode": "offline", "answer": "[{\"word\":\"fit\",\"start\":1.11,\"end\":1.35}]"}
{"utterance_id": "6241-61946-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0001.flac", "task_type": "lexical_detection", "query": "steady", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"steady\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.6, 4.06, "steady"]], "response_mode": "offline", "answer": "[{\"word\":\"steady\",\"start\":3.6,\"end\":4.06}]"}
{"utterance_id": "6319-57405-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.07, 6.55, "bones"], [7.9, 8.43, "shoulders"]], "response_mode": "offline", "answer": "[{\"word\":\"bones\",\"start\":6.07,\"end\":6.55},{\"word\":\"shoulders\",\"start\":7.9,\"end\":8.43}]"}
{"utterance_id": "1462-170142-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0016.flac", "task_type": "lexical_detection", "query": "spoils", "prompt": "When the utterance is complete, report all instances of the word \"spoils\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.96, 5.43, "spoils"]], "response_mode": "offline", "answer": "[{\"word\":\"spoils\",\"start\":4.96,\"end\":5.43}]"}
{"utterance_id": "6241-66616-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0014.flac", "task_type": "lexical_detection", "query": "price", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"price\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.17, 4.49, "price"]], "response_mode": "offline", "answer": "[{\"word\":\"price\",\"start\":4.17,\"end\":4.49}]"}
{"utterance_id": "3853-163249-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0042.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Process the full audio offline. Find every instance of a money or currency term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.45, 11.87, "pension"]], "response_mode": "offline", "answer": "[{\"word\":\"pension\",\"start\":11.45,\"end\":11.87}]"}
{"utterance_id": "8297-275156-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.93, 5.31, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.93,\"end\":5.31}]"}
{"utterance_id": "5895-34622-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0017.flac", "task_type": "lexical_detection", "query": "dressing", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"dressing\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.13, 6.55, "dressing"]], "response_mode": "offline", "answer": "[{\"word\":\"dressing\",\"start\":6.13,\"end\":6.55}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.56, 6.87, "one"], [8.37, 8.66, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.56,\"end\":6.87},{\"word\":\"one\",\"start\":8.37,\"end\":8.66}]"}
{"utterance_id": "1673-143396-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[11.18, 11.68, "flesh"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":11.18,\"end\":11.68}]"}
{"utterance_id": "3752-4944-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0001.flac", "task_type": "lexical_detection", "query": "almost", "prompt": "Once the audio has finished, identify every instance of the word \"almost\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.32, 4.16, "almost"]], "response_mode": "offline", "answer": "[{\"word\":\"almost\",\"start\":3.32,\"end\":4.16}]"}
{"utterance_id": "5338-284437-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0021.flac", "task_type": "lexical_detection", "query": "pay", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"pay\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.08, 6.36, "pay"]], "response_mode": "offline", "answer": "[{\"word\":\"pay\",\"start\":6.08,\"end\":6.36}]"}
{"utterance_id": "1993-147149-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0004.flac", "task_type": "lexical_detection", "query": "jem", "prompt": "When the utterance is complete, report all instances of the word \"jem\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.45, "jem"]], "response_mode": "offline", "answer": "[{\"word\":\"jem\",\"start\":1.23,\"end\":1.45}]"}
{"utterance_id": "6295-64301-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0022.flac", "task_type": "lexical_detection", "query": "hast", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hast\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.34, 0.62, "hast"]], "response_mode": "offline", "answer": "[{\"word\":\"hast\",\"start\":0.34,\"end\":0.62}]"}
{"utterance_id": "8297-275154-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0005.flac", "task_type": "lexical_detection", "query": "beauty", "prompt": "Process the full audio offline. Find every instance of the word \"beauty\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.2, "beauty"]], "response_mode": "offline", "answer": "[{\"word\":\"beauty\",\"start\":0.85,\"end\":1.2}]"}
{"utterance_id": "3000-15664-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0027.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.71, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":1.21,\"end\":1.71}]"}
{"utterance_id": "3536-23268-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0003.flac", "task_type": "lexical_detection", "query": "abroad", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"abroad\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.36, 3.86, "abroad"]], "response_mode": "offline", "answer": "[{\"word\":\"abroad\",\"start\":3.36,\"end\":3.86}]"}
{"utterance_id": "7850-281318-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0010.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.55, 2.23, "magpie"], [3.59, 4.32, "magpie"]], "response_mode": "offline", "answer": "[{\"word\":\"magpie\",\"start\":1.55,\"end\":2.23},{\"word\":\"magpie\",\"start\":3.59,\"end\":4.32}]"}
{"utterance_id": "2035-152373-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.43, 2.86, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":2.43,\"end\":2.86}]"}
{"utterance_id": "422-122949-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0012.flac", "task_type": "lexical_counting", "query": "individual", "prompt": "When the utterance is complete, report how the word \"individual\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.83, 12.48, "one"], [13.52, 14.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":11.83,\"end\":12.48},{\"word\":\"two\",\"start\":13.52,\"end\":14.34}]"}
{"utterance_id": "2428-83705-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0009.flac", "task_type": "lexical_detection", "query": "that's", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"that's\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.35, 0.64, "that's"]], "response_mode": "offline", "answer": "[{\"word\":\"that's\",\"start\":0.35,\"end\":0.64}]"}
{"utterance_id": "7850-281318-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0010.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Process the full audio offline. Find every instance of a time of day and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.22, 9.68, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":9.22,\"end\":9.68}]"}
{"utterance_id": "84-121123-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.29, 0.71, "father"], [2.46, 2.9, "father"], [8.6, 9.07, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":0.29,\"end\":0.71},{\"word\":\"father\",\"start\":2.46,\"end\":2.9},{\"word\":\"father\",\"start\":8.6,\"end\":9.07}]"}
{"utterance_id": "3000-15664-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0006.flac", "task_type": "lexical_detection", "query": "required", "prompt": "When the utterance is complete, report all instances of the word \"required\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.63, 2.33, "required"]], "response_mode": "offline", "answer": "[{\"word\":\"required\",\"start\":1.63,\"end\":2.33}]"}
{"utterance_id": "7850-73752-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0010.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Process the full audio offline. For every occurrence of a person's name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.91, 11.45, "one"], [12.97, 13.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":10.91,\"end\":11.45},{\"word\":\"two\",\"start\":12.97,\"end\":13.68}]"}
{"utterance_id": "3752-4944-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0015.flac", "task_type": "lexical_detection", "query": "meekin", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"meekin\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.51, "meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"meekin\",\"start\":1.21,\"end\":1.51}]"}
{"utterance_id": "5895-34629-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 0.9, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":0.35,\"end\":0.9}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_counting", "query": "organization", "prompt": "When the utterance is complete, report how an organization name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.59, 6.4, "one"], [8.25, 8.78, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.59,\"end\":6.4},{\"word\":\"two\",\"start\":8.25,\"end\":8.78}]"}
{"utterance_id": "6345-93302-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0019.flac", "task_type": "lexical_detection", "query": "wounded", "prompt": "When the utterance is complete, report all instances of the word \"wounded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.06, "wounded"]], "response_mode": "offline", "answer": "[{\"word\":\"wounded\",\"start\":1.64,\"end\":2.06}]"}
{"utterance_id": "3536-23268-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0026.flac", "task_type": "lexical_detection", "query": "duties", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"duties\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.22, "duties"]], "response_mode": "offline", "answer": "[{\"word\":\"duties\",\"start\":0.85,\"end\":1.22}]"}
{"utterance_id": "5694-64038-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0023.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.79, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"third\",\"start\":0.54,\"end\":0.79}]"}
{"utterance_id": "8842-302201-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0000.flac", "task_type": "lexical_counting", "query": "things", "prompt": "Listen to the entire clip, then report the count sequence for the word \"things\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.42, "one"], [6.11, 6.5, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.13,\"end\":4.42},{\"word\":\"two\",\"start\":6.11,\"end\":6.5}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[4.01, 4.37, "one"], [7.07, 7.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.01,\"end\":4.37},{\"word\":\"two\",\"start\":7.07,\"end\":7.47}]"}
{"utterance_id": "3752-4944-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0050.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.07, 1.54, "officer"]], "response_mode": "offline", "answer": "[{\"word\":\"officer\",\"start\":1.07,\"end\":1.54}]"}
{"utterance_id": "251-137823-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.78, 1.22, "engineer"]], "response_mode": "offline", "answer": "[{\"word\":\"engineer\",\"start\":0.78,\"end\":1.22}]"}
{"utterance_id": "3536-23268-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0024.flac", "task_type": "lexical_detection", "query": "dorriforth", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"dorriforth\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.14, "dorriforth"]], "response_mode": "offline", "answer": "[{\"word\":\"dorriforth\",\"start\":0.53,\"end\":1.14}]"}
{"utterance_id": "2277-149897-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0018.flac", "task_type": "lexical_detection", "query": "post", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"post\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.38, 2.67, "post"]], "response_mode": "offline", "answer": "[{\"word\":\"post\",\"start\":2.38,\"end\":2.67}]"}
{"utterance_id": "5895-34615-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0018.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.6, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":1.26,\"end\":1.6}]"}
{"utterance_id": "5338-284437-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0029.flac", "task_type": "lexical_detection", "query": "big", "prompt": "When the utterance is complete, report all instances of the word \"big\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.67, 4.99, "big"]], "response_mode": "offline", "answer": "[{\"word\":\"big\",\"start\":4.67,\"end\":4.99}]"}
{"utterance_id": "1988-148538-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0005.flac", "task_type": "lexical_detection", "query": "hesitate", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"hesitate\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.26, 7.8, "hesitate"]], "response_mode": "offline", "answer": "[{\"word\":\"hesitate\",\"start\":7.26,\"end\":7.8}]"}
{"utterance_id": "6313-76958-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.37, 2.2, "curley adams"]], "response_mode": "offline", "answer": "[{\"word\":\"curley adams\",\"start\":1.37,\"end\":2.2}]"}
{"utterance_id": "6295-244435-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.24, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":0.85,\"end\":1.24}]"}
{"utterance_id": "5895-34622-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0001.flac", "task_type": "lexical_detection", "query": "therefore", "prompt": "When the utterance is complete, report all instances of the word \"therefore\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.25, 4.67, "therefore"]], "response_mode": "offline", "answer": "[{\"word\":\"therefore\",\"start\":4.25,\"end\":4.67}]"}
{"utterance_id": "1919-142785-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.59, "salt"], [10.11, 10.45, "salt"]], "response_mode": "offline", "answer": "[{\"word\":\"salt\",\"start\":1.2,\"end\":1.59},{\"word\":\"salt\",\"start\":10.11,\"end\":10.45}]"}
{"utterance_id": "5338-24615-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0010.flac", "task_type": "lexical_detection", "query": "stables", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"stables\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.21, 0.75, "stables"]], "response_mode": "offline", "answer": "[{\"word\":\"stables\",\"start\":0.21,\"end\":0.75}]"}
{"utterance_id": "251-137823-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0000.flac", "task_type": "lexical_detection", "query": "replied", "prompt": "After the audio ends, extract all occurrences of the word \"replied\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.39, 3.0, "replied"]], "response_mode": "offline", "answer": "[{\"word\":\"replied\",\"start\":2.39,\"end\":3.0}]"}
{"utterance_id": "6313-66125-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0007.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a person's name with word/start/end. If none, reply [].", "labels": [[1.33, 1.72, "one"], [3.89, 4.7, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.33,\"end\":1.72},{\"word\":\"two\",\"start\":3.89,\"end\":4.7}]"}
{"utterance_id": "1272-141231-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.46, 2.89, "muscles"]], "response_mode": "offline", "answer": "[{\"word\":\"muscles\",\"start\":2.46,\"end\":2.89}]"}
{"utterance_id": "7976-105575-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0002.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.75, 1.21, "brigade"]], "response_mode": "offline", "answer": "[{\"word\":\"brigade\",\"start\":0.75,\"end\":1.21}]"}
{"utterance_id": "6313-66125-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 0.84, "lige"], [6.78, 7.14, "tad's"]], "response_mode": "offline", "answer": "[{\"word\":\"lige\",\"start\":0.26,\"end\":0.84},{\"word\":\"tad's\",\"start\":6.78,\"end\":7.14}]"}
{"utterance_id": "3081-166546-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0044.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.41, 3.98, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":3.41,\"end\":3.98}]"}
{"utterance_id": "5694-64038-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0017.flac", "task_type": "lexical_counting", "query": "how", "prompt": "Listen to the full audio. Count every occurrence of the word \"how\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.75, "one"], [3.32, 3.49, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.75},{\"word\":\"two\",\"start\":3.32,\"end\":3.49}]"}
{"utterance_id": "251-136532-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 3.01, "martians"], [4.25, 4.74, "martians"]], "response_mode": "offline", "answer": "[{\"word\":\"martians\",\"start\":2.29,\"end\":3.01},{\"word\":\"martians\",\"start\":4.25,\"end\":4.74}]"}
{"utterance_id": "3752-4944-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0048.flac", "task_type": "lexical_detection", "query": "god", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"god\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.67, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":1.25,\"end\":1.67}]"}
{"utterance_id": "1919-142785-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0050.flac", "task_type": "lexical_detection", "query": "prevent", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"prevent\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.28, "prevent"]], "response_mode": "offline", "answer": "[{\"word\":\"prevent\",\"start\":3.86,\"end\":4.28}]"}
{"utterance_id": "2078-142845-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0038.flac", "task_type": "lexical_detection", "query": "bake", "prompt": "After the audio ends, extract all occurrences of the word \"bake\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.02, 3.34, "bake"]], "response_mode": "offline", "answer": "[{\"word\":\"bake\",\"start\":3.02,\"end\":3.34}]"}
{"utterance_id": "3576-138058-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0039.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.18, 0.38, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.18,\"end\":0.38}]"}
{"utterance_id": "2803-154320-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0007.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.42, 0.93, "one"], [3.65, 4.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.42,\"end\":0.93},{\"word\":\"two\",\"start\":3.65,\"end\":4.18}]"}
{"utterance_id": "1272-135031-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.26, "betsy"]], "response_mode": "offline", "answer": "[{\"word\":\"betsy\",\"start\":1.84,\"end\":2.26}]"}
{"utterance_id": "1673-143397-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0001.flac", "task_type": "lexical_detection", "query": "multitude", "prompt": "Once the audio has finished, identify every instance of the word \"multitude\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.67, 7.2, "multitude"]], "response_mode": "offline", "answer": "[{\"word\":\"multitude\",\"start\":6.67,\"end\":7.2}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.96, "sancho"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho\",\"start\":0.47,\"end\":0.96}]"}
{"utterance_id": "2035-147960-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0001.flac", "task_type": "lexical_detection", "query": "adventure", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"adventure\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.55, "adventure"]], "response_mode": "offline", "answer": "[{\"word\":\"adventure\",\"start\":2.05,\"end\":2.55}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After the audio ends, tally every instance of a food name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[6.97, 7.31, "one"], [9.57, 10.02, "two"], [15.12, 15.67, "three"], [19.24, 19.62, "four"], [21.69, 22.11, "five"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.97,\"end\":7.31},{\"word\":\"two\",\"start\":9.57,\"end\":10.02},{\"word\":\"three\",\"start\":15.12,\"end\":15.67},{\"word\":\"four\",\"start\":19.24,\"end\":19.62},{\"word\":\"five\",\"start\":21.69,\"end\":22.11}]"}
{"utterance_id": "6313-66125-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0002.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.42, 2.91, "bushes"]], "response_mode": "offline", "answer": "[{\"word\":\"bushes\",\"start\":2.42,\"end\":2.91}]"}
{"utterance_id": "174-84280-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0015.flac", "task_type": "lexical_detection", "query": "destroyed", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"destroyed\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.96, 1.64, "destroyed"]], "response_mode": "offline", "answer": "[{\"word\":\"destroyed\",\"start\":0.96,\"end\":1.64}]"}
{"utterance_id": "7850-281318-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.85, 7.38, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":6.85,\"end\":7.38}]"}
{"utterance_id": "652-129742-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0001.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.34, 1.88, "one"], [4.67, 5.1, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.34,\"end\":1.88},{\"word\":\"two\",\"start\":4.67,\"end\":5.1}]"}
{"utterance_id": "2428-83699-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0001.flac", "task_type": "lexical_detection", "query": "festive", "prompt": "Once the audio has finished, identify every instance of the word \"festive\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.4, 0.99, "festive"]], "response_mode": "offline", "answer": "[{\"word\":\"festive\",\"start\":0.4,\"end\":0.99}]"}
{"utterance_id": "3853-163249-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0034.flac", "task_type": "lexical_detection", "query": "satisfied", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"satisfied\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 2.45, "satisfied"]], "response_mode": "offline", "answer": "[{\"word\":\"satisfied\",\"start\":1.56,\"end\":2.45}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 1.23, "reverend meekin"], [3.58, 4.35, "reverend meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"reverend meekin\",\"start\":0.33,\"end\":1.23},{\"word\":\"reverend meekin\",\"start\":3.58,\"end\":4.35}]"}
{"utterance_id": "777-126732-0061", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0061.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[3.66, 4.15, "forehead"]], "response_mode": "offline", "answer": "[{\"word\":\"forehead\",\"start\":3.66,\"end\":4.15}]"}
{"utterance_id": "422-122949-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.34, 9.86, "forest"]], "response_mode": "offline", "answer": "[{\"word\":\"forest\",\"start\":9.34,\"end\":9.86}]"}
{"utterance_id": "3081-166546-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0013.flac", "task_type": "lexical_detection", "query": "eyes", "prompt": "When the utterance is complete, report all instances of the word \"eyes\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.6, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.2,\"end\":2.6}]"}
{"utterance_id": "1673-143396-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0011.flac", "task_type": "lexical_detection", "query": "maturity", "prompt": "Process the full audio offline. Find every instance of the word \"maturity\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.58, 4.28, "maturity"]], "response_mode": "offline", "answer": "[{\"word\":\"maturity\",\"start\":3.58,\"end\":4.28}]"}
{"utterance_id": "2035-147960-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 1.83, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.71,\"end\":1.83}]"}
{"utterance_id": "2803-154328-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0022.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.41, 3.94, "flax"]], "response_mode": "offline", "answer": "[{\"word\":\"flax\",\"start\":3.41,\"end\":3.94}]"}
{"utterance_id": "7976-105575-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0023.flac", "task_type": "lexical_detection", "query": "lay", "prompt": "Once the audio has finished, identify every instance of the word \"lay\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.53, 0.79, "lay"]], "response_mode": "offline", "answer": "[{\"word\":\"lay\",\"start\":0.53,\"end\":0.79}]"}
{"utterance_id": "6313-66125-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0017.flac", "task_type": "lexical_detection", "query": "sharp", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sharp\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.19, 2.56, "sharp"]], "response_mode": "offline", "answer": "[{\"word\":\"sharp\",\"start\":2.19,\"end\":2.56}]"}
{"utterance_id": "5694-64025-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0004.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.76, 2.21, "courier"]], "response_mode": "offline", "answer": "[{\"word\":\"courier\",\"start\":1.76,\"end\":2.21}]"}
{"utterance_id": "8297-275154-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0018.flac", "task_type": "lexical_detection", "query": "afraid", "prompt": "After the audio ends, extract all occurrences of the word \"afraid\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.25, "afraid"]], "response_mode": "offline", "answer": "[{\"word\":\"afraid\",\"start\":0.74,\"end\":1.25}]"}
{"utterance_id": "5536-43358-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0017.flac", "task_type": "lexical_detection", "query": "essence", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"essence\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.44, "essence"]], "response_mode": "offline", "answer": "[{\"word\":\"essence\",\"start\":2.08,\"end\":2.44}]"}
{"utterance_id": "7976-105575-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0005.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.18, "battery"]], "response_mode": "offline", "answer": "[{\"word\":\"battery\",\"start\":0.71,\"end\":1.18}]"}
{"utterance_id": "2803-154320-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0006.flac", "task_type": "lexical_detection", "query": "lord", "prompt": "After the audio ends, extract all occurrences of the word \"lord\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.85, "lord"]], "response_mode": "offline", "answer": "[{\"word\":\"lord\",\"start\":1.39,\"end\":1.85}]"}
{"utterance_id": "2412-153948-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0013.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a money or currency term with fields word, start, end (seconds). If none, reply [].", "labels": [[0.95, 1.37, "money"], [5.39, 5.97, "capital"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":0.95,\"end\":1.37},{\"word\":\"capital\",\"start\":5.39,\"end\":5.97}]"}
{"utterance_id": "3000-15664-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0007.flac", "task_type": "lexical_detection", "query": "utter", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"utter\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.34, 4.58, "utter"]], "response_mode": "offline", "answer": "[{\"word\":\"utter\",\"start\":4.34,\"end\":4.58}]"}
{"utterance_id": "2428-83705-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0008.flac", "task_type": "lexical_counting", "query": "gasped", "prompt": "Listen to the full audio. Count every occurrence of the word \"gasped\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.16, "one"], [2.64, 3.24, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.6,\"end\":1.16},{\"word\":\"two\",\"start\":2.64,\"end\":3.24}]"}
{"utterance_id": "1272-141231-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0027.flac", "task_type": "lexical_detection", "query": "sounded", "prompt": "Once the audio has finished, identify every instance of the word \"sounded\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.08, 1.62, "sounded"]], "response_mode": "offline", "answer": "[{\"word\":\"sounded\",\"start\":1.08,\"end\":1.62}]"}
{"utterance_id": "2803-161169-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0012.flac", "task_type": "lexical_counting", "query": "trees", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"trees\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[2.06, 2.66, "one"], [11.56, 12.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.66},{\"word\":\"two\",\"start\":11.56,\"end\":12.3}]"}
{"utterance_id": "5338-24615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[14.36, 14.71, "boughs"]], "response_mode": "offline", "answer": "[{\"word\":\"boughs\",\"start\":14.36,\"end\":14.71}]"}
{"utterance_id": "3853-163249-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0036.flac", "task_type": "lexical_detection", "query": "admire", "prompt": "Process the full audio offline. Find every instance of the word \"admire\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.73, 11.38, "admire"]], "response_mode": "offline", "answer": "[{\"word\":\"admire\",\"start\":10.73,\"end\":11.38}]"}
{"utterance_id": "652-129742-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.55, 0.72, "two"], [2.72, 2.92, "two"], [5.78, 6.06, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.55,\"end\":0.72},{\"word\":\"two\",\"start\":2.72,\"end\":2.92},{\"word\":\"half\",\"start\":5.78,\"end\":6.06}]"}
{"utterance_id": "1462-170138-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.96, 5.36, "philly's"]], "response_mode": "offline", "answer": "[{\"word\":\"philly's\",\"start\":4.96,\"end\":5.36}]"}
{"utterance_id": "1462-170138-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0018.flac", "task_type": "lexical_detection", "query": "girl's", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"girl's\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.98, 2.31, "girl's"]], "response_mode": "offline", "answer": "[{\"word\":\"girl's\",\"start\":1.98,\"end\":2.31}]"}
{"utterance_id": "2428-83699-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0031.flac", "task_type": "lexical_detection", "query": "suggested", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"suggested\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.21, 1.69, "suggested"]], "response_mode": "offline", "answer": "[{\"word\":\"suggested\",\"start\":1.21,\"end\":1.69}]"}
{"utterance_id": "1272-128104-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[17.02, 17.94, "upholsterer"]], "response_mode": "offline", "answer": "[{\"word\":\"upholsterer\",\"start\":17.02,\"end\":17.94}]"}
{"utterance_id": "6313-66125-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0020.flac", "task_type": "lexical_detection", "query": "reach", "prompt": "Process the full audio offline. Find every instance of the word \"reach\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.59, "reach"]], "response_mode": "offline", "answer": "[{\"word\":\"reach\",\"start\":2.37,\"end\":2.59}]"}
{"utterance_id": "6345-93302-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0012.flac", "task_type": "lexical_detection", "query": "shared", "prompt": "After listening to the whole recording, output every instance of the word \"shared\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.01, "shared"]], "response_mode": "offline", "answer": "[{\"word\":\"shared\",\"start\":0.66,\"end\":1.01}]"}
{"utterance_id": "7850-73752-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0019.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.02, "papa"]], "response_mode": "offline", "answer": "[{\"word\":\"papa\",\"start\":0.7,\"end\":1.02}]"}
{"utterance_id": "1919-142785-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0028.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.38, "soup"]], "response_mode": "offline", "answer": "[{\"word\":\"soup\",\"start\":2.04,\"end\":2.38}]"}
{"utterance_id": "6295-244435-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0012.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.36, "cotton"]], "response_mode": "offline", "answer": "[{\"word\":\"cotton\",\"start\":1.78,\"end\":2.36}]"}
{"utterance_id": "777-126732-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.92, 5.33, "fifteen"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen\",\"start\":4.92,\"end\":5.33}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "lexical_counting", "query": "blood", "prompt": "When the utterance is complete, report how the word \"blood\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.88, "one"], [17.03, 17.41, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.62,\"end\":1.88},{\"word\":\"two\",\"start\":17.03,\"end\":17.41}]"}
{"utterance_id": "6319-275224-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0015.flac", "task_type": "lexical_detection", "query": "farm", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"farm\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.44, "farm"]], "response_mode": "offline", "answer": "[{\"word\":\"farm\",\"start\":4.13,\"end\":4.44}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.17, 3.53, "esther"]], "response_mode": "offline", "answer": "[{\"word\":\"esther\",\"start\":3.17,\"end\":3.53}]"}
{"utterance_id": "8297-275155-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0015.flac", "task_type": "lexical_detection", "query": "mercy", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mercy\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.41, 5.96, "mercy"]], "response_mode": "offline", "answer": "[{\"word\":\"mercy\",\"start\":5.41,\"end\":5.96}]"}
{"utterance_id": "2428-83705-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0026.flac", "task_type": "lexical_counting", "query": "plain", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"plain\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[4.25, 4.71, "one"], [5.84, 6.15, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.25,\"end\":4.71},{\"word\":\"two\",\"start\":5.84,\"end\":6.15}]"}
{"utterance_id": "2086-149220-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0017.flac", "task_type": "lexical_detection", "query": "vulgar", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"vulgar\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.57, 6.94, "vulgar"]], "response_mode": "offline", "answer": "[{\"word\":\"vulgar\",\"start\":6.57,\"end\":6.94}]"}
{"utterance_id": "2035-147960-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0016.flac", "task_type": "lexical_detection", "query": "size", "prompt": "Once the audio has finished, identify every instance of the word \"size\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.14, 1.61, "size"]], "response_mode": "offline", "answer": "[{\"word\":\"size\",\"start\":1.14,\"end\":1.61}]"}
{"utterance_id": "2035-147960-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0013.flac", "task_type": "lexical_detection", "query": "fuchs", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fuchs\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.74, 1.14, "fuchs"]], "response_mode": "offline", "answer": "[{\"word\":\"fuchs\",\"start\":0.74,\"end\":1.14}]"}
{"utterance_id": "5895-34615-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0020.flac", "task_type": "lexical_detection", "query": "brow", "prompt": "After listening to the whole recording, output every instance of the word \"brow\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.11, 6.61, "brow"]], "response_mode": "offline", "answer": "[{\"word\":\"brow\",\"start\":6.11,\"end\":6.61}]"}
{"utterance_id": "3853-163249-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0043.flac", "task_type": "lexical_detection", "query": "part", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"part\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.9, 1.18, "part"]], "response_mode": "offline", "answer": "[{\"word\":\"part\",\"start\":0.9,\"end\":1.18}]"}
{"utterance_id": "2428-83705-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0014.flac", "task_type": "lexical_detection", "query": "pigstye", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pigstye\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.53, 3.1, "pigstye"]], "response_mode": "offline", "answer": "[{\"word\":\"pigstye\",\"start\":2.53,\"end\":3.1}]"}
{"utterance_id": "3853-163249-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0031.flac", "task_type": "lexical_detection", "query": "swallowed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"swallowed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.11, 2.53, "swallowed"]], "response_mode": "offline", "answer": "[{\"word\":\"swallowed\",\"start\":2.11,\"end\":2.53}]"}
{"utterance_id": "6313-66129-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0029.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.69, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.41,\"end\":0.69}]"}
{"utterance_id": "2035-152373-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0015.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.97, 2.46, "ferns"]], "response_mode": "offline", "answer": "[{\"word\":\"ferns\",\"start\":1.97,\"end\":2.46}]"}
{"utterance_id": "251-136532-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0022.flac", "task_type": "lexical_detection", "query": "lectures", "prompt": "After the audio ends, extract all occurrences of the word \"lectures\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.32, "lectures"]], "response_mode": "offline", "answer": "[{\"word\":\"lectures\",\"start\":0.51,\"end\":1.32}]"}
{"utterance_id": "5694-64029-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0003.flac", "task_type": "lexical_detection", "query": "picket", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"picket\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.07, 3.43, "picket"]], "response_mode": "offline", "answer": "[{\"word\":\"picket\",\"start\":3.07,\"end\":3.43}]"}
{"utterance_id": "6345-93306-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0014.flac", "task_type": "lexical_detection", "query": "kind", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"kind\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.34, 1.92, "kind"]], "response_mode": "offline", "answer": "[{\"word\":\"kind\",\"start\":1.34,\"end\":1.92}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "semantic_counting", "query": "money", "prompt": "After listening to the whole recording, output timed count words for a money or currency term (one, two, three, ...) as JSON. If none, reply [].", "labels": [[6.25, 6.57, "one"], [9.68, 9.93, "two"], [12.74, 13.59, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.25,\"end\":6.57},{\"word\":\"two\",\"start\":9.68,\"end\":9.93},{\"word\":\"three\",\"start\":12.74,\"end\":13.59}]"}
{"utterance_id": "1993-147149-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.08, 7.61, "hand's"], [8.81, 9.24, "frame"], [12.65, 13.14, "ear"]], "response_mode": "offline", "answer": "[{\"word\":\"hand's\",\"start\":7.08,\"end\":7.61},{\"word\":\"frame\",\"start\":8.81,\"end\":9.24},{\"word\":\"ear\",\"start\":12.65,\"end\":13.14}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "You will hear a complete utterance. When it ends, list every instance of an organization name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.78, 8.4, "del monte"]], "response_mode": "offline", "answer": "[{\"word\":\"del monte\",\"start\":7.78,\"end\":8.4}]"}
{"utterance_id": "2803-154328-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0010.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "After the audio ends, tally every instance of an ordinal number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.72, 1.26, "one"], [5.15, 5.76, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.72,\"end\":1.26},{\"word\":\"two\",\"start\":5.15,\"end\":5.76}]"}
{"utterance_id": "2902-9008-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.06, 4.77, "hypatia"]], "response_mode": "offline", "answer": "[{\"word\":\"hypatia\",\"start\":4.06,\"end\":4.77}]"}
{"utterance_id": "5694-64038-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0016.flac", "task_type": "lexical_detection", "query": "supper", "prompt": "When the utterance is complete, report all instances of the word \"supper\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.58, "supper"]], "response_mode": "offline", "answer": "[{\"word\":\"supper\",\"start\":1.25,\"end\":1.58}]"}
{"utterance_id": "2428-83699-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0012.flac", "task_type": "lexical_detection", "query": "absolutely", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"absolutely\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.44, 7.0, "absolutely"]], "response_mode": "offline", "answer": "[{\"word\":\"absolutely\",\"start\":6.44,\"end\":7.0}]"}
{"utterance_id": "1673-143396-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0016.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 2.16, "one"], [3.42, 3.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.26,\"end\":2.16},{\"word\":\"two\",\"start\":3.42,\"end\":3.97}]"}
{"utterance_id": "1993-147964-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0007.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "You will hear a complete utterance. When it ends, list every instance of a state or province name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.55, "virginia"]], "response_mode": "offline", "answer": "[{\"word\":\"virginia\",\"start\":2.95,\"end\":3.55}]"}
{"utterance_id": "7976-105575-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0005.flac", "task_type": "lexical_detection", "query": "war", "prompt": "Once the audio has finished, identify every instance of the word \"war\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.45, 2.85, "war"]], "response_mode": "offline", "answer": "[{\"word\":\"war\",\"start\":2.45,\"end\":2.85}]"}
{"utterance_id": "6319-57405-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0009.flac", "task_type": "lexical_detection", "query": "bones", "prompt": "When the utterance is complete, report all instances of the word \"bones\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.07, 6.55, "bones"]], "response_mode": "offline", "answer": "[{\"word\":\"bones\",\"start\":6.07,\"end\":6.55}]"}
{"utterance_id": "3081-166546-0064", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0064.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 1.16, "sweetwater"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":0.35,\"end\":1.16}]"}
{"utterance_id": "8297-275155-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.58, "missus norman"]], "response_mode": "offline", "answer": "[{\"word\":\"missus norman\",\"start\":0.5,\"end\":1.58}]"}
{"utterance_id": "6241-61943-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0019.flac", "task_type": "lexical_detection", "query": "lose", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lose\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 2.74, "lose"]], "response_mode": "offline", "answer": "[{\"word\":\"lose\",\"start\":2.29,\"end\":2.74}]"}
{"utterance_id": "3536-23268-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0001.flac", "task_type": "lexical_detection", "query": "passion", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"passion\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.42, 3.07, "passion"]], "response_mode": "offline", "answer": "[{\"word\":\"passion\",\"start\":2.42,\"end\":3.07}]"}
{"utterance_id": "6295-244435-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0025.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.09, 1.83, "charleston"]], "response_mode": "offline", "answer": "[{\"word\":\"charleston\",\"start\":1.09,\"end\":1.83}]"}
{"utterance_id": "3536-8226-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0024.flac", "task_type": "lexical_detection", "query": "right", "prompt": "When the utterance is complete, report all instances of the word \"right\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.01, 3.42, "right"]], "response_mode": "offline", "answer": "[{\"word\":\"right\",\"start\":3.01,\"end\":3.42}]"}
{"utterance_id": "5536-43358-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.32, 17.58, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":17.32,\"end\":17.58}]"}
{"utterance_id": "5338-284437-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0029.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.44, "frog"]], "response_mode": "offline", "answer": "[{\"word\":\"frog\",\"start\":4.99,\"end\":5.44}]"}
{"utterance_id": "8297-275156-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0002.flac", "task_type": "lexical_detection", "query": "clear", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"clear\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.42, 4.71, "clear"]], "response_mode": "offline", "answer": "[{\"word\":\"clear\",\"start\":4.42,\"end\":4.71}]"}
{"utterance_id": "84-121123-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0015.flac", "task_type": "lexical_detection", "query": "understand", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"understand\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.84, 1.33, "understand"]], "response_mode": "offline", "answer": "[{\"word\":\"understand\",\"start\":0.84,\"end\":1.33}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.82, 4.46, "three"], [12.85, 13.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":3.82,\"end\":4.46},{\"word\":\"two\",\"start\":12.85,\"end\":13.32}]"}
{"utterance_id": "5895-34629-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.13, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":1.84,\"end\":2.13}]"}
{"utterance_id": "1993-147149-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0022.flac", "task_type": "lexical_detection", "query": "wilson", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wilson\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.76, 3.23, "wilson"]], "response_mode": "offline", "answer": "[{\"word\":\"wilson\",\"start\":2.76,\"end\":3.23}]"}
{"utterance_id": "2277-149874-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.38, 5.63, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":5.38,\"end\":5.63}]"}
{"utterance_id": "422-122949-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0027.flac", "task_type": "lexical_counting", "query": "action", "prompt": "Listen to the full audio. Count every occurrence of the word \"action\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.04, 6.46, "one"], [8.33, 8.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.04,\"end\":6.46},{\"word\":\"two\",\"start\":8.33,\"end\":8.79}]"}
{"utterance_id": "1988-147956-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0003.flac", "task_type": "lexical_detection", "query": "otto", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"otto\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.15, "otto"]], "response_mode": "offline", "answer": "[{\"word\":\"otto\",\"start\":1.74,\"end\":2.15}]"}
{"utterance_id": "2428-83699-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0033.flac", "task_type": "lexical_detection", "query": "bell", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bell\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 0.83, "bell"]], "response_mode": "offline", "answer": "[{\"word\":\"bell\",\"start\":0.43,\"end\":0.83}]"}
{"utterance_id": "251-136532-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0013.flac", "task_type": "lexical_detection", "query": "corners", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"corners\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.28, "corners"]], "response_mode": "offline", "answer": "[{\"word\":\"corners\",\"start\":3.7,\"end\":4.28}]"}
{"utterance_id": "777-126732-0078", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0078.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Once the audio has finished, identify every instance of a money or currency term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.86, 1.39, "halfpenny"]], "response_mode": "offline", "answer": "[{\"word\":\"halfpenny\",\"start\":0.86,\"end\":1.39}]"}
{"utterance_id": "5694-64038-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0001.flac", "task_type": "lexical_detection", "query": "yank", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"yank\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 0.86, "yank"]], "response_mode": "offline", "answer": "[{\"word\":\"yank\",\"start\":0.53,\"end\":0.86}]"}
{"utterance_id": "174-168635-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0014.flac", "task_type": "lexical_detection", "query": "longer", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"longer\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.95, "longer"]], "response_mode": "offline", "answer": "[{\"word\":\"longer\",\"start\":1.46,\"end\":1.95}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a color name (word, start, end). If none, reply [].", "labels": [[2.2, 2.69, "one"], [14.18, 14.57, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.2,\"end\":2.69},{\"word\":\"two\",\"start\":14.18,\"end\":14.57}]"}
{"utterance_id": "2078-142845-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.29, "kirkleatham"]], "response_mode": "offline", "answer": "[{\"word\":\"kirkleatham\",\"start\":0.51,\"end\":1.29}]"}
{"utterance_id": "1462-170142-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0028.flac", "task_type": "lexical_detection", "query": "stole", "prompt": "After the audio ends, extract all occurrences of the word \"stole\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.38, "stole"]], "response_mode": "offline", "answer": "[{\"word\":\"stole\",\"start\":1.1,\"end\":1.38}]"}
{"utterance_id": "7850-281318-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0004.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.81, 6.11, "bird"]], "response_mode": "offline", "answer": "[{\"word\":\"bird\",\"start\":5.81,\"end\":6.11}]"}
{"utterance_id": "3752-4944-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.73, 3.14, "meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"meekin\",\"start\":2.73,\"end\":3.14}]"}
{"utterance_id": "1272-141231-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.34, 3.0, "twenties"]], "response_mode": "offline", "answer": "[{\"word\":\"twenties\",\"start\":2.34,\"end\":3.0}]"}
{"utterance_id": "174-84280-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0000.flac", "task_type": "lexical_detection", "query": "how", "prompt": "Process the full audio offline. Find every instance of the word \"how\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.81, "how"]], "response_mode": "offline", "answer": "[{\"word\":\"how\",\"start\":0.54,\"end\":0.81}]"}
{"utterance_id": "2086-149220-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0035.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.16, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":1.75,\"end\":2.16}]"}
{"utterance_id": "1988-24833-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0024.flac", "task_type": "lexical_detection", "query": "job", "prompt": "Once the audio has finished, identify every instance of the word \"job\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.37, 5.68, "job"]], "response_mode": "offline", "answer": "[{\"word\":\"job\",\"start\":5.37,\"end\":5.68}]"}
{"utterance_id": "8297-275155-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0002.flac", "task_type": "lexical_detection", "query": "columns", "prompt": "Once the audio has finished, identify every instance of the word \"columns\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.77, 4.4, "columns"]], "response_mode": "offline", "answer": "[{\"word\":\"columns\",\"start\":3.77,\"end\":4.4}]"}
{"utterance_id": "6295-64301-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0029.flac", "task_type": "lexical_detection", "query": "tearing", "prompt": "Process the full audio offline. Find every instance of the word \"tearing\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.14, 5.72, "tearing"]], "response_mode": "offline", "answer": "[{\"word\":\"tearing\",\"start\":5.14,\"end\":5.72}]"}
{"utterance_id": "2277-149896-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0017.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "You will hear a complete utterance. When it ends, list every instance of a time of day you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.15, "one thirty"]], "response_mode": "offline", "answer": "[{\"word\":\"one thirty\",\"start\":0.67,\"end\":1.15}]"}
{"utterance_id": "3536-8226-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0005.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "When the utterance is complete, report how a profession accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.21, "one"], [11.31, 11.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.9,\"end\":2.21},{\"word\":\"two\",\"start\":11.31,\"end\":11.85}]"}
{"utterance_id": "84-121550-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0016.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.8, "rome"]], "response_mode": "offline", "answer": "[{\"word\":\"rome\",\"start\":1.29,\"end\":1.8}]"}
{"utterance_id": "3853-163249-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0014.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[4.52, 4.98, "one"], [6.18, 6.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.52,\"end\":4.98},{\"word\":\"two\",\"start\":6.18,\"end\":6.54}]"}
{"utterance_id": "251-136532-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0011.flac", "task_type": "lexical_detection", "query": "burned", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"burned\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.38, "burned"]], "response_mode": "offline", "answer": "[{\"word\":\"burned\",\"start\":1.02,\"end\":1.38}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.69, "greenest"], [14.18, 14.57, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"greenest\",\"start\":2.2,\"end\":2.69},{\"word\":\"yellow\",\"start\":14.18,\"end\":14.57}]"}
{"utterance_id": "2277-149897-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0024.flac", "task_type": "lexical_detection", "query": "family", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"family\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.02, 1.33, "family"]], "response_mode": "offline", "answer": "[{\"word\":\"family\",\"start\":1.02,\"end\":1.33}]"}
{"utterance_id": "2412-153947-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0015.flac", "task_type": "lexical_detection", "query": "erewhon", "prompt": "When the utterance is complete, report all instances of the word \"erewhon\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 3.0, "erewhon"]], "response_mode": "offline", "answer": "[{\"word\":\"erewhon\",\"start\":2.55,\"end\":3.0}]"}
{"utterance_id": "1988-147956-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[1.68, 2.02, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.68,\"end\":2.02}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.36, "eldest"], [2.97, 3.44, "youngest"]], "response_mode": "offline", "answer": "[{\"word\":\"eldest\",\"start\":0.91,\"end\":1.36},{\"word\":\"youngest\",\"start\":2.97,\"end\":3.44}]"}
{"utterance_id": "8297-275155-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0026.flac", "task_type": "lexical_detection", "query": "decision", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"decision\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.57, 2.28, "decision"]], "response_mode": "offline", "answer": "[{\"word\":\"decision\",\"start\":1.57,\"end\":2.28}]"}
{"utterance_id": "6241-66616-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0009.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.57, 4.89, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":4.57,\"end\":4.89}]"}
{"utterance_id": "1993-147149-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0017.flac", "task_type": "lexical_detection", "query": "bear", "prompt": "When the utterance is complete, report all instances of the word \"bear\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.51, 2.69, "bear"]], "response_mode": "offline", "answer": "[{\"word\":\"bear\",\"start\":2.51,\"end\":2.69}]"}
{"utterance_id": "1919-142785-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0019.flac", "task_type": "lexical_detection", "query": "cultivated", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"cultivated\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.38, 2.0, "cultivated"]], "response_mode": "offline", "answer": "[{\"word\":\"cultivated\",\"start\":1.38,\"end\":2.0}]"}
{"utterance_id": "2428-83705-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0032.flac", "task_type": "lexical_detection", "query": "fail", "prompt": "After listening to the whole recording, output every instance of the word \"fail\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.69, 5.19, "fail"]], "response_mode": "offline", "answer": "[{\"word\":\"fail\",\"start\":4.69,\"end\":5.19}]"}
{"utterance_id": "652-130726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 1.45, "san francisco's"]], "response_mode": "offline", "answer": "[{\"word\":\"san francisco's\",\"start\":0.47,\"end\":1.45}]"}
{"utterance_id": "1462-170142-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0015.flac", "task_type": "lexical_detection", "query": "tortures", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"tortures\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.4, 1.83, "tortures"]], "response_mode": "offline", "answer": "[{\"word\":\"tortures\",\"start\":1.4,\"end\":1.83}]"}
{"utterance_id": "2803-154328-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0003.flac", "task_type": "lexical_detection", "query": "slept", "prompt": "Process the full audio offline. Find every instance of the word \"slept\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.08, 7.53, "slept"]], "response_mode": "offline", "answer": "[{\"word\":\"slept\",\"start\":7.08,\"end\":7.53}]"}
{"utterance_id": "7976-110523-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.24, 5.6, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":5.24,\"end\":5.6}]"}
{"utterance_id": "251-137823-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0018.flac", "task_type": "lexical_detection", "query": "anyone", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"anyone\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.7, 5.08, "anyone"]], "response_mode": "offline", "answer": "[{\"word\":\"anyone\",\"start\":4.7,\"end\":5.08}]"}
{"utterance_id": "1462-170145-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0018.flac", "task_type": "lexical_detection", "query": "completely", "prompt": "Once the audio has finished, identify every instance of the word \"completely\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.15, 1.84, "completely"]], "response_mode": "offline", "answer": "[{\"word\":\"completely\",\"start\":1.15,\"end\":1.84}]"}
{"utterance_id": "1993-147149-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0023.flac", "task_type": "lexical_detection", "query": "plainly", "prompt": "Once the audio has finished, identify every instance of the word \"plainly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[9.43, 10.17, "plainly"]], "response_mode": "offline", "answer": "[{\"word\":\"plainly\",\"start\":9.43,\"end\":10.17}]"}
{"utterance_id": "5895-34629-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0019.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.57, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":1.0,\"end\":1.57}]"}
{"utterance_id": "2277-149897-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0030.flac", "task_type": "lexical_detection", "query": "within", "prompt": "After the audio ends, extract all occurrences of the word \"within\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.23, "within"]], "response_mode": "offline", "answer": "[{\"word\":\"within\",\"start\":0.95,\"end\":1.23}]"}
{"utterance_id": "1988-24833-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0021.flac", "task_type": "lexical_detection", "query": "hilda", "prompt": "Once the audio has finished, identify every instance of the word \"hilda\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.95, 1.2, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":0.95,\"end\":1.2}]"}
{"utterance_id": "5895-34615-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0016.flac", "task_type": "lexical_detection", "query": "suffering", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"suffering\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.05, 8.67, "suffering"]], "response_mode": "offline", "answer": "[{\"word\":\"suffering\",\"start\":8.05,\"end\":8.67}]"}
{"utterance_id": "2428-83699-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0015.flac", "task_type": "lexical_detection", "query": "reason", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"reason\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.58, 2.1, "reason"]], "response_mode": "offline", "answer": "[{\"word\":\"reason\",\"start\":1.58,\"end\":2.1}]"}
{"utterance_id": "3853-163249-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0045.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.2, "one"], [6.2, 6.51, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.86,\"end\":4.2},{\"word\":\"two\",\"start\":6.2,\"end\":6.51}]"}
{"utterance_id": "1993-147966-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0002.flac", "task_type": "lexical_detection", "query": "people's", "prompt": "After listening to the whole recording, output every instance of the word \"people's\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.2, "people's"]], "response_mode": "offline", "answer": "[{\"word\":\"people's\",\"start\":2.74,\"end\":3.2}]"}
{"utterance_id": "7850-281318-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0006.flac", "task_type": "lexical_detection", "query": "little", "prompt": "Process the full audio offline. Find every instance of the word \"little\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.48, 6.85, "little"]], "response_mode": "offline", "answer": "[{\"word\":\"little\",\"start\":6.48,\"end\":6.85}]"}
{"utterance_id": "777-126732-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0059.flac", "task_type": "lexical_detection", "query": "midst", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"midst\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.56, "midst"]], "response_mode": "offline", "answer": "[{\"word\":\"midst\",\"start\":3.26,\"end\":3.56}]"}
{"utterance_id": "6313-66125-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0022.flac", "task_type": "lexical_detection", "query": "tad's", "prompt": "Process the full audio offline. Find every instance of the word \"tad's\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.78, 7.14, "tad's"]], "response_mode": "offline", "answer": "[{\"word\":\"tad's\",\"start\":6.78,\"end\":7.14}]"}
{"utterance_id": "7850-281318-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0010.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, produce a running count of an animal name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.55, 2.23, "one"], [3.59, 4.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.55,\"end\":2.23},{\"word\":\"two\",\"start\":3.59,\"end\":4.32}]"}
{"utterance_id": "5338-284437-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0001.flac", "task_type": "lexical_detection", "query": "tourmaline", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"tourmaline\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.22, 7.84, "tourmaline"]], "response_mode": "offline", "answer": "[{\"word\":\"tourmaline\",\"start\":7.22,\"end\":7.84}]"}
{"utterance_id": "1993-147964-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0008.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.83, 6.66, "christmas eve"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas eve\",\"start\":5.83,\"end\":6.66}]"}
{"utterance_id": "2086-149220-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.15, "miss hepzibah"], [6.6, 7.07, "pyncheon"]], "response_mode": "offline", "answer": "[{\"word\":\"miss hepzibah\",\"start\":0.51,\"end\":1.15},{\"word\":\"pyncheon\",\"start\":6.6,\"end\":7.07}]"}
{"utterance_id": "7976-110124-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0010.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.35, "sisters"]], "response_mode": "offline", "answer": "[{\"word\":\"sisters\",\"start\":3.81,\"end\":4.35}]"}
{"utterance_id": "8842-302203-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.57, 1.99, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.57,\"end\":1.99}]"}
{"utterance_id": "3853-163249-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0056.flac", "task_type": "lexical_detection", "query": "honor", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"honor\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.17, 3.49, "honor"]], "response_mode": "offline", "answer": "[{\"word\":\"honor\",\"start\":3.17,\"end\":3.49}]"}
{"utterance_id": "1988-24833-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0020.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.54, 5.07, "grandfather"]], "response_mode": "offline", "answer": "[{\"word\":\"grandfather\",\"start\":4.54,\"end\":5.07}]"}
{"utterance_id": "2428-83699-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0034.flac", "task_type": "lexical_detection", "query": "window", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"window\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.8, 7.2, "window"]], "response_mode": "offline", "answer": "[{\"word\":\"window\",\"start\":6.8,\"end\":7.2}]"}
{"utterance_id": "2086-149220-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0016.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.15, "one"], [6.6, 7.07, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.15},{\"word\":\"two\",\"start\":6.6,\"end\":7.07}]"}
{"utterance_id": "3081-166546-0070", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0070.flac", "task_type": "lexical_detection", "query": "questions", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"questions\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.62, 11.39, "questions"]], "response_mode": "offline", "answer": "[{\"word\":\"questions\",\"start\":10.62,\"end\":11.39}]"}
{"utterance_id": "3170-137482-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0005.flac", "task_type": "lexical_detection", "query": "spend", "prompt": "Process the full audio offline. Find every instance of the word \"spend\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.54, "spend"]], "response_mode": "offline", "answer": "[{\"word\":\"spend\",\"start\":1.27,\"end\":1.54}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "lexical_detection", "query": "turban", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"turban\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.35, 5.75, "turban"]], "response_mode": "offline", "answer": "[{\"word\":\"turban\",\"start\":5.35,\"end\":5.75}]"}
{"utterance_id": "3853-163249-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0016.flac", "task_type": "lexical_detection", "query": "younger", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"younger\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.52, 6.97, "younger"]], "response_mode": "offline", "answer": "[{\"word\":\"younger\",\"start\":6.52,\"end\":6.97}]"}
{"utterance_id": "652-130737-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.51, 1.15, "sauterne"], [4.26, 4.65, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"sauterne\",\"start\":0.51,\"end\":1.15},{\"word\":\"wine\",\"start\":4.26,\"end\":4.65}]"}
{"utterance_id": "5895-34629-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0026.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the full audio. After it finishes, find every instance of a weather term. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.03, 11.74, "clouds"]], "response_mode": "offline", "answer": "[{\"word\":\"clouds\",\"start\":11.03,\"end\":11.74}]"}
{"utterance_id": "1462-170145-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0012.flac", "task_type": "lexical_detection", "query": "bent", "prompt": "Once the audio has finished, identify every instance of the word \"bent\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.47, 0.74, "bent"]], "response_mode": "offline", "answer": "[{\"word\":\"bent\",\"start\":0.47,\"end\":0.74}]"}
{"utterance_id": "174-50561-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.61, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":1.11,\"end\":1.61}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[10.95, 11.55, "emperor"]], "response_mode": "offline", "answer": "[{\"word\":\"emperor\",\"start\":10.95,\"end\":11.55}]"}
{"utterance_id": "2803-154328-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0004.flac", "task_type": "lexical_counting", "query": "die", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"die\" with word/start/end. If none, reply [].", "labels": [[1.8, 2.13, "one"], [3.55, 3.77, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.8,\"end\":2.13},{\"word\":\"two\",\"start\":3.55,\"end\":3.77}]"}
{"utterance_id": "1272-141231-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0018.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.64, 0.85, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":0.64,\"end\":0.85}]"}
{"utterance_id": "5338-284437-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0015.flac", "task_type": "lexical_detection", "query": "trying", "prompt": "After listening to the whole recording, output every instance of the word \"trying\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.92, 1.23, "trying"]], "response_mode": "offline", "answer": "[{\"word\":\"trying\",\"start\":0.92,\"end\":1.23}]"}
{"utterance_id": "1919-142785-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0044.flac", "task_type": "lexical_detection", "query": "mix", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"mix\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 3.94, "mix"]], "response_mode": "offline", "answer": "[{\"word\":\"mix\",\"start\":3.7,\"end\":3.94}]"}
{"utterance_id": "1988-147956-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.58, 2.92, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":2.58,\"end\":2.92}]"}
{"utterance_id": "3000-15664-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.42, 16.03, "six hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"six hundred\",\"start\":15.42,\"end\":16.03}]"}
{"utterance_id": "777-126732-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0050.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.87, 5.2, "stevie"]], "response_mode": "offline", "answer": "[{\"word\":\"stevie\",\"start\":4.87,\"end\":5.2}]"}
{"utterance_id": "7976-105575-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0024.flac", "task_type": "lexical_detection", "query": "tree", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"tree\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.76, 3.01, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":2.76,\"end\":3.01}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.32, 0.79, "roses"], [5.3, 5.78, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"roses\",\"start\":0.32,\"end\":0.79},{\"word\":\"flowers\",\"start\":5.3,\"end\":5.78}]"}
{"utterance_id": "174-168635-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0002.flac", "task_type": "lexical_detection", "query": "sister's", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sister's\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.08, "sister's"]], "response_mode": "offline", "answer": "[{\"word\":\"sister's\",\"start\":1.64,\"end\":2.08}]"}
{"utterance_id": "84-121123-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0009.flac", "task_type": "lexical_detection", "query": "morrel", "prompt": "Once the audio has finished, identify every instance of the word \"morrel\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.01, 2.43, "morrel"]], "response_mode": "offline", "answer": "[{\"word\":\"morrel\",\"start\":2.01,\"end\":2.43}]"}
{"utterance_id": "251-136532-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0001.flac", "task_type": "lexical_detection", "query": "syrtis", "prompt": "When the utterance is complete, report all instances of the word \"syrtis\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.76, 6.45, "syrtis"]], "response_mode": "offline", "answer": "[{\"word\":\"syrtis\",\"start\":5.76,\"end\":6.45}]"}
{"utterance_id": "422-122949-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0024.flac", "task_type": "lexical_detection", "query": "wherein", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wherein\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.49, 5.94, "wherein"]], "response_mode": "offline", "answer": "[{\"word\":\"wherein\",\"start\":5.49,\"end\":5.94}]"}
{"utterance_id": "5895-34629-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0015.flac", "task_type": "lexical_detection", "query": "small", "prompt": "After listening to the whole recording, output every instance of the word \"small\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.34, "small"]], "response_mode": "offline", "answer": "[{\"word\":\"small\",\"start\":1.03,\"end\":1.34}]"}
{"utterance_id": "5536-43359-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0012.flac", "task_type": "lexical_detection", "query": "notable", "prompt": "When the utterance is complete, report all instances of the word \"notable\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.56, 3.07, "notable"]], "response_mode": "offline", "answer": "[{\"word\":\"notable\",\"start\":2.56,\"end\":3.07}]"}
{"utterance_id": "3536-23268-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.42, 6.13, "missus horton"]], "response_mode": "offline", "answer": "[{\"word\":\"missus horton\",\"start\":5.42,\"end\":6.13}]"}
{"utterance_id": "2086-149220-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.58, 2.99, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":2.58,\"end\":2.99}]"}
{"utterance_id": "2035-147961-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0007.flac", "task_type": "lexical_detection", "query": "smell", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"smell\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.01, "smell"]], "response_mode": "offline", "answer": "[{\"word\":\"smell\",\"start\":0.72,\"end\":1.01}]"}
{"utterance_id": "2428-83705-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0028.flac", "task_type": "lexical_detection", "query": "wishes", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"wishes\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.79, 4.38, "wishes"]], "response_mode": "offline", "answer": "[{\"word\":\"wishes\",\"start\":3.79,\"end\":4.38}]"}
{"utterance_id": "1993-147149-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0006.flac", "task_type": "lexical_detection", "query": "rocking", "prompt": "Process the full audio offline. Find every instance of the word \"rocking\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.37, 4.78, "rocking"]], "response_mode": "offline", "answer": "[{\"word\":\"rocking\",\"start\":4.37,\"end\":4.78}]"}
{"utterance_id": "2078-142845-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0007.flac", "task_type": "lexical_detection", "query": "dresser", "prompt": "Process the full audio offline. Find every instance of the word \"dresser\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.0, 3.84, "dresser"]], "response_mode": "offline", "answer": "[{\"word\":\"dresser\",\"start\":3.0,\"end\":3.84}]"}
{"utterance_id": "2277-149896-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0034.flac", "task_type": "lexical_detection", "query": "harder", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"harder\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.15, "harder"]], "response_mode": "offline", "answer": "[{\"word\":\"harder\",\"start\":1.64,\"end\":2.15}]"}
{"utterance_id": "3576-138058-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0004.flac", "task_type": "lexical_detection", "query": "regardless", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"regardless\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.67, 4.44, "regardless"]], "response_mode": "offline", "answer": "[{\"word\":\"regardless\",\"start\":3.67,\"end\":4.44}]"}
{"utterance_id": "1462-170142-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0014.flac", "task_type": "lexical_detection", "query": "worse", "prompt": "When the utterance is complete, report all instances of the word \"worse\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.06, "worse"]], "response_mode": "offline", "answer": "[{\"word\":\"worse\",\"start\":1.7,\"end\":2.06}]"}
{"utterance_id": "7976-110523-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.57, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.18,\"end\":2.57}]"}
{"utterance_id": "84-121550-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.31, "venus"]], "response_mode": "offline", "answer": "[{\"word\":\"venus\",\"start\":3.78,\"end\":4.31}]"}
{"utterance_id": "6319-64726-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0013.flac", "task_type": "lexical_detection", "query": "apartment", "prompt": "Once the audio has finished, identify every instance of the word \"apartment\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.83, 2.45, "apartment"]], "response_mode": "offline", "answer": "[{\"word\":\"apartment\",\"start\":1.83,\"end\":2.45}]"}
{"utterance_id": "1919-142785-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0048.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.7, "forcemeat"]], "response_mode": "offline", "answer": "[{\"word\":\"forcemeat\",\"start\":0.99,\"end\":1.7}]"}
{"utterance_id": "1988-147956-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0017.flac", "task_type": "lexical_detection", "query": "virginia", "prompt": "Once the audio has finished, identify every instance of the word \"virginia\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.25, 5.8, "virginia"]], "response_mode": "offline", "answer": "[{\"word\":\"virginia\",\"start\":5.25,\"end\":5.8}]"}
{"utterance_id": "84-121550-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0013.flac", "task_type": "lexical_detection", "query": "whence", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"whence\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.44, 15.74, "whence"]], "response_mode": "offline", "answer": "[{\"word\":\"whence\",\"start\":15.44,\"end\":15.74}]"}
{"utterance_id": "251-137823-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.99, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom\",\"start\":0.5,\"end\":0.99}]"}
{"utterance_id": "6241-61946-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0005.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.18, 2.65, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":2.18,\"end\":2.65}]"}
{"utterance_id": "6295-64301-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0016.flac", "task_type": "lexical_detection", "query": "compelled", "prompt": "Once the audio has finished, identify every instance of the word \"compelled\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.43, 2.96, "compelled"]], "response_mode": "offline", "answer": "[{\"word\":\"compelled\",\"start\":2.43,\"end\":2.96}]"}
{"utterance_id": "5694-64029-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0020.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.91, 2.35, "yankees"]], "response_mode": "offline", "answer": "[{\"word\":\"yankees\",\"start\":1.91,\"end\":2.35}]"}
{"utterance_id": "3752-4944-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.49, "mister meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"mister meekin\",\"start\":1.78,\"end\":2.49}]"}
{"utterance_id": "1919-142785-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0028.flac", "task_type": "lexical_detection", "query": "soup", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"soup\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.38, "soup"]], "response_mode": "offline", "answer": "[{\"word\":\"soup\",\"start\":2.04,\"end\":2.38}]"}
{"utterance_id": "5536-43358-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0013.flac", "task_type": "lexical_detection", "query": "quickening", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"quickening\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.24, 3.67, "quickening"]], "response_mode": "offline", "answer": "[{\"word\":\"quickening\",\"start\":3.24,\"end\":3.67}]"}
{"utterance_id": "174-168635-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0018.flac", "task_type": "lexical_detection", "query": "eclipse", "prompt": "After the audio ends, extract all occurrences of the word \"eclipse\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[16.28, 16.96, "eclipse"]], "response_mode": "offline", "answer": "[{\"word\":\"eclipse\",\"start\":16.28,\"end\":16.96}]"}
{"utterance_id": "1988-24833-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0008.flac", "task_type": "lexical_detection", "query": "lucky", "prompt": "Once the audio has finished, identify every instance of the word \"lucky\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.47, 0.82, "lucky"]], "response_mode": "offline", "answer": "[{\"word\":\"lucky\",\"start\":0.47,\"end\":0.82}]"}
{"utterance_id": "1462-170142-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.85, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":0.51,\"end\":0.85}]"}
{"utterance_id": "2035-152373-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.62, "missionary"]], "response_mode": "offline", "answer": "[{\"word\":\"missionary\",\"start\":1.0,\"end\":1.62}]"}
{"utterance_id": "2428-83699-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0025.flac", "task_type": "lexical_detection", "query": "quest", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"quest\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.96, 7.36, "quest"]], "response_mode": "offline", "answer": "[{\"word\":\"quest\",\"start\":6.96,\"end\":7.36}]"}
{"utterance_id": "6241-66616-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0009.flac", "task_type": "lexical_detection", "query": "crown", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"crown\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.1, 4.38, "crown"]], "response_mode": "offline", "answer": "[{\"word\":\"crown\",\"start\":4.1,\"end\":4.38}]"}
{"utterance_id": "2428-83705-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0038.flac", "task_type": "lexical_detection", "query": "unless", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"unless\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.07, 1.46, "unless"]], "response_mode": "offline", "answer": "[{\"word\":\"unless\",\"start\":1.07,\"end\":1.46}]"}
{"utterance_id": "3853-163249-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0020.flac", "task_type": "lexical_detection", "query": "place", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"place\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.28, 3.6, "place"]], "response_mode": "offline", "answer": "[{\"word\":\"place\",\"start\":3.28,\"end\":3.6}]"}
{"utterance_id": "1988-24833-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.73, 4.03, "kate's"]], "response_mode": "offline", "answer": "[{\"word\":\"kate's\",\"start\":3.73,\"end\":4.03}]"}
{"utterance_id": "1993-147965-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0007.flac", "task_type": "lexical_detection", "query": "sick", "prompt": "Process the full audio offline. Find every instance of the word \"sick\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.17, 4.41, "sick"]], "response_mode": "offline", "answer": "[{\"word\":\"sick\",\"start\":4.17,\"end\":4.41}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "semantic_counting", "query": "date", "prompt": "Listen to the full audio. Count every occurrence of a date. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.52, 7.26, "one"], [8.88, 9.79, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.52,\"end\":7.26},{\"word\":\"two\",\"start\":8.88,\"end\":9.79}]"}
{"utterance_id": "251-137823-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[7.84, 8.22, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":7.84,\"end\":8.22}]"}
{"utterance_id": "1272-135031-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.18, 3.89, "polychrome"]], "response_mode": "offline", "answer": "[{\"word\":\"polychrome\",\"start\":3.18,\"end\":3.89}]"}
{"utterance_id": "5694-64029-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0004.flac", "task_type": "lexical_detection", "query": "yankee", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"yankee\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.63, 0.96, "yankee"]], "response_mode": "offline", "answer": "[{\"word\":\"yankee\",\"start\":0.63,\"end\":0.96}]"}
{"utterance_id": "3752-4944-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0026.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.26, "doctor"]], "response_mode": "offline", "answer": "[{\"word\":\"doctor\",\"start\":1.77,\"end\":2.26}]"}
{"utterance_id": "3170-137482-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0028.flac", "task_type": "lexical_detection", "query": "result", "prompt": "After the audio ends, extract all occurrences of the word \"result\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.4, "result"]], "response_mode": "offline", "answer": "[{\"word\":\"result\",\"start\":1.91,\"end\":2.4}]"}
{"utterance_id": "7976-110523-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.81, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":1.44,\"end\":1.81}]"}
{"utterance_id": "8842-302203-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.5, "eyes"], [4.86, 5.18, "brain"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":3.29,\"end\":3.5},{\"word\":\"brain\",\"start\":4.86,\"end\":5.18}]"}
{"utterance_id": "2277-149896-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0018.flac", "task_type": "lexical_detection", "query": "devil", "prompt": "After listening to the whole recording, output every instance of the word \"devil\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.26, "devil"]], "response_mode": "offline", "answer": "[{\"word\":\"devil\",\"start\":3.79,\"end\":4.26}]"}
{"utterance_id": "1673-143396-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.24, 2.62, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.24,\"end\":2.62}]"}
{"utterance_id": "2277-149874-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0014.flac", "task_type": "lexical_detection", "query": "mantel", "prompt": "Process the full audio offline. Find every instance of the word \"mantel\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.82, 3.27, "mantel"]], "response_mode": "offline", "answer": "[{\"word\":\"mantel\",\"start\":2.82,\"end\":3.27}]"}
{"utterance_id": "2803-154320-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0004.flac", "task_type": "semantic_counting", "query": "country", "prompt": "When the utterance is complete, report how a country name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.38, 6.28, "one"], [8.01, 8.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.38,\"end\":6.28},{\"word\":\"two\",\"start\":8.01,\"end\":8.83}]"}
{"utterance_id": "5694-64025-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0007.flac", "task_type": "lexical_detection", "query": "hail", "prompt": "Once the audio has finished, identify every instance of the word \"hail\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.45, 2.79, "hail"]], "response_mode": "offline", "answer": "[{\"word\":\"hail\",\"start\":2.45,\"end\":2.79}]"}
{"utterance_id": "6241-61943-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0005.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Do not answer until the audio ends. Then list each instance of a country name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.11, 4.64, "denmark"]], "response_mode": "offline", "answer": "[{\"word\":\"denmark\",\"start\":4.11,\"end\":4.64}]"}
{"utterance_id": "3000-15664-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0026.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 1.25, "mule deer"]], "response_mode": "offline", "answer": "[{\"word\":\"mule deer\",\"start\":0.54,\"end\":1.25}]"}
{"utterance_id": "2078-142845-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0009.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.44, 3.07, "millet"]], "response_mode": "offline", "answer": "[{\"word\":\"millet\",\"start\":2.44,\"end\":3.07}]"}
{"utterance_id": "2428-83699-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0014.flac", "task_type": "lexical_detection", "query": "horrible", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"horrible\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.93, 1.53, "horrible"]], "response_mode": "offline", "answer": "[{\"word\":\"horrible\",\"start\":0.93,\"end\":1.53}]"}
{"utterance_id": "2035-147961-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 1.04, "antonia's"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia's\",\"start\":0.48,\"end\":1.04}]"}
{"utterance_id": "6295-244435-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0022.flac", "task_type": "lexical_detection", "query": "changes", "prompt": "Once the audio has finished, identify every instance of the word \"changes\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.72, 2.46, "changes"]], "response_mode": "offline", "answer": "[{\"word\":\"changes\",\"start\":1.72,\"end\":2.46}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.11, "neck"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":3.78,\"end\":4.11}]"}
{"utterance_id": "3081-166546-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0036.flac", "task_type": "lexical_detection", "query": "gryce", "prompt": "Process the full audio offline. Find every instance of the word \"gryce\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.57, "gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"gryce\",\"start\":2.05,\"end\":2.57}]"}
{"utterance_id": "1988-147956-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0021.flac", "task_type": "lexical_detection", "query": "blazing", "prompt": "When the utterance is complete, report all instances of the word \"blazing\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.82, "blazing"]], "response_mode": "offline", "answer": "[{\"word\":\"blazing\",\"start\":2.3,\"end\":2.82}]"}
{"utterance_id": "2277-149896-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0017.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.52, "lunch"]], "response_mode": "offline", "answer": "[{\"word\":\"lunch\",\"start\":2.16,\"end\":2.52}]"}
{"utterance_id": "1272-135031-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0015.flac", "task_type": "lexical_detection", "query": "cavern", "prompt": "Process the full audio offline. Find every instance of the word \"cavern\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 3.43, "cavern"]], "response_mode": "offline", "answer": "[{\"word\":\"cavern\",\"start\":2.71,\"end\":3.43}]"}
{"utterance_id": "1272-135031-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.22, 2.84, "kaliko"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":2.22,\"end\":2.84}]"}
{"utterance_id": "1919-142785-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0050.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[8.01, 8.34, "butter"]], "response_mode": "offline", "answer": "[{\"word\":\"butter\",\"start\":8.01,\"end\":8.34}]"}
{"utterance_id": "6295-64301-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0025.flac", "task_type": "lexical_detection", "query": "bosom", "prompt": "After the audio ends, extract all occurrences of the word \"bosom\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.82, 5.35, "bosom"]], "response_mode": "offline", "answer": "[{\"word\":\"bosom\",\"start\":4.82,\"end\":5.35}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "semantic_detection", "query": "color", "prompt": "You will hear a complete utterance. When it ends, list every instance of a color name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.8, "silver"], [6.67, 7.0, "silver"]], "response_mode": "offline", "answer": "[{\"word\":\"silver\",\"start\":0.36,\"end\":0.8},{\"word\":\"silver\",\"start\":6.67,\"end\":7.0}]"}
{"utterance_id": "2086-149220-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0001.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 0.97, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":0.64,\"end\":0.97}]"}
{"utterance_id": "422-122949-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0011.flac", "task_type": "lexical_counting", "query": "qualities", "prompt": "Listen to the entire clip, then report the count sequence for the word \"qualities\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.77, "one"], [12.05, 12.72, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.15,\"end\":3.77},{\"word\":\"two\",\"start\":12.05,\"end\":12.72}]"}
{"utterance_id": "1988-24833-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0021.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.53, 0.77, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.53,\"end\":0.77}]"}
{"utterance_id": "2035-147961-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0035.flac", "task_type": "lexical_detection", "query": "unfortunate", "prompt": "After the audio ends, extract all occurrences of the word \"unfortunate\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.49, 5.19, "unfortunate"]], "response_mode": "offline", "answer": "[{\"word\":\"unfortunate\",\"start\":4.49,\"end\":5.19}]"}
{"utterance_id": "84-121550-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0018.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 1.11, "second"], [5.56, 6.12, "third"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.53,\"end\":1.11},{\"word\":\"third\",\"start\":5.56,\"end\":6.12}]"}
{"utterance_id": "2803-154328-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0007.flac", "task_type": "lexical_detection", "query": "helena", "prompt": "Once the audio has finished, identify every instance of the word \"helena\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.24, 4.84, "helena"]], "response_mode": "offline", "answer": "[{\"word\":\"helena\",\"start\":4.24,\"end\":4.84}]"}
{"utterance_id": "8842-302201-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0014.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "Listen to the full audio. Count every occurrence of an ordinal number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.64, "one"], [2.56, 2.96, "two"], [8.36, 8.78, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.39,\"end\":0.64},{\"word\":\"two\",\"start\":2.56,\"end\":2.96},{\"word\":\"three\",\"start\":8.36,\"end\":8.78}]"}
{"utterance_id": "2086-149220-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0022.flac", "task_type": "lexical_detection", "query": "artist", "prompt": "After the audio ends, extract all occurrences of the word \"artist\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.33, "artist"]], "response_mode": "offline", "answer": "[{\"word\":\"artist\",\"start\":1.82,\"end\":2.33}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a person's name: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[2.43, 2.87, "one"], [4.27, 4.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.43,\"end\":2.87},{\"word\":\"two\",\"start\":4.27,\"end\":4.66}]"}
{"utterance_id": "2086-149220-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0035.flac", "task_type": "lexical_detection", "query": "earnestly", "prompt": "After the audio ends, extract all occurrences of the word \"earnestly\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.85, 4.32, "earnestly"]], "response_mode": "offline", "answer": "[{\"word\":\"earnestly\",\"start\":3.85,\"end\":4.32}]"}
{"utterance_id": "5694-64025-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0017.flac", "task_type": "lexical_detection", "query": "captured", "prompt": "Once the audio has finished, identify every instance of the word \"captured\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.65, 2.1, "captured"]], "response_mode": "offline", "answer": "[{\"word\":\"captured\",\"start\":1.65,\"end\":2.1}]"}
{"utterance_id": "7976-110523-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0017.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[7.02, 7.32, "one"], [12.33, 12.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.02,\"end\":7.32},{\"word\":\"two\",\"start\":12.33,\"end\":12.88}]"}
{"utterance_id": "3000-15664-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0035.flac", "task_type": "lexical_counting", "query": "river", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"river\" with word/start/end. If none, reply [].", "labels": [[11.38, 11.85, "one"], [16.24, 16.62, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":11.38,\"end\":11.85},{\"word\":\"two\",\"start\":16.24,\"end\":16.62}]"}
{"utterance_id": "7850-111771-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0009.flac", "task_type": "lexical_detection", "query": "perform", "prompt": "After listening to the whole recording, output every instance of the word \"perform\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.63, 5.23, "perform"]], "response_mode": "offline", "answer": "[{\"word\":\"perform\",\"start\":4.63,\"end\":5.23}]"}
{"utterance_id": "174-50561-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0016.flac", "task_type": "lexical_counting", "query": "singer", "prompt": "Listen to the full audio. Count every occurrence of the word \"singer\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.85, "one"], [8.43, 8.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.44,\"end\":1.85},{\"word\":\"two\",\"start\":8.43,\"end\":8.8}]"}
{"utterance_id": "3752-4943-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0001.flac", "task_type": "lexical_detection", "query": "gang", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"gang\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.24, "gang"]], "response_mode": "offline", "answer": "[{\"word\":\"gang\",\"start\":1.87,\"end\":2.24}]"}
{"utterance_id": "652-129742-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0001.flac", "task_type": "lexical_detection", "query": "lettuce", "prompt": "Once the audio has finished, identify every instance of the word \"lettuce\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.67, 5.1, "lettuce"]], "response_mode": "offline", "answer": "[{\"word\":\"lettuce\",\"start\":4.67,\"end\":5.1}]"}
{"utterance_id": "1673-143397-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.7, 9.35, "alexandria"]], "response_mode": "offline", "answer": "[{\"word\":\"alexandria\",\"start\":8.7,\"end\":9.35}]"}
{"utterance_id": "2428-83705-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0010.flac", "task_type": "lexical_detection", "query": "cousin", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"cousin\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.84, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":0.52,\"end\":0.84}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.67, 15.15, "ears"]], "response_mode": "offline", "answer": "[{\"word\":\"ears\",\"start\":14.67,\"end\":15.15}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen to the full audio. After it finishes, find every instance of a state or province name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.35, 11.99, "northumbria"]], "response_mode": "offline", "answer": "[{\"word\":\"northumbria\",\"start\":11.35,\"end\":11.99}]"}
{"utterance_id": "2035-147960-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0006.flac", "task_type": "lexical_detection", "query": "scared", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"scared\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.98, "scared"]], "response_mode": "offline", "answer": "[{\"word\":\"scared\",\"start\":2.26,\"end\":2.98}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.3, 7.69, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":7.3,\"end\":7.69}]"}
{"utterance_id": "5536-43359-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.69, 2.88, "red"], [6.84, 7.15, "black"], [10.71, 10.97, "white"], [12.06, 12.35, "green"], [14.96, 15.27, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":2.69,\"end\":2.88},{\"word\":\"black\",\"start\":6.84,\"end\":7.15},{\"word\":\"white\",\"start\":10.71,\"end\":10.97},{\"word\":\"green\",\"start\":12.06,\"end\":12.35},{\"word\":\"red\",\"start\":14.96,\"end\":15.27}]"}
{"utterance_id": "3752-4943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0007.flac", "task_type": "semantic_counting", "query": "color", "prompt": "Listen to the full audio. Count every occurrence of a color name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.5, "one"], [5.8, 6.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.01,\"end\":2.5},{\"word\":\"two\",\"start\":5.8,\"end\":6.16}]"}
{"utterance_id": "5694-64038-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0002.flac", "task_type": "lexical_detection", "query": "passed", "prompt": "Once the audio has finished, identify every instance of the word \"passed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.65, 1.02, "passed"]], "response_mode": "offline", "answer": "[{\"word\":\"passed\",\"start\":0.65,\"end\":1.02}]"}
{"utterance_id": "2428-83699-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.17, 10.61, "lucy's"]], "response_mode": "offline", "answer": "[{\"word\":\"lucy's\",\"start\":10.17,\"end\":10.61}]"}
{"utterance_id": "6313-76958-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 2.93, "big foot"]], "response_mode": "offline", "answer": "[{\"word\":\"big foot\",\"start\":2.46,\"end\":2.93}]"}
{"utterance_id": "251-118436-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.37, 5.88, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":5.37,\"end\":5.88}]"}
{"utterance_id": "6313-66129-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0006.flac", "task_type": "lexical_detection", "query": "young", "prompt": "When the utterance is complete, report all instances of the word \"young\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.39, "young"]], "response_mode": "offline", "answer": "[{\"word\":\"young\",\"start\":1.11,\"end\":1.39}]"}
{"utterance_id": "7850-286674-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.3, 4.61, "back"]], "response_mode": "offline", "answer": "[{\"word\":\"back\",\"start\":4.3,\"end\":4.61}]"}
{"utterance_id": "1993-147149-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0011.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.19, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":0.93,\"end\":1.19}]"}
{"utterance_id": "174-84280-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0003.flac", "task_type": "lexical_detection", "query": "understanding", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"understanding\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.11, 6.99, "understanding"]], "response_mode": "offline", "answer": "[{\"word\":\"understanding\",\"start\":6.11,\"end\":6.99}]"}
{"utterance_id": "2277-149874-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.42, 0.85, "minnie's"]], "response_mode": "offline", "answer": "[{\"word\":\"minnie's\",\"start\":0.42,\"end\":0.85}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.9, 3.32, "horse"], [6.92, 7.48, "centaur"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.9,\"end\":3.32},{\"word\":\"centaur\",\"start\":6.92,\"end\":7.48}]"}
{"utterance_id": "6319-64726-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0015.flac", "task_type": "lexical_detection", "query": "heads", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"heads\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.22, 4.55, "heads"]], "response_mode": "offline", "answer": "[{\"word\":\"heads\",\"start\":4.22,\"end\":4.55}]"}
{"utterance_id": "3536-23268-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 1.17, "missus horton"]], "response_mode": "offline", "answer": "[{\"word\":\"missus horton\",\"start\":0.47,\"end\":1.17}]"}
{"utterance_id": "1673-143397-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0014.flac", "task_type": "lexical_detection", "query": "emperor", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"emperor\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.8, "emperor"]], "response_mode": "offline", "answer": "[{\"word\":\"emperor\",\"start\":2.44,\"end\":2.8}]"}
{"utterance_id": "2277-149874-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0006.flac", "task_type": "lexical_detection", "query": "rooms", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"rooms\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.26, "rooms"]], "response_mode": "offline", "answer": "[{\"word\":\"rooms\",\"start\":0.89,\"end\":1.26}]"}
{"utterance_id": "1272-141231-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0017.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.22, "handler"]], "response_mode": "offline", "answer": "[{\"word\":\"handler\",\"start\":0.73,\"end\":1.22}]"}
{"utterance_id": "2035-147961-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0024.flac", "task_type": "lexical_detection", "query": "left", "prompt": "Once the audio has finished, identify every instance of the word \"left\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.1, 2.56, "left"]], "response_mode": "offline", "answer": "[{\"word\":\"left\",\"start\":2.1,\"end\":2.56}]"}
{"utterance_id": "2803-154328-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.18, 2.76, "hearts"]], "response_mode": "offline", "answer": "[{\"word\":\"hearts\",\"start\":2.18,\"end\":2.76}]"}
{"utterance_id": "1673-143397-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.24, 1.64, "cyril"], [3.32, 3.77, "julian"]], "response_mode": "offline", "answer": "[{\"word\":\"cyril\",\"start\":1.24,\"end\":1.64},{\"word\":\"julian\",\"start\":3.32,\"end\":3.77}]"}
{"utterance_id": "6295-244435-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0005.flac", "task_type": "lexical_detection", "query": "flag", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"flag\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.19, "flag"]], "response_mode": "offline", "answer": "[{\"word\":\"flag\",\"start\":0.73,\"end\":1.19}]"}
{"utterance_id": "7976-110523-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0008.flac", "task_type": "lexical_detection", "query": "slice", "prompt": "After listening to the whole recording, output every instance of the word \"slice\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.79, 5.11, "slice"]], "response_mode": "offline", "answer": "[{\"word\":\"slice\",\"start\":4.79,\"end\":5.11}]"}
{"utterance_id": "6313-66125-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.7, 3.12, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":2.7,\"end\":3.12}]"}
{"utterance_id": "652-130737-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0004.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.62, 1.88, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.62,\"end\":1.88}]"}
{"utterance_id": "1462-170145-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0019.flac", "task_type": "lexical_detection", "query": "talking", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"talking\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.35, 1.8, "talking"]], "response_mode": "offline", "answer": "[{\"word\":\"talking\",\"start\":1.35,\"end\":1.8}]"}
{"utterance_id": "1919-142785-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0019.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "After listening to the whole recording, output timed count words for an ordinal number (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.06, 1.38, "one"], [3.82, 4.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.06,\"end\":1.38},{\"word\":\"two\",\"start\":3.82,\"end\":4.43}]"}
{"utterance_id": "8297-275156-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0013.flac", "task_type": "lexical_detection", "query": "marriage", "prompt": "Process the full audio offline. Find every instance of the word \"marriage\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.66, 4.1, "marriage"]], "response_mode": "offline", "answer": "[{\"word\":\"marriage\",\"start\":3.66,\"end\":4.1}]"}
{"utterance_id": "2078-142845-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0028.flac", "task_type": "lexical_detection", "query": "nicely", "prompt": "After the audio ends, extract all occurrences of the word \"nicely\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.89, 3.42, "nicely"]], "response_mode": "offline", "answer": "[{\"word\":\"nicely\",\"start\":2.89,\"end\":3.42}]"}
{"utterance_id": "6241-61946-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0017.flac", "task_type": "lexical_detection", "query": "horse", "prompt": "Once the audio has finished, identify every instance of the word \"horse\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.55, 2.98, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.55,\"end\":2.98}]"}
{"utterance_id": "2428-83705-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0002.flac", "task_type": "lexical_detection", "query": "ann's", "prompt": "When the utterance is complete, report all instances of the word \"ann's\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.29, 11.68, "ann's"]], "response_mode": "offline", "answer": "[{\"word\":\"ann's\",\"start\":11.29,\"end\":11.68}]"}
{"utterance_id": "777-126732-0075", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0075.flac", "task_type": "lexical_detection", "query": "went", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"went\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.38, 3.61, "went"]], "response_mode": "offline", "answer": "[{\"word\":\"went\",\"start\":3.38,\"end\":3.61}]"}
{"utterance_id": "5536-43363-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0006.flac", "task_type": "lexical_detection", "query": "slain", "prompt": "When the utterance is complete, report all instances of the word \"slain\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.45, "slain"]], "response_mode": "offline", "answer": "[{\"word\":\"slain\",\"start\":1.14,\"end\":1.45}]"}
{"utterance_id": "1673-143396-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0014.flac", "task_type": "lexical_detection", "query": "abandoned", "prompt": "Once the audio has finished, identify every instance of the word \"abandoned\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.6, 9.22, "abandoned"]], "response_mode": "offline", "answer": "[{\"word\":\"abandoned\",\"start\":8.6,\"end\":9.22}]"}
{"utterance_id": "3752-4943-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0008.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 2.04, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.59,\"end\":2.04}]"}
{"utterance_id": "174-168635-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0017.flac", "task_type": "lexical_detection", "query": "thoughts", "prompt": "After listening to the whole recording, output every instance of the word \"thoughts\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.46, 4.11, "thoughts"]], "response_mode": "offline", "answer": "[{\"word\":\"thoughts\",\"start\":3.46,\"end\":4.11}]"}
{"utterance_id": "3752-4943-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0015.flac", "task_type": "lexical_detection", "query": "impertinent", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"impertinent\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.11, 0.92, "impertinent"]], "response_mode": "offline", "answer": "[{\"word\":\"impertinent\",\"start\":0.11,\"end\":0.92}]"}
{"utterance_id": "1462-170145-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0020.flac", "task_type": "lexical_detection", "query": "yes", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"yes\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.51, 0.86, "yes"]], "response_mode": "offline", "answer": "[{\"word\":\"yes\",\"start\":0.51,\"end\":0.86}]"}
{"utterance_id": "1272-141231-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0025.flac", "task_type": "lexical_counting", "query": "during", "prompt": "After listening to the whole recording, output timed count words for the word \"during\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.98, 2.22, "one"], [3.96, 4.33, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.98,\"end\":2.22},{\"word\":\"two\",\"start\":3.96,\"end\":4.33}]"}
{"utterance_id": "6241-61943-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0000.flac", "task_type": "lexical_detection", "query": "valkyrie", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"valkyrie\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.03, 6.6, "valkyrie"]], "response_mode": "offline", "answer": "[{\"word\":\"valkyrie\",\"start\":6.03,\"end\":6.6}]"}
{"utterance_id": "6313-66129-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.47, 3.79, "ned"], [5.97, 6.37, "jimmie"]], "response_mode": "offline", "answer": "[{\"word\":\"ned\",\"start\":3.47,\"end\":3.79},{\"word\":\"jimmie\",\"start\":5.97,\"end\":6.37}]"}
{"utterance_id": "3536-23268-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0007.flac", "task_type": "lexical_counting", "query": "seat", "prompt": "Listen to the full audio. Count every occurrence of the word \"seat\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.12, "one"], [6.78, 7.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.75,\"end\":2.12},{\"word\":\"two\",\"start\":6.78,\"end\":7.0}]"}
{"utterance_id": "174-50561-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0016.flac", "task_type": "lexical_detection", "query": "drowsy", "prompt": "When the utterance is complete, report all instances of the word \"drowsy\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.35, 6.79, "drowsy"]], "response_mode": "offline", "answer": "[{\"word\":\"drowsy\",\"start\":6.35,\"end\":6.79}]"}
{"utterance_id": "5338-24640-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0008.flac", "task_type": "lexical_detection", "query": "cruelty", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cruelty\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.25, 5.9, "cruelty"]], "response_mode": "offline", "answer": "[{\"word\":\"cruelty\",\"start\":5.25,\"end\":5.9}]"}
{"utterance_id": "6313-66125-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0005.flac", "task_type": "lexical_detection", "query": "allowing", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"allowing\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.0, 1.33, "allowing"]], "response_mode": "offline", "answer": "[{\"word\":\"allowing\",\"start\":1.0,\"end\":1.33}]"}
{"utterance_id": "2803-154320-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0001.flac", "task_type": "lexical_detection", "query": "mangles", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"mangles\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.95, 1.6, "mangles"]], "response_mode": "offline", "answer": "[{\"word\":\"mangles\",\"start\":0.95,\"end\":1.6}]"}
{"utterance_id": "1462-170142-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.58, 2.08, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":1.58,\"end\":2.08}]"}
{"utterance_id": "6319-64726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0002.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "Do not answer until the audio ends. Then list each instance of a disease name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[11.25, 11.72, "wound"]], "response_mode": "offline", "answer": "[{\"word\":\"wound\",\"start\":11.25,\"end\":11.72}]"}
{"utterance_id": "1462-170145-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0014.flac", "task_type": "lexical_detection", "query": "mistaken", "prompt": "When the utterance is complete, report all instances of the word \"mistaken\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.45, "mistaken"]], "response_mode": "offline", "answer": "[{\"word\":\"mistaken\",\"start\":1.8,\"end\":2.45}]"}
{"utterance_id": "2035-152373-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0016.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.35, "cattle"]], "response_mode": "offline", "answer": "[{\"word\":\"cattle\",\"start\":4.99,\"end\":5.35}]"}
{"utterance_id": "84-121550-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.22, 6.6, "neck"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":6.22,\"end\":6.6}]"}
{"utterance_id": "5895-34622-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0005.flac", "task_type": "lexical_detection", "query": "grounds", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"grounds\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.51, 5.94, "grounds"]], "response_mode": "offline", "answer": "[{\"word\":\"grounds\",\"start\":5.51,\"end\":5.94}]"}
{"utterance_id": "1988-147956-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.5, 1.83, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.5,\"end\":1.83}]"}
{"utterance_id": "6295-64301-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0024.flac", "task_type": "lexical_counting", "query": "whom", "prompt": "Once the audio has finished, list each occurrence of the word \"whom\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.13, 2.34, "one"], [4.74, 5.02, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.13,\"end\":2.34},{\"word\":\"two\",\"start\":4.74,\"end\":5.02}]"}
{"utterance_id": "7976-105575-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0019.flac", "task_type": "lexical_detection", "query": "carried", "prompt": "After listening to the whole recording, output every instance of the word \"carried\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.83, 3.23, "carried"]], "response_mode": "offline", "answer": "[{\"word\":\"carried\",\"start\":2.83,\"end\":3.23}]"}
{"utterance_id": "7976-105575-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0026.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 1.12, "rosecrans"]], "response_mode": "offline", "answer": "[{\"word\":\"rosecrans\",\"start\":0.41,\"end\":1.12}]"}
{"utterance_id": "6345-64257-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.51, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":1.23,\"end\":1.51}]"}
{"utterance_id": "3170-137482-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0017.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the full audio. Count every occurrence of a number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.61, "one"], [8.09, 8.69, "two"], [10.69, 10.96, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.44,\"end\":1.61},{\"word\":\"two\",\"start\":8.09,\"end\":8.69},{\"word\":\"three\",\"start\":10.69,\"end\":10.96}]"}
{"utterance_id": "2412-153947-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0009.flac", "task_type": "lexical_detection", "query": "ultimately", "prompt": "Process the full audio offline. Find every instance of the word \"ultimately\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.23, 3.72, "ultimately"]], "response_mode": "offline", "answer": "[{\"word\":\"ultimately\",\"start\":3.23,\"end\":3.72}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "lexical_counting", "query": "features", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"features\" with word/start/end. If none, reply [].", "labels": [[8.51, 9.04, "one"], [20.78, 21.25, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.51,\"end\":9.04},{\"word\":\"two\",\"start\":20.78,\"end\":21.25}]"}
{"utterance_id": "1673-143396-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0005.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a family relation (word, start, end). If none, reply [].", "labels": [[0.51, 0.81, "one"], [10.47, 10.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.81},{\"word\":\"two\",\"start\":10.47,\"end\":10.88}]"}
{"utterance_id": "2277-149896-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.25, 2.59, "eyes"], [4.13, 4.37, "brow"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":2.25,\"end\":2.59},{\"word\":\"brow\",\"start\":4.13,\"end\":4.37}]"}
{"utterance_id": "2428-83699-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.49, 5.85, "chop"]], "response_mode": "offline", "answer": "[{\"word\":\"chop\",\"start\":5.49,\"end\":5.85}]"}
{"utterance_id": "3081-166546-0071", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0071.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.39, 4.13, "sweetwater"], [5.16, 6.02, "mister gryce"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":3.39,\"end\":4.13},{\"word\":\"mister gryce\",\"start\":5.16,\"end\":6.02}]"}
{"utterance_id": "6241-61946-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0021.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.24, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":1.83,\"end\":2.24}]"}
{"utterance_id": "1919-142785-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0049.flac", "task_type": "lexical_detection", "query": "consist", "prompt": "Once the audio has finished, identify every instance of the word \"consist\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.04, 6.57, "consist"]], "response_mode": "offline", "answer": "[{\"word\":\"consist\",\"start\":6.04,\"end\":6.57}]"}
{"utterance_id": "3752-4943-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0021.flac", "task_type": "lexical_detection", "query": "beneath", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"beneath\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.37, 3.68, "beneath"]], "response_mode": "offline", "answer": "[{\"word\":\"beneath\",\"start\":3.37,\"end\":3.68}]"}
{"utterance_id": "8297-275154-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.6, "brains"]], "response_mode": "offline", "answer": "[{\"word\":\"brains\",\"start\":2.98,\"end\":3.6}]"}
{"utterance_id": "1988-147956-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0026.flac", "task_type": "lexical_detection", "query": "sky", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sky\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.27, 3.67, "sky"]], "response_mode": "offline", "answer": "[{\"word\":\"sky\",\"start\":3.27,\"end\":3.67}]"}
{"utterance_id": "174-168635-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.65, 2.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.65,\"end\":2.83}]"}
{"utterance_id": "5536-43359-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.92, "grandparents"]], "response_mode": "offline", "answer": "[{\"word\":\"grandparents\",\"start\":2.19,\"end\":2.92}]"}
{"utterance_id": "3170-137482-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0038.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[4.84, 5.26, "hermit"]], "response_mode": "offline", "answer": "[{\"word\":\"hermit\",\"start\":4.84,\"end\":5.26}]"}
{"utterance_id": "3576-138058-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0037.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.6, 1.13, "captains"], [3.82, 4.3, "regent's"]], "response_mode": "offline", "answer": "[{\"word\":\"captains\",\"start\":0.6,\"end\":1.13},{\"word\":\"regent's\",\"start\":3.82,\"end\":4.3}]"}
{"utterance_id": "3853-163249-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0009.flac", "task_type": "lexical_detection", "query": "proud", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"proud\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.73, 2.14, "proud"]], "response_mode": "offline", "answer": "[{\"word\":\"proud\",\"start\":1.73,\"end\":2.14}]"}
{"utterance_id": "2035-147960-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.4, 5.7, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":5.4,\"end\":5.7}]"}
{"utterance_id": "6295-64301-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0007.flac", "task_type": "lexical_detection", "query": "mood", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"mood\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.56, "mood"]], "response_mode": "offline", "answer": "[{\"word\":\"mood\",\"start\":2.19,\"end\":2.56}]"}
{"utterance_id": "5338-24615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0005.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.96, 7.15, "horse chestnuts"], [8.87, 9.56, "sycamores"]], "response_mode": "offline", "answer": "[{\"word\":\"horse chestnuts\",\"start\":5.96,\"end\":7.15},{\"word\":\"sycamores\",\"start\":8.87,\"end\":9.56}]"}
{"utterance_id": "1272-128104-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0011.flac", "task_type": "lexical_detection", "query": "severe", "prompt": "Once the audio has finished, identify every instance of the word \"severe\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.86, 2.35, "severe"]], "response_mode": "offline", "answer": "[{\"word\":\"severe\",\"start\":1.86,\"end\":2.35}]"}
{"utterance_id": "251-118436-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0023.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.33, "soul"]], "response_mode": "offline", "answer": "[{\"word\":\"soul\",\"start\":0.88,\"end\":1.33}]"}
{"utterance_id": "1919-142785-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0012.flac", "task_type": "lexical_detection", "query": "saucepan", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"saucepan\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.27, 3.92, "saucepan"]], "response_mode": "offline", "answer": "[{\"word\":\"saucepan\",\"start\":3.27,\"end\":3.92}]"}
{"utterance_id": "5536-43363-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0002.flac", "task_type": "lexical_detection", "query": "blacken", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"blacken\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.16, "blacken"]], "response_mode": "offline", "answer": "[{\"word\":\"blacken\",\"start\":0.84,\"end\":1.16}]"}
{"utterance_id": "5895-34615-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0019.flac", "task_type": "lexical_detection", "query": "touch", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"touch\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.04, 6.54, "touch"]], "response_mode": "offline", "answer": "[{\"word\":\"touch\",\"start\":6.04,\"end\":6.54}]"}
{"utterance_id": "6295-244435-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0033.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.39, 0.94, "afternoon"]], "response_mode": "offline", "answer": "[{\"word\":\"afternoon\",\"start\":0.39,\"end\":0.94}]"}
{"utterance_id": "6313-66125-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0009.flac", "task_type": "lexical_detection", "query": "protest", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"protest\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.7, 1.59, "protest"]], "response_mode": "offline", "answer": "[{\"word\":\"protest\",\"start\":0.7,\"end\":1.59}]"}
{"utterance_id": "6319-275224-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0005.flac", "task_type": "lexical_detection", "query": "advantages", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"advantages\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.13, 4.77, "advantages"]], "response_mode": "offline", "answer": "[{\"word\":\"advantages\",\"start\":4.13,\"end\":4.77}]"}
{"utterance_id": "1272-128104-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0013.flac", "task_type": "lexical_detection", "query": "missed", "prompt": "After the audio ends, extract all occurrences of the word \"missed\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.88, "missed"]], "response_mode": "offline", "answer": "[{\"word\":\"missed\",\"start\":1.51,\"end\":1.88}]"}
{"utterance_id": "5338-284437-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0010.flac", "task_type": "lexical_detection", "query": "rosy", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"rosy\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.71, 4.1, "rosy"]], "response_mode": "offline", "answer": "[{\"word\":\"rosy\",\"start\":3.71,\"end\":4.1}]"}
{"utterance_id": "3576-138058-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0028.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "After the audio ends, tally every instance of a family relation. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.78, 4.07, "one"], [9.59, 10.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.78,\"end\":4.07},{\"word\":\"two\",\"start\":9.59,\"end\":10.18}]"}
{"utterance_id": "2078-142845-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0035.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.84, "crumpets"]], "response_mode": "offline", "answer": "[{\"word\":\"crumpets\",\"start\":2.17,\"end\":2.84}]"}
{"utterance_id": "3000-15664-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0011.flac", "task_type": "semantic_counting", "query": "weather", "prompt": "Listen to the full audio. Count every occurrence of a weather term. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.59, "one"], [9.25, 9.6, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.83,\"end\":1.59},{\"word\":\"two\",\"start\":9.25,\"end\":9.6}]"}
{"utterance_id": "5338-284437-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0007.flac", "task_type": "lexical_detection", "query": "palace", "prompt": "After listening to the whole recording, output every instance of the word \"palace\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.68, 3.29, "palace"]], "response_mode": "offline", "answer": "[{\"word\":\"palace\",\"start\":2.68,\"end\":3.29}]"}
{"utterance_id": "84-121123-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0003.flac", "task_type": "lexical_detection", "query": "thus", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"thus\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.27, 3.47, "thus"]], "response_mode": "offline", "answer": "[{\"word\":\"thus\",\"start\":3.27,\"end\":3.47}]"}
{"utterance_id": "1272-128104-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.18, 5.89, "roast beef"]], "response_mode": "offline", "answer": "[{\"word\":\"roast beef\",\"start\":5.18,\"end\":5.89}]"}
{"utterance_id": "2078-142845-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0028.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.7, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":2.3,\"end\":2.7}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.09, 6.3, "ear"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":6.09,\"end\":6.3}]"}
{"utterance_id": "777-126732-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0043.flac", "task_type": "lexical_detection", "query": "satisfied", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"satisfied\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.42, "satisfied"]], "response_mode": "offline", "answer": "[{\"word\":\"satisfied\",\"start\":0.96,\"end\":1.42}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.21, 3.1, "one"], [9.58, 10.03, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.21,\"end\":3.1},{\"word\":\"two\",\"start\":9.58,\"end\":10.03}]"}
{"utterance_id": "2803-161169-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0008.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.14, "nut"], [9.6, 10.06, "nut"], [12.23, 12.94, "chestnut"]], "response_mode": "offline", "answer": "[{\"word\":\"nut\",\"start\":3.75,\"end\":4.14},{\"word\":\"nut\",\"start\":9.6,\"end\":10.06},{\"word\":\"chestnut\",\"start\":12.23,\"end\":12.94}]"}
{"utterance_id": "84-121550-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0017.flac", "task_type": "lexical_detection", "query": "circle", "prompt": "When the utterance is complete, report all instances of the word \"circle\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.78, "circle"]], "response_mode": "offline", "answer": "[{\"word\":\"circle\",\"start\":2.33,\"end\":2.78}]"}
{"utterance_id": "1919-142785-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0021.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.58, "eggs"]], "response_mode": "offline", "answer": "[{\"word\":\"eggs\",\"start\":1.06,\"end\":1.58}]"}
{"utterance_id": "1462-170145-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0009.flac", "task_type": "lexical_detection", "query": "write", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"write\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.27, "write"]], "response_mode": "offline", "answer": "[{\"word\":\"write\",\"start\":2.04,\"end\":2.27}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "lexical_detection", "query": "sending", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sending\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.85, 7.23, "sending"]], "response_mode": "offline", "answer": "[{\"word\":\"sending\",\"start\":6.85,\"end\":7.23}]"}
{"utterance_id": "2902-9006-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0017.flac", "task_type": "lexical_detection", "query": "climate", "prompt": "Once the audio has finished, identify every instance of the word \"climate\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.46, 1.13, "climate"]], "response_mode": "offline", "answer": "[{\"word\":\"climate\",\"start\":0.46,\"end\":1.13}]"}
{"utterance_id": "5895-34622-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0022.flac", "task_type": "lexical_detection", "query": "preachers", "prompt": "Process the full audio offline. Find every instance of the word \"preachers\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.05, 7.71, "preachers"]], "response_mode": "offline", "answer": "[{\"word\":\"preachers\",\"start\":7.05,\"end\":7.71}]"}
{"utterance_id": "1993-147149-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.87, 2.24, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":1.87,\"end\":2.24}]"}
{"utterance_id": "2428-83705-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0033.flac", "task_type": "lexical_counting", "query": "giving", "prompt": "When the utterance is complete, report how the word \"giving\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.49, "one"], [5.73, 6.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.14,\"end\":1.49},{\"word\":\"two\",\"start\":5.73,\"end\":6.12}]"}
{"utterance_id": "5694-64038-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0025.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[3.97, 4.27, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":3.97,\"end\":4.27}]"}
{"utterance_id": "2078-142845-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a plant name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.46, 2.93, "millet"]], "response_mode": "offline", "answer": "[{\"word\":\"millet\",\"start\":2.46,\"end\":2.93}]"}
{"utterance_id": "1993-147149-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0018.flac", "task_type": "lexical_detection", "query": "fearing", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fearing\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.63, "fearing"]], "response_mode": "offline", "answer": "[{\"word\":\"fearing\",\"start\":2.26,\"end\":2.63}]"}
{"utterance_id": "6313-66125-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0015.flac", "task_type": "lexical_detection", "query": "cautiously", "prompt": "Process the full audio offline. Find every instance of the word \"cautiously\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.18, 0.73, "cautiously"]], "response_mode": "offline", "answer": "[{\"word\":\"cautiously\",\"start\":0.18,\"end\":0.73}]"}
{"utterance_id": "2277-149896-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0014.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.57, "madison"]], "response_mode": "offline", "answer": "[{\"word\":\"madison\",\"start\":3.11,\"end\":3.57}]"}
{"utterance_id": "5338-284437-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.82, 4.21, "sailor's"]], "response_mode": "offline", "answer": "[{\"word\":\"sailor's\",\"start\":3.82,\"end\":4.21}]"}
{"utterance_id": "2412-153954-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.5, 6.69, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":6.5,\"end\":6.69}]"}
{"utterance_id": "6313-66125-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.51, 3.8, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":3.51,\"end\":3.8}]"}
{"utterance_id": "1919-142785-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0010.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a food name (word, start, end). If none, reply [].", "labels": [[1.2, 1.59, "one"], [10.11, 10.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.2,\"end\":1.59},{\"word\":\"two\",\"start\":10.11,\"end\":10.45}]"}
{"utterance_id": "5895-34622-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0018.flac", "task_type": "lexical_detection", "query": "caravan", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"caravan\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.63, 1.19, "caravan"]], "response_mode": "offline", "answer": "[{\"word\":\"caravan\",\"start\":0.63,\"end\":1.19}]"}
{"utterance_id": "2277-149896-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0012.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.8, 2.23, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.8,\"end\":2.23}]"}
{"utterance_id": "3000-15664-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0003.flac", "task_type": "lexical_detection", "query": "alone", "prompt": "After the audio ends, extract all occurrences of the word \"alone\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.8, 2.45, "alone"]], "response_mode": "offline", "answer": "[{\"word\":\"alone\",\"start\":1.8,\"end\":2.45}]"}
{"utterance_id": "2902-9006-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.53, 4.2, "salvian"]], "response_mode": "offline", "answer": "[{\"word\":\"salvian\",\"start\":3.53,\"end\":4.2}]"}
{"utterance_id": "7976-110523-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0014.flac", "task_type": "lexical_detection", "query": "witch", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"witch\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.36, 4.66, "witch"]], "response_mode": "offline", "answer": "[{\"word\":\"witch\",\"start\":4.36,\"end\":4.66}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.48, 5.69, "two"], [7.3, 7.49, "two"], [11.21, 11.57, "six"], [15.56, 15.9, "couple"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":5.48,\"end\":5.69},{\"word\":\"two\",\"start\":7.3,\"end\":7.49},{\"word\":\"six\",\"start\":11.21,\"end\":11.57},{\"word\":\"couple\",\"start\":15.56,\"end\":15.9}]"}
{"utterance_id": "7976-110523-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0010.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "After the audio ends, tally every instance of a person's name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.46, 0.89, "one"], [5.59, 5.93, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.46,\"end\":0.89},{\"word\":\"two\",\"start\":5.59,\"end\":5.93}]"}
{"utterance_id": "652-129742-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0003.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.39, "cabbage salad"], [5.28, 5.75, "cabbage"]], "response_mode": "offline", "answer": "[{\"word\":\"cabbage salad\",\"start\":0.51,\"end\":1.39},{\"word\":\"cabbage\",\"start\":5.28,\"end\":5.75}]"}
{"utterance_id": "8297-275155-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0005.flac", "task_type": "lexical_detection", "query": "norman", "prompt": "Process the full audio offline. Find every instance of the word \"norman\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.58, "norman"]], "response_mode": "offline", "answer": "[{\"word\":\"norman\",\"start\":1.06,\"end\":1.58}]"}
{"utterance_id": "3081-166546-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0004.flac", "task_type": "lexical_detection", "query": "belong", "prompt": "Once the audio has finished, identify every instance of the word \"belong\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.75, 1.14, "belong"]], "response_mode": "offline", "answer": "[{\"word\":\"belong\",\"start\":0.75,\"end\":1.14}]"}
{"utterance_id": "1919-142785-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0026.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.07, "jamaica"]], "response_mode": "offline", "answer": "[{\"word\":\"jamaica\",\"start\":0.59,\"end\":1.07}]"}
{"utterance_id": "2086-149220-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[13.41, 13.72, "seven"]], "response_mode": "offline", "answer": "[{\"word\":\"seven\",\"start\":13.41,\"end\":13.72}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.87, 9.16, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":8.87,\"end\":9.16}]"}
{"utterance_id": "7850-73752-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0002.flac", "task_type": "lexical_detection", "query": "self", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"self\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.46, 1.79, "self"]], "response_mode": "offline", "answer": "[{\"word\":\"self\",\"start\":1.46,\"end\":1.79}]"}
{"utterance_id": "5694-64038-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0004.flac", "task_type": "lexical_detection", "query": "united", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"united\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.2, "united"]], "response_mode": "offline", "answer": "[{\"word\":\"united\",\"start\":0.84,\"end\":1.2}]"}
{"utterance_id": "3081-166546-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.82, "breakfast"]], "response_mode": "offline", "answer": "[{\"word\":\"breakfast\",\"start\":2.3,\"end\":2.82}]"}
{"utterance_id": "6241-61943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0024.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.04, 3.31, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":3.04,\"end\":3.31}]"}
{"utterance_id": "1919-142785-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0033.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen carefully to the entire clip, then report every instance of a color name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.85, 6.23, "brown"]], "response_mode": "offline", "answer": "[{\"word\":\"brown\",\"start\":5.85,\"end\":6.23}]"}
{"utterance_id": "6241-66616-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.02, 4.25, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":4.02,\"end\":4.25}]"}
{"utterance_id": "6345-93302-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0022.flac", "task_type": "lexical_detection", "query": "both", "prompt": "After the audio ends, extract all occurrences of the word \"both\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.03, "both"]], "response_mode": "offline", "answer": "[{\"word\":\"both\",\"start\":1.79,\"end\":2.03}]"}
{"utterance_id": "8842-304647-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0004.flac", "task_type": "semantic_detection", "query": "disease", "prompt": "After listening to the whole recording, output every instance of a disease name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.9, 7.48, "disease"]], "response_mode": "offline", "answer": "[{\"word\":\"disease\",\"start\":6.9,\"end\":7.48}]"}
{"utterance_id": "2277-149897-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0012.flac", "task_type": "lexical_detection", "query": "wouldn't", "prompt": "When the utterance is complete, report all instances of the word \"wouldn't\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.11, 2.41, "wouldn't"]], "response_mode": "offline", "answer": "[{\"word\":\"wouldn't\",\"start\":2.11,\"end\":2.41}]"}
{"utterance_id": "1993-147149-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.6, 12.38, "miss simmonds"]], "response_mode": "offline", "answer": "[{\"word\":\"miss simmonds\",\"start\":11.6,\"end\":12.38}]"}
{"utterance_id": "7976-105575-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.12, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":3.79,\"end\":4.12}]"}
{"utterance_id": "422-122949-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0010.flac", "task_type": "lexical_counting", "query": "masters", "prompt": "Process the full audio offline. For every occurrence of the word \"masters\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.37, 7.86, "one"], [11.61, 12.22, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.37,\"end\":7.86},{\"word\":\"two\",\"start\":11.61,\"end\":12.22}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.2, "groom's"]], "response_mode": "offline", "answer": "[{\"word\":\"groom's\",\"start\":1.82,\"end\":2.2}]"}
{"utterance_id": "1988-147956-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0007.flac", "task_type": "lexical_detection", "query": "loud", "prompt": "After the audio ends, extract all occurrences of the word \"loud\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 2.9, "loud"]], "response_mode": "offline", "answer": "[{\"word\":\"loud\",\"start\":2.52,\"end\":2.9}]"}
{"utterance_id": "2035-147960-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0003.flac", "task_type": "lexical_detection", "query": "frost", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"frost\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.59, 1.96, "frost"]], "response_mode": "offline", "answer": "[{\"word\":\"frost\",\"start\":1.59,\"end\":1.96}]"}
{"utterance_id": "2035-147960-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.64, "antonia"]], "response_mode": "offline", "answer": "[{\"word\":\"antonia\",\"start\":1.12,\"end\":1.64}]"}
{"utterance_id": "7850-281318-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0018.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.81, 4.51, "wood pigeon"]], "response_mode": "offline", "answer": "[{\"word\":\"wood pigeon\",\"start\":3.81,\"end\":4.51}]"}
{"utterance_id": "174-168635-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.59, 4.18, "cosette"]], "response_mode": "offline", "answer": "[{\"word\":\"cosette\",\"start\":3.59,\"end\":4.18}]"}
{"utterance_id": "1988-148538-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0002.flac", "task_type": "lexical_detection", "query": "americans", "prompt": "After the audio ends, extract all occurrences of the word \"americans\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.17, "americans"]], "response_mode": "offline", "answer": "[{\"word\":\"americans\",\"start\":1.61,\"end\":2.17}]"}
{"utterance_id": "1462-170142-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":0.97,\"end\":1.16}]"}
{"utterance_id": "1462-170145-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0004.flac", "task_type": "lexical_detection", "query": "yesterday", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"yesterday\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.56, "yesterday"]], "response_mode": "offline", "answer": "[{\"word\":\"yesterday\",\"start\":1.89,\"end\":2.56}]"}
{"utterance_id": "2412-153947-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0013.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.98, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.67,\"end\":0.98}]"}
{"utterance_id": "2803-161169-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0011.flac", "task_type": "lexical_detection", "query": "everybody", "prompt": "After the audio ends, extract all occurrences of the word \"everybody\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.07, 2.81, "everybody"]], "response_mode": "offline", "answer": "[{\"word\":\"everybody\",\"start\":2.07,\"end\":2.81}]"}
{"utterance_id": "652-130737-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0003.flac", "task_type": "lexical_detection", "query": "burgundy", "prompt": "After listening to the whole recording, output every instance of the word \"burgundy\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.69, "burgundy"]], "response_mode": "offline", "answer": "[{\"word\":\"burgundy\",\"start\":1.05,\"end\":1.69}]"}
{"utterance_id": "2428-83705-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0039.flac", "task_type": "lexical_detection", "query": "wedding", "prompt": "Once the audio has finished, identify every instance of the word \"wedding\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.62, 3.05, "wedding"]], "response_mode": "offline", "answer": "[{\"word\":\"wedding\",\"start\":2.62,\"end\":3.05}]"}
{"utterance_id": "652-130726-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0016.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.89, 3.2, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.89,\"end\":3.2}]"}
{"utterance_id": "5895-34629-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0019.flac", "task_type": "lexical_detection", "query": "bridge", "prompt": "Once the audio has finished, identify every instance of the word \"bridge\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.86, 3.29, "bridge"]], "response_mode": "offline", "answer": "[{\"word\":\"bridge\",\"start\":2.86,\"end\":3.29}]"}
{"utterance_id": "1993-147149-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0005.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[1.42, 1.75, "one"], [4.44, 4.92, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.42,\"end\":1.75},{\"word\":\"two\",\"start\":4.44,\"end\":4.92}]"}
{"utterance_id": "7976-105575-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0001.flac", "task_type": "lexical_detection", "query": "advance", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"advance\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.55, "advance"]], "response_mode": "offline", "answer": "[{\"word\":\"advance\",\"start\":1.83,\"end\":2.55}]"}
{"utterance_id": "1988-24833-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0011.flac", "task_type": "lexical_detection", "query": "alike", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"alike\" with fields word, start, end (seconds). If none, reply [].", "labels": [[5.46, 5.97, "alike"]], "response_mode": "offline", "answer": "[{\"word\":\"alike\",\"start\":5.46,\"end\":5.97}]"}
{"utterance_id": "7976-105575-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0029.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.89, 8.33, "bodies"]], "response_mode": "offline", "answer": "[{\"word\":\"bodies\",\"start\":7.89,\"end\":8.33}]"}
{"utterance_id": "7976-110523-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0002.flac", "task_type": "lexical_detection", "query": "hunger", "prompt": "After the audio ends, extract all occurrences of the word \"hunger\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.57, "hunger"]], "response_mode": "offline", "answer": "[{\"word\":\"hunger\",\"start\":3.97,\"end\":4.57}]"}
{"utterance_id": "1919-142785-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0007.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.69, 3.4, "cucumbers"], [5.13, 5.64, "seeds"], [8.33, 9.14, "cabbage leaf"], [20.67, 21.3, "cabbage leaf"]], "response_mode": "offline", "answer": "[{\"word\":\"cucumbers\",\"start\":2.69,\"end\":3.4},{\"word\":\"seeds\",\"start\":5.13,\"end\":5.64},{\"word\":\"cabbage leaf\",\"start\":8.33,\"end\":9.14},{\"word\":\"cabbage leaf\",\"start\":20.67,\"end\":21.3}]"}
{"utterance_id": "1272-135031-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.9, 2.38, "shaggy"]], "response_mode": "offline", "answer": "[{\"word\":\"shaggy\",\"start\":1.9,\"end\":2.38}]"}
{"utterance_id": "1272-141231-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.29, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":1.79,\"end\":2.29}]"}
{"utterance_id": "3536-23268-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0024.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.77, 4.16, "servant"]], "response_mode": "offline", "answer": "[{\"word\":\"servant\",\"start\":3.77,\"end\":4.16}]"}
{"utterance_id": "777-126732-0078", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0078.flac", "task_type": "lexical_detection", "query": "lot", "prompt": "Once the audio has finished, identify every instance of the word \"lot\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.84, 2.22, "lot"]], "response_mode": "offline", "answer": "[{\"word\":\"lot\",\"start\":1.84,\"end\":2.22}]"}
{"utterance_id": "84-121550-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0003.flac", "task_type": "lexical_detection", "query": "custom", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"custom\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.21, 7.73, "custom"]], "response_mode": "offline", "answer": "[{\"word\":\"custom\",\"start\":7.21,\"end\":7.73}]"}
{"utterance_id": "6319-57405-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.53, 5.87, "son"]], "response_mode": "offline", "answer": "[{\"word\":\"son\",\"start\":5.53,\"end\":5.87}]"}
{"utterance_id": "2412-153948-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0005.flac", "task_type": "lexical_detection", "query": "delighted", "prompt": "Once the audio has finished, identify every instance of the word \"delighted\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.77, 1.23, "delighted"]], "response_mode": "offline", "answer": "[{\"word\":\"delighted\",\"start\":0.77,\"end\":1.23}]"}
{"utterance_id": "2803-154328-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0002.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.63, 1.02, "meal"]], "response_mode": "offline", "answer": "[{\"word\":\"meal\",\"start\":0.63,\"end\":1.02}]"}
{"utterance_id": "6241-66616-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0003.flac", "task_type": "lexical_detection", "query": "driven", "prompt": "Once the audio has finished, identify every instance of the word \"driven\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.8, 3.09, "driven"]], "response_mode": "offline", "answer": "[{\"word\":\"driven\",\"start\":2.8,\"end\":3.09}]"}
{"utterance_id": "5338-284437-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0010.flac", "task_type": "semantic_counting", "query": "color", "prompt": "After listening to the whole recording, output timed count words for a color name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.06, 2.49, "one"], [3.71, 4.1, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.06,\"end\":2.49},{\"word\":\"two\",\"start\":3.71,\"end\":4.1}]"}
{"utterance_id": "2902-9006-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0006.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.6, 3.12, "empire"]], "response_mode": "offline", "answer": "[{\"word\":\"empire\",\"start\":2.6,\"end\":3.12}]"}
{"utterance_id": "1988-24833-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.43, 0.72, "pop"]], "response_mode": "offline", "answer": "[{\"word\":\"pop\",\"start\":0.43,\"end\":0.72}]"}
{"utterance_id": "3000-15664-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.45, 1.62, "thousand thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"thousand thousand\",\"start\":0.45,\"end\":1.62}]"}
{"utterance_id": "3000-15664-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0023.flac", "task_type": "lexical_detection", "query": "stronghold", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"stronghold\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.95, 2.68, "stronghold"]], "response_mode": "offline", "answer": "[{\"word\":\"stronghold\",\"start\":1.95,\"end\":2.68}]"}
{"utterance_id": "3576-138058-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0040.flac", "task_type": "lexical_detection", "query": "dared", "prompt": "After the audio ends, extract all occurrences of the word \"dared\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.08, 3.41, "dared"]], "response_mode": "offline", "answer": "[{\"word\":\"dared\",\"start\":3.08,\"end\":3.41}]"}
{"utterance_id": "652-130737-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0010.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.18, 2.76, "bordeaux"]], "response_mode": "offline", "answer": "[{\"word\":\"bordeaux\",\"start\":2.18,\"end\":2.76}]"}
{"utterance_id": "5895-34615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.4, 4.72, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":4.4,\"end\":4.72}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a country name with fields word, start, end (seconds). If none, reply [].", "labels": [[3.49, 3.85, "german"]], "response_mode": "offline", "answer": "[{\"word\":\"german\",\"start\":3.49,\"end\":3.85}]"}
{"utterance_id": "1988-24833-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0010.flac", "task_type": "lexical_detection", "query": "cares", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cares\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.78, "cares"]], "response_mode": "offline", "answer": "[{\"word\":\"cares\",\"start\":1.21,\"end\":1.78}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.21, "jordan"]], "response_mode": "offline", "answer": "[{\"word\":\"jordan\",\"start\":1.81,\"end\":2.21}]"}
{"utterance_id": "3536-8226-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0005.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a family relation: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[3.42, 3.84, "one"], [8.94, 9.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.42,\"end\":3.84},{\"word\":\"two\",\"start\":8.94,\"end\":9.38}]"}
{"utterance_id": "5338-24615-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0000.flac", "task_type": "lexical_detection", "query": "noon", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"noon\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.34, "noon"]], "response_mode": "offline", "answer": "[{\"word\":\"noon\",\"start\":1.03,\"end\":1.34}]"}
{"utterance_id": "5895-34629-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0032.flac", "task_type": "lexical_detection", "query": "famous", "prompt": "After the audio ends, extract all occurrences of the word \"famous\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.66, "famous"]], "response_mode": "offline", "answer": "[{\"word\":\"famous\",\"start\":1.17,\"end\":1.66}]"}
{"utterance_id": "7850-286674-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0003.flac", "task_type": "lexical_detection", "query": "plenty", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"plenty\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.28, 2.88, "plenty"]], "response_mode": "offline", "answer": "[{\"word\":\"plenty\",\"start\":2.28,\"end\":2.88}]"}
{"utterance_id": "1272-141231-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.65, 3.97, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":3.65,\"end\":3.97}]"}
{"utterance_id": "652-129742-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0011.flac", "task_type": "lexical_detection", "query": "chill", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"chill\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.69, 2.07, "chill"]], "response_mode": "offline", "answer": "[{\"word\":\"chill\",\"start\":1.69,\"end\":2.07}]"}
{"utterance_id": "7976-110124-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0003.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.89, 1.51, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":0.89,\"end\":1.51}]"}
{"utterance_id": "2902-9008-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0012.flac", "task_type": "lexical_detection", "query": "content", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"content\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.04, 2.5, "content"]], "response_mode": "offline", "answer": "[{\"word\":\"content\",\"start\":2.04,\"end\":2.5}]"}
{"utterance_id": "777-126732-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0055.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.8, "animal"]], "response_mode": "offline", "answer": "[{\"word\":\"animal\",\"start\":2.44,\"end\":2.8}]"}
{"utterance_id": "5338-284437-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0004.flac", "task_type": "lexical_detection", "query": "residences", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"residences\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.74, "residences"]], "response_mode": "offline", "answer": "[{\"word\":\"residences\",\"start\":2.87,\"end\":3.74}]"}
{"utterance_id": "422-122949-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0026.flac", "task_type": "lexical_counting", "query": "instinct", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"instinct\" (word, start, end). If none, reply [].", "labels": [[0.77, 1.28, "one"], [11.34, 11.87, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.77,\"end\":1.28},{\"word\":\"two\",\"start\":11.34,\"end\":11.87}]"}
{"utterance_id": "251-137823-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0009.flac", "task_type": "lexical_detection", "query": "tom's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"tom's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.83, "tom's"]], "response_mode": "offline", "answer": "[{\"word\":\"tom's\",\"start\":0.3,\"end\":0.83}]"}
{"utterance_id": "6345-64257-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0003.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "After listening to the whole recording, output every instance of a weather term with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.21, 6.67, "wind"]], "response_mode": "offline", "answer": "[{\"word\":\"wind\",\"start\":6.21,\"end\":6.67}]"}
{"utterance_id": "3170-137482-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0029.flac", "task_type": "lexical_detection", "query": "truth", "prompt": "Once the audio has finished, identify every instance of the word \"truth\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.92, 2.49, "truth"]], "response_mode": "offline", "answer": "[{\"word\":\"truth\",\"start\":1.92,\"end\":2.49}]"}
{"utterance_id": "6295-244435-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0024.flac", "task_type": "lexical_detection", "query": "world", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"world\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.65, "world"]], "response_mode": "offline", "answer": "[{\"word\":\"world\",\"start\":1.1,\"end\":1.65}]"}
{"utterance_id": "5536-43358-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0016.flac", "task_type": "lexical_detection", "query": "superstition", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"superstition\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.49, 2.27, "superstition"]], "response_mode": "offline", "answer": "[{\"word\":\"superstition\",\"start\":1.49,\"end\":2.27}]"}
{"utterance_id": "6241-66616-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0009.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.38, 7.84, "lynx"]], "response_mode": "offline", "answer": "[{\"word\":\"lynx\",\"start\":7.38,\"end\":7.84}]"}
{"utterance_id": "6345-93302-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0020.flac", "task_type": "lexical_detection", "query": "couldn't", "prompt": "Once the audio has finished, identify every instance of the word \"couldn't\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.25, 0.65, "couldn't"]], "response_mode": "offline", "answer": "[{\"word\":\"couldn't\",\"start\":0.25,\"end\":0.65}]"}
{"utterance_id": "174-168635-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.3, 2.88, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":2.3,\"end\":2.88}]"}
{"utterance_id": "1993-147965-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0000.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.17, 3.5, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":3.17,\"end\":3.5}]"}
{"utterance_id": "3000-15664-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.4, 4.8, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":4.4,\"end\":4.8}]"}
{"utterance_id": "174-168635-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.12,\"end\":1.36}]"}
{"utterance_id": "3752-4944-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0013.flac", "task_type": "lexical_detection", "query": "bishop", "prompt": "After the audio ends, extract all occurrences of the word \"bishop\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.02, 5.5, "bishop"]], "response_mode": "offline", "answer": "[{\"word\":\"bishop\",\"start\":5.02,\"end\":5.5}]"}
{"utterance_id": "2428-83699-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0041.flac", "task_type": "lexical_detection", "query": "i'm", "prompt": "Once the audio has finished, identify every instance of the word \"i'm\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.18, 0.39, "i'm"]], "response_mode": "offline", "answer": "[{\"word\":\"i'm\",\"start\":0.18,\"end\":0.39}]"}
{"utterance_id": "2086-149220-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0033.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.09, 4.44, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":4.09,\"end\":4.44}]"}
{"utterance_id": "5895-34622-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0015.flac", "task_type": "lexical_detection", "query": "venus", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"venus\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.53, 1.87, "venus"]], "response_mode": "offline", "answer": "[{\"word\":\"venus\",\"start\":1.53,\"end\":1.87}]"}
{"utterance_id": "251-137823-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0026.flac", "task_type": "lexical_detection", "query": "twinkled", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"twinkled\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.58, "twinkled"]], "response_mode": "offline", "answer": "[{\"word\":\"twinkled\",\"start\":2.01,\"end\":2.58}]"}
{"utterance_id": "5338-24615-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.7, 10.54, "mister dent's"]], "response_mode": "offline", "answer": "[{\"word\":\"mister dent's\",\"start\":9.7,\"end\":10.54}]"}
{"utterance_id": "5694-64025-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0011.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.71, 2.06, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.71,\"end\":2.06}]"}
{"utterance_id": "6313-66129-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0001.flac", "task_type": "lexical_detection", "query": "dashed", "prompt": "When the utterance is complete, report all instances of the word \"dashed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 1.92, "dashed"]], "response_mode": "offline", "answer": "[{\"word\":\"dashed\",\"start\":1.56,\"end\":1.92}]"}
{"utterance_id": "84-121550-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0013.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[14.9, 15.44, "colours"]], "response_mode": "offline", "answer": "[{\"word\":\"colours\",\"start\":14.9,\"end\":15.44}]"}
{"utterance_id": "2078-142845-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0015.flac", "task_type": "lexical_detection", "query": "rice", "prompt": "Process the full audio offline. Find every instance of the word \"rice\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.34, "rice"], [6.8, 7.42, "rice"]], "response_mode": "offline", "answer": "[{\"word\":\"rice\",\"start\":1.89,\"end\":2.34},{\"word\":\"rice\",\"start\":6.8,\"end\":7.42}]"}
{"utterance_id": "1988-147956-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.53, 7.05, "grandmother"]], "response_mode": "offline", "answer": "[{\"word\":\"grandmother\",\"start\":6.53,\"end\":7.05}]"}
{"utterance_id": "3752-4944-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.17, 1.46, "meekin"]], "response_mode": "offline", "answer": "[{\"word\":\"meekin\",\"start\":1.17,\"end\":1.46}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "lexical_detection", "query": "saturday", "prompt": "After the audio ends, extract all occurrences of the word \"saturday\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.07, 1.72, "saturday"]], "response_mode": "offline", "answer": "[{\"word\":\"saturday\",\"start\":1.07,\"end\":1.72}]"}
{"utterance_id": "6319-64726-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0011.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.37, 5.76, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":5.37,\"end\":5.76}]"}
{"utterance_id": "2277-149896-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0025.flac", "task_type": "lexical_detection", "query": "arose", "prompt": "After the audio ends, extract all occurrences of the word \"arose\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.02, "arose"]], "response_mode": "offline", "answer": "[{\"word\":\"arose\",\"start\":0.7,\"end\":1.02}]"}
{"utterance_id": "6295-64301-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0019.flac", "task_type": "lexical_detection", "query": "eternity", "prompt": "Process the full audio offline. Find every instance of the word \"eternity\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.43, "eternity"]], "response_mode": "offline", "answer": "[{\"word\":\"eternity\",\"start\":2.74,\"end\":3.43}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.19, 0.41, "one"], [2.41, 2.55, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.19,\"end\":0.41},{\"word\":\"one\",\"start\":2.41,\"end\":2.55}]"}
{"utterance_id": "2902-9008-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0006.flac", "task_type": "lexical_detection", "query": "welcomed", "prompt": "After listening to the whole recording, output every instance of the word \"welcomed\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 1.16, "welcomed"]], "response_mode": "offline", "answer": "[{\"word\":\"welcomed\",\"start\":0.58,\"end\":1.16}]"}
{"utterance_id": "2277-149874-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0018.flac", "task_type": "lexical_detection", "query": "circumstances", "prompt": "After listening to the whole recording, output every instance of the word \"circumstances\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.51, "circumstances"]], "response_mode": "offline", "answer": "[{\"word\":\"circumstances\",\"start\":1.69,\"end\":2.51}]"}
{"utterance_id": "3853-163249-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0048.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.43, 2.87, "helen"], [4.27, 4.66, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"helen\",\"start\":2.43,\"end\":2.87},{\"word\":\"david\",\"start\":4.27,\"end\":4.66}]"}
{"utterance_id": "5895-34629-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0021.flac", "task_type": "lexical_detection", "query": "sold", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sold\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.35, 4.76, "sold"]], "response_mode": "offline", "answer": "[{\"word\":\"sold\",\"start\":4.35,\"end\":4.76}]"}
{"utterance_id": "2078-142845-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0034.flac", "task_type": "lexical_detection", "query": "crumpets", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"crumpets\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.99, 1.53, "crumpets"]], "response_mode": "offline", "answer": "[{\"word\":\"crumpets\",\"start\":0.99,\"end\":1.53}]"}
{"utterance_id": "2277-149874-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0015.flac", "task_type": "lexical_detection", "query": "relations", "prompt": "Once the audio has finished, identify every instance of the word \"relations\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.11, 2.66, "relations"]], "response_mode": "offline", "answer": "[{\"word\":\"relations\",\"start\":2.11,\"end\":2.66}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the entire clip, then report the count sequence for a family relation as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.17, "one"], [7.27, 7.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.93,\"end\":3.17},{\"word\":\"two\",\"start\":7.27,\"end\":7.68}]"}
{"utterance_id": "1272-141231-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.24, "muscles"]], "response_mode": "offline", "answer": "[{\"word\":\"muscles\",\"start\":2.72,\"end\":3.24}]"}
{"utterance_id": "2902-9006-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0014.flac", "task_type": "lexical_detection", "query": "sons", "prompt": "Once the audio has finished, identify every instance of the word \"sons\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.74, 3.14, "sons"]], "response_mode": "offline", "answer": "[{\"word\":\"sons\",\"start\":2.74,\"end\":3.14}]"}
{"utterance_id": "3000-15664-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0041.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[20.97, 21.87, "sugar pine"]], "response_mode": "offline", "answer": "[{\"word\":\"sugar pine\",\"start\":20.97,\"end\":21.87}]"}
{"utterance_id": "3081-166546-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0023.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.08, 4.55, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":4.08,\"end\":4.55}]"}
{"utterance_id": "2035-147961-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0032.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.61, 0.94, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.61,\"end\":0.94}]"}
{"utterance_id": "2803-161169-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0012.flac", "task_type": "lexical_detection", "query": "palm", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"palm\" with fields word, start, end (seconds). If none, reply [].", "labels": [[12.55, 13.05, "palm"]], "response_mode": "offline", "answer": "[{\"word\":\"palm\",\"start\":12.55,\"end\":13.05}]"}
{"utterance_id": "3081-166546-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.49, "miss challoner"]], "response_mode": "offline", "answer": "[{\"word\":\"miss challoner\",\"start\":1.76,\"end\":2.49}]"}
{"utterance_id": "1919-142785-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0036.flac", "task_type": "lexical_detection", "query": "graceful", "prompt": "Process the full audio offline. Find every instance of the word \"graceful\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.36, 4.8, "graceful"]], "response_mode": "offline", "answer": "[{\"word\":\"graceful\",\"start\":4.36,\"end\":4.8}]"}
{"utterance_id": "6313-76958-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0000.flac", "task_type": "lexical_detection", "query": "chapter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"chapter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.17, 0.5, "chapter"]], "response_mode": "offline", "answer": "[{\"word\":\"chapter\",\"start\":0.17,\"end\":0.5}]"}
{"utterance_id": "2035-147961-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0012.flac", "task_type": "lexical_detection", "query": "breath", "prompt": "Process the full audio offline. Find every instance of the word \"breath\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.72, 2.1, "breath"]], "response_mode": "offline", "answer": "[{\"word\":\"breath\",\"start\":1.72,\"end\":2.1}]"}
{"utterance_id": "1919-142785-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0057.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the full audio. Count every occurrence of a food name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.56, 0.88, "one"], [2.91, 3.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.56,\"end\":0.88},{\"word\":\"two\",\"start\":2.91,\"end\":3.31}]"}
{"utterance_id": "6295-64301-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0018.flac", "task_type": "lexical_counting", "query": "mood", "prompt": "Listen to the full audio. Count every occurrence of the word \"mood\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.26, "one"], [5.22, 5.7, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.96,\"end\":1.26},{\"word\":\"two\",\"start\":5.22,\"end\":5.7}]"}
{"utterance_id": "3081-166546-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0052.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.4, 2.07, "musicians"]], "response_mode": "offline", "answer": "[{\"word\":\"musicians\",\"start\":1.4,\"end\":2.07}]"}
{"utterance_id": "1462-170142-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.65, 1.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.65,\"end\":1.88}]"}
{"utterance_id": "3081-166546-0080", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0080.flac", "task_type": "lexical_detection", "query": "going", "prompt": "When the utterance is complete, report all instances of the word \"going\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.79, 3.03, "going"]], "response_mode": "offline", "answer": "[{\"word\":\"going\",\"start\":2.79,\"end\":3.03}]"}
{"utterance_id": "1673-143397-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0020.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.47, 2.11, "blemmyes"]], "response_mode": "offline", "answer": "[{\"word\":\"blemmyes\",\"start\":1.47,\"end\":2.11}]"}
{"utterance_id": "2086-149220-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0041.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[5.47, 5.61, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.47,\"end\":5.61}]"}
{"utterance_id": "6295-244435-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0012.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.24, 0.75, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"europe\",\"start\":0.24,\"end\":0.75}]"}
{"utterance_id": "652-129742-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0016.flac", "task_type": "lexical_detection", "query": "tomato", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"tomato\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 0.88, "tomato"], [3.13, 3.48, "tomato"]], "response_mode": "offline", "answer": "[{\"word\":\"tomato\",\"start\":0.48,\"end\":0.88},{\"word\":\"tomato\",\"start\":3.13,\"end\":3.48}]"}
{"utterance_id": "2902-9008-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0003.flac", "task_type": "lexical_detection", "query": "gardens", "prompt": "Process the full audio offline. Find every instance of the word \"gardens\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 5.98, "gardens"]], "response_mode": "offline", "answer": "[{\"word\":\"gardens\",\"start\":5.52,\"end\":5.98}]"}
{"utterance_id": "3170-137482-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0044.flac", "task_type": "lexical_detection", "query": "phenomenon", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"phenomenon\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.74, 12.31, "phenomenon"]], "response_mode": "offline", "answer": "[{\"word\":\"phenomenon\",\"start\":11.74,\"end\":12.31}]"}
{"utterance_id": "1462-170145-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0011.flac", "task_type": "lexical_detection", "query": "coming", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"coming\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.66, 2.13, "coming"]], "response_mode": "offline", "answer": "[{\"word\":\"coming\",\"start\":1.66,\"end\":2.13}]"}
{"utterance_id": "2086-149220-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0034.flac", "task_type": "lexical_detection", "query": "stern", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"stern\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.53, 4.86, "stern"]], "response_mode": "offline", "answer": "[{\"word\":\"stern\",\"start\":4.53,\"end\":4.86}]"}
{"utterance_id": "3081-166546-0065", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0065.flac", "task_type": "lexical_detection", "query": "altogether", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"altogether\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.36, 2.04, "altogether"]], "response_mode": "offline", "answer": "[{\"word\":\"altogether\",\"start\":1.36,\"end\":2.04}]"}
{"utterance_id": "3170-137482-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0001.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.2, 17.6, "army"]], "response_mode": "offline", "answer": "[{\"word\":\"army\",\"start\":17.2,\"end\":17.6}]"}
{"utterance_id": "3752-4944-0055", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0055.flac", "task_type": "lexical_detection", "query": "murderous", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"murderous\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.17, 2.01, "murderous"]], "response_mode": "offline", "answer": "[{\"word\":\"murderous\",\"start\":1.17,\"end\":2.01}]"}
{"utterance_id": "5694-64029-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0014.flac", "task_type": "lexical_detection", "query": "bad", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"bad\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.31, 0.75, "bad"]], "response_mode": "offline", "answer": "[{\"word\":\"bad\",\"start\":0.31,\"end\":0.75}]"}
{"utterance_id": "3576-138058-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0032.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Do not answer until the audio ends. Then list each instance of a money or currency term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.45, 5.2, "reals"]], "response_mode": "offline", "answer": "[{\"word\":\"reals\",\"start\":4.45,\"end\":5.2}]"}
{"utterance_id": "6241-66616-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.46, 2.01, "wabigoon's"]], "response_mode": "offline", "answer": "[{\"word\":\"wabigoon's\",\"start\":1.46,\"end\":2.01}]"}
{"utterance_id": "6313-76958-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0029.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.84, "pong"]], "response_mode": "offline", "answer": "[{\"word\":\"pong\",\"start\":0.41,\"end\":0.84}]"}
{"utterance_id": "8842-302201-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0001.flac", "task_type": "lexical_detection", "query": "having", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"having\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.44, 4.67, "having"]], "response_mode": "offline", "answer": "[{\"word\":\"having\",\"start\":4.44,\"end\":4.67}]"}
{"utterance_id": "7976-105575-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.61, 3.23, "six hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"six hundred\",\"start\":2.61,\"end\":3.23}]"}
{"utterance_id": "1272-141231-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.09, 1.5, "chest"], [5.09, 5.66, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"chest\",\"start\":1.09,\"end\":1.5},{\"word\":\"eyes\",\"start\":5.09,\"end\":5.66}]"}
{"utterance_id": "6345-93302-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0008.flac", "task_type": "lexical_detection", "query": "excuse", "prompt": "When the utterance is complete, report all instances of the word \"excuse\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.85, "excuse"]], "response_mode": "offline", "answer": "[{\"word\":\"excuse\",\"start\":2.19,\"end\":2.85}]"}
{"utterance_id": "3752-4944-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0007.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 2.42, "sarah island"]], "response_mode": "offline", "answer": "[{\"word\":\"sarah island\",\"start\":1.47,\"end\":2.42}]"}
{"utterance_id": "174-168635-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.84, "cosette"]], "response_mode": "offline", "answer": "[{\"word\":\"cosette\",\"start\":3.26,\"end\":3.84}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "After listening to the whole recording, output timed count words for a profession (one, two, three, ...) as JSON. If none, reply [].", "labels": [[1.25, 1.54, "one"], [4.28, 4.59, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.25,\"end\":1.54},{\"word\":\"two\",\"start\":4.28,\"end\":4.59}]"}
{"utterance_id": "1919-142785-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0041.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.39, 3.02, "ingredients"], [7.3, 7.74, "egg"]], "response_mode": "offline", "answer": "[{\"word\":\"ingredients\",\"start\":2.39,\"end\":3.02},{\"word\":\"egg\",\"start\":7.3,\"end\":7.74}]"}
{"utterance_id": "6295-244435-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 2.26, "president davis"]], "response_mode": "offline", "answer": "[{\"word\":\"president davis\",\"start\":1.27,\"end\":2.26}]"}
{"utterance_id": "2803-154328-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0012.flac", "task_type": "lexical_detection", "query": "possible", "prompt": "After the audio ends, extract all occurrences of the word \"possible\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.62, 5.3, "possible"]], "response_mode": "offline", "answer": "[{\"word\":\"possible\",\"start\":4.62,\"end\":5.3}]"}
{"utterance_id": "1988-147956-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0018.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After the audio ends, extract all occurrences of a color name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.06, 1.49, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.06,\"end\":1.49}]"}
{"utterance_id": "6319-64726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.64, 3.13, "daughter"]], "response_mode": "offline", "answer": "[{\"word\":\"daughter\",\"start\":2.64,\"end\":3.13}]"}
{"utterance_id": "1919-142785-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0041.flac", "task_type": "semantic_counting", "query": "food", "prompt": "After listening to the whole recording, output timed count words for a food name (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.39, 3.02, "one"], [7.3, 7.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.39,\"end\":3.02},{\"word\":\"two\",\"start\":7.3,\"end\":7.74}]"}
{"utterance_id": "6319-275224-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0010.flac", "task_type": "lexical_detection", "query": "fearful", "prompt": "When the utterance is complete, report all instances of the word \"fearful\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.47, 4.0, "fearful"]], "response_mode": "offline", "answer": "[{\"word\":\"fearful\",\"start\":3.47,\"end\":4.0}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "semantic_counting", "query": "weather", "prompt": "Listen to the entire clip, then report the count sequence for a weather term as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.16, "one"], [5.09, 5.48, "two"], [7.58, 7.91, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.73,\"end\":2.16},{\"word\":\"two\",\"start\":5.09,\"end\":5.48},{\"word\":\"three\",\"start\":7.58,\"end\":7.91}]"}
{"utterance_id": "3081-166546-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0015.flac", "task_type": "lexical_detection", "query": "looking", "prompt": "When the utterance is complete, report all instances of the word \"looking\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.34, "looking"]], "response_mode": "offline", "answer": "[{\"word\":\"looking\",\"start\":0.88,\"end\":1.34}]"}
{"utterance_id": "6319-64726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0007.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "When the utterance is complete, report how a number accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.81, "one"], [12.9, 13.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.81},{\"word\":\"two\",\"start\":12.9,\"end\":13.13}]"}
{"utterance_id": "2078-142845-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0006.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.03, 6.43, "yeast"], [8.11, 8.64, "flour"], [13.84, 14.35, "dough"]], "response_mode": "offline", "answer": "[{\"word\":\"yeast\",\"start\":6.03,\"end\":6.43},{\"word\":\"flour\",\"start\":8.11,\"end\":8.64},{\"word\":\"dough\",\"start\":13.84,\"end\":14.35}]"}
{"utterance_id": "84-121550-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0023.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.51, 4.82, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":4.51,\"end\":4.82}]"}
{"utterance_id": "777-126732-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0053.flac", "task_type": "lexical_detection", "query": "kitchen", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"kitchen\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.09, 3.57, "kitchen"]], "response_mode": "offline", "answer": "[{\"word\":\"kitchen\",\"start\":3.09,\"end\":3.57}]"}
{"utterance_id": "7850-73752-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.97, "ferdinand"]], "response_mode": "offline", "answer": "[{\"word\":\"ferdinand\",\"start\":0.51,\"end\":0.97}]"}
{"utterance_id": "7850-111771-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0007.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.95, 5.43, "one"], [7.33, 7.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.95,\"end\":5.43},{\"word\":\"two\",\"start\":7.33,\"end\":7.73}]"}
{"utterance_id": "6295-64301-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0011.flac", "task_type": "lexical_detection", "query": "equal", "prompt": "After listening to the whole recording, output every instance of the word \"equal\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.54, "equal"]], "response_mode": "offline", "answer": "[{\"word\":\"equal\",\"start\":2.16,\"end\":2.54}]"}
{"utterance_id": "1673-143396-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.73, 13.27, "three hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"three hundred\",\"start\":12.73,\"end\":13.27}]"}
{"utterance_id": "8842-304647-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/304647/8842-304647-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.23, 6.66, "vein"]], "response_mode": "offline", "answer": "[{\"word\":\"vein\",\"start\":6.23,\"end\":6.66}]"}
{"utterance_id": "3853-163249-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0038.flac", "task_type": "lexical_detection", "query": "visits", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"visits\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.84, 3.23, "visits"]], "response_mode": "offline", "answer": "[{\"word\":\"visits\",\"start\":2.84,\"end\":3.23}]"}
{"utterance_id": "5694-64025-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0005.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After the audio ends, extract all occurrences of a date. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.0, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":0.7,\"end\":1.0}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.95, 11.27, "foot"]], "response_mode": "offline", "answer": "[{\"word\":\"foot\",\"start\":10.95,\"end\":11.27}]"}
{"utterance_id": "8297-275155-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0003.flac", "task_type": "lexical_detection", "query": "randal", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"randal\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.39, 0.76, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":0.39,\"end\":0.76}]"}
{"utterance_id": "251-137823-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0021.flac", "task_type": "lexical_counting", "query": "swift", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"swift\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.85, 1.3, "one"], [4.72, 5.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.85,\"end\":1.3},{\"word\":\"two\",\"start\":4.72,\"end\":5.13}]"}
{"utterance_id": "6313-66129-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0012.flac", "task_type": "lexical_detection", "query": "filled", "prompt": "When the utterance is complete, report all instances of the word \"filled\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.47, "filled"]], "response_mode": "offline", "answer": "[{\"word\":\"filled\",\"start\":1.17,\"end\":1.47}]"}
{"utterance_id": "1272-135031-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.51, 0.9, "kaliko"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":0.51,\"end\":0.9}]"}
{"utterance_id": "5694-64029-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0028.flac", "task_type": "lexical_detection", "query": "struggle", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"struggle\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.22, 4.67, "struggle"]], "response_mode": "offline", "answer": "[{\"word\":\"struggle\",\"start\":4.22,\"end\":4.67}]"}
{"utterance_id": "2277-149897-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0020.flac", "task_type": "lexical_detection", "query": "helpless", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"helpless\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 0.82, "helpless"]], "response_mode": "offline", "answer": "[{\"word\":\"helpless\",\"start\":0.37,\"end\":0.82}]"}
{"utterance_id": "251-118436-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0022.flac", "task_type": "lexical_detection", "query": "obey", "prompt": "Once the audio has finished, identify every instance of the word \"obey\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.47, 3.01, "obey"]], "response_mode": "offline", "answer": "[{\"word\":\"obey\",\"start\":2.47,\"end\":3.01}]"}
{"utterance_id": "3752-4944-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0066.flac", "task_type": "lexical_detection", "query": "twice", "prompt": "Once the audio has finished, identify every instance of the word \"twice\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.26, 0.69, "twice"], [3.35, 3.86, "twice"]], "response_mode": "offline", "answer": "[{\"word\":\"twice\",\"start\":0.26,\"end\":0.69},{\"word\":\"twice\",\"start\":3.35,\"end\":3.86}]"}
{"utterance_id": "5338-24615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0005.flac", "task_type": "lexical_detection", "query": "straight", "prompt": "After listening to the whole recording, output every instance of the word \"straight\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.33, 1.84, "straight"]], "response_mode": "offline", "answer": "[{\"word\":\"straight\",\"start\":1.33,\"end\":1.84}]"}
{"utterance_id": "8297-275154-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.73, 9.27, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":8.73,\"end\":9.27}]"}
{"utterance_id": "5694-64029-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0024.flac", "task_type": "lexical_counting", "query": "country", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"country\" (word, start, end). If none, reply [].", "labels": [[4.27, 4.74, "one"], [5.74, 6.27, "two"], [8.23, 8.81, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.27,\"end\":4.74},{\"word\":\"two\",\"start\":5.74,\"end\":6.27},{\"word\":\"three\",\"start\":8.23,\"end\":8.81}]"}
{"utterance_id": "6313-76958-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0016.flac", "task_type": "lexical_detection", "query": "wake", "prompt": "When the utterance is complete, report all instances of the word \"wake\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.66, "wake"]], "response_mode": "offline", "answer": "[{\"word\":\"wake\",\"start\":1.39,\"end\":1.66}]"}
{"utterance_id": "6241-61943-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0027.flac", "task_type": "lexical_detection", "query": "limits", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"limits\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[16.06, 16.44, "limits"]], "response_mode": "offline", "answer": "[{\"word\":\"limits\",\"start\":16.06,\"end\":16.44}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "lexical_detection", "query": "triumphant", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"triumphant\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.93, "triumphant"]], "response_mode": "offline", "answer": "[{\"word\":\"triumphant\",\"start\":1.31,\"end\":1.93}]"}
{"utterance_id": "5694-64038-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.76, 3.21, "johnny"]], "response_mode": "offline", "answer": "[{\"word\":\"johnny\",\"start\":2.76,\"end\":3.21}]"}
{"utterance_id": "6241-61946-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.1, 4.39, "legs"]], "response_mode": "offline", "answer": "[{\"word\":\"legs\",\"start\":4.1,\"end\":4.39}]"}
{"utterance_id": "5694-64025-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0020.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Process the full audio offline. For every occurrence of a body part, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.11, 5.45, "one"], [6.45, 7.01, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.11,\"end\":5.45},{\"word\":\"two\",\"start\":6.45,\"end\":7.01}]"}
{"utterance_id": "7976-110124-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0014.flac", "task_type": "lexical_detection", "query": "powder", "prompt": "Once the audio has finished, identify every instance of the word \"powder\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.98, 3.29, "powder"]], "response_mode": "offline", "answer": "[{\"word\":\"powder\",\"start\":2.98,\"end\":3.29}]"}
{"utterance_id": "1988-147956-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0000.flac", "task_type": "lexical_detection", "query": "fuchs", "prompt": "Once the audio has finished, identify every instance of the word \"fuchs\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.53, 0.94, "fuchs"]], "response_mode": "offline", "answer": "[{\"word\":\"fuchs\",\"start\":0.53,\"end\":0.94}]"}
{"utterance_id": "777-126732-0054", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0054.flac", "task_type": "lexical_detection", "query": "watched", "prompt": "Process the full audio offline. Find every instance of the word \"watched\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.88, "watched"]], "response_mode": "offline", "answer": "[{\"word\":\"watched\",\"start\":0.46,\"end\":0.88}]"}
{"utterance_id": "5895-34615-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0021.flac", "task_type": "lexical_detection", "query": "operation", "prompt": "Once the audio has finished, identify every instance of the word \"operation\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.78, 1.52, "operation"]], "response_mode": "offline", "answer": "[{\"word\":\"operation\",\"start\":0.78,\"end\":1.52}]"}
{"utterance_id": "6295-64301-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.82, 5.35, "bosom"]], "response_mode": "offline", "answer": "[{\"word\":\"bosom\",\"start\":4.82,\"end\":5.35}]"}
{"utterance_id": "2078-142845-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0013.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.73, 6.43, "puddings"]], "response_mode": "offline", "answer": "[{\"word\":\"puddings\",\"start\":5.73,\"end\":6.43}]"}
{"utterance_id": "1272-141231-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0023.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.71, 5.89, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":5.71,\"end\":5.89}]"}
{"utterance_id": "1919-142785-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0018.flac", "task_type": "semantic_detection", "query": "country", "prompt": "You will hear a complete utterance. When it ends, list every instance of a country name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.12, 5.48, "europe"]], "response_mode": "offline", "answer": "[{\"word\":\"europe\",\"start\":5.12,\"end\":5.48}]"}
{"utterance_id": "1272-141231-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.83, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.83}]"}
{"utterance_id": "2035-147961-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0040.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Once the audio has finished, identify every instance of a weather term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.62, 3.13, "snows"]], "response_mode": "offline", "answer": "[{\"word\":\"snows\",\"start\":2.62,\"end\":3.13}]"}
{"utterance_id": "2035-147961-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.64, "spirits"]], "response_mode": "offline", "answer": "[{\"word\":\"spirits\",\"start\":1.11,\"end\":1.64}]"}
{"utterance_id": "3081-166546-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0062.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.12, 5.34, "ten"]], "response_mode": "offline", "answer": "[{\"word\":\"ten\",\"start\":5.12,\"end\":5.34}]"}
{"utterance_id": "2078-142845-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0043.flac", "task_type": "lexical_counting", "query": "stir", "prompt": "After listening to the whole recording, output timed count words for the word \"stir\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.79, 3.18, "one"], [9.58, 10.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.79,\"end\":3.18},{\"word\":\"two\",\"start\":9.58,\"end\":10.0}]"}
{"utterance_id": "5694-64025-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0022.flac", "task_type": "lexical_counting", "query": "rope", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"rope\" with word/start/end. If none, reply [].", "labels": [[0.49, 0.78, "one"], [5.73, 6.18, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.49,\"end\":0.78},{\"word\":\"two\",\"start\":5.73,\"end\":6.18}]"}
{"utterance_id": "3853-163249-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0015.flac", "task_type": "lexical_detection", "query": "terrible", "prompt": "Once the audio has finished, identify every instance of the word \"terrible\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.58, 1.15, "terrible"]], "response_mode": "offline", "answer": "[{\"word\":\"terrible\",\"start\":0.58,\"end\":1.15}]"}
{"utterance_id": "5338-24615-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0005.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Listen to the full audio. Count every occurrence of a plant name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.96, 7.15, "one"], [8.87, 9.56, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.96,\"end\":7.15},{\"word\":\"two\",\"start\":8.87,\"end\":9.56}]"}
{"utterance_id": "2277-149897-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0032.flac", "task_type": "lexical_detection", "query": "watchman", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"watchman\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.43, "watchman"]], "response_mode": "offline", "answer": "[{\"word\":\"watchman\",\"start\":4.99,\"end\":5.43}]"}
{"utterance_id": "2035-152373-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0013.flac", "task_type": "lexical_detection", "query": "exiles", "prompt": "When the utterance is complete, report all instances of the word \"exiles\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.32, 1.89, "exiles"]], "response_mode": "offline", "answer": "[{\"word\":\"exiles\",\"start\":1.32,\"end\":1.89}]"}
{"utterance_id": "6295-244435-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0018.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Once the audio has finished, identify every instance of a time of day. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.63, 2.15, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.63,\"end\":2.15}]"}
{"utterance_id": "2277-149874-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0010.flac", "task_type": "lexical_detection", "query": "walking", "prompt": "Process the full audio offline. Find every instance of the word \"walking\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 2.07, "walking"]], "response_mode": "offline", "answer": "[{\"word\":\"walking\",\"start\":1.56,\"end\":2.07}]"}
{"utterance_id": "2428-83705-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0000.flac", "task_type": "lexical_detection", "query": "remarkable", "prompt": "After the audio ends, extract all occurrences of the word \"remarkable\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.91, "remarkable"]], "response_mode": "offline", "answer": "[{\"word\":\"remarkable\",\"start\":1.35,\"end\":1.91}]"}
{"utterance_id": "652-129742-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.62, "asparagus salad"], [2.89, 3.59, "asparagus"]], "response_mode": "offline", "answer": "[{\"word\":\"asparagus salad\",\"start\":0.51,\"end\":1.62},{\"word\":\"asparagus\",\"start\":2.89,\"end\":3.59}]"}
{"utterance_id": "5895-34615-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0017.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 1.99, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":1.48,\"end\":1.99}]"}
{"utterance_id": "8297-275156-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.7, 9.22, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.7,\"end\":9.22}]"}
{"utterance_id": "5536-43363-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.19, 7.43, "hair"]], "response_mode": "offline", "answer": "[{\"word\":\"hair\",\"start\":7.19,\"end\":7.43}]"}
{"utterance_id": "5338-24615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.36, 5.91, "grandam"]], "response_mode": "offline", "answer": "[{\"word\":\"grandam\",\"start\":5.36,\"end\":5.91}]"}
{"utterance_id": "3536-8226-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.17, "trevelyan"]], "response_mode": "offline", "answer": "[{\"word\":\"trevelyan\",\"start\":0.53,\"end\":1.17}]"}
{"utterance_id": "5338-284437-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 0.96, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.62,\"end\":0.96}]"}
{"utterance_id": "5895-34615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.4, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.72,\"end\":1.4}]"}
{"utterance_id": "251-137823-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0017.flac", "task_type": "lexical_detection", "query": "help", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"help\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.88, "help"]], "response_mode": "offline", "answer": "[{\"word\":\"help\",\"start\":1.51,\"end\":1.88}]"}
{"utterance_id": "777-126732-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0039.flac", "task_type": "lexical_detection", "query": "call", "prompt": "Process the full audio offline. Find every instance of the word \"call\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.58, "call"]], "response_mode": "offline", "answer": "[{\"word\":\"call\",\"start\":0.36,\"end\":0.58}]"}
{"utterance_id": "3081-166546-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0029.flac", "task_type": "lexical_detection", "query": "suggestion", "prompt": "After the audio ends, extract all occurrences of the word \"suggestion\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.43, 1.18, "suggestion"]], "response_mode": "offline", "answer": "[{\"word\":\"suggestion\",\"start\":0.43,\"end\":1.18}]"}
{"utterance_id": "1993-147149-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0021.flac", "task_type": "lexical_detection", "query": "stairs", "prompt": "Process the full audio offline. Find every instance of the word \"stairs\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.58, "stairs"]], "response_mode": "offline", "answer": "[{\"word\":\"stairs\",\"start\":1.17,\"end\":1.58}]"}
{"utterance_id": "3853-163249-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0011.flac", "task_type": "lexical_detection", "query": "wait", "prompt": "After listening to the whole recording, output every instance of the word \"wait\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.87, 1.08, "wait"]], "response_mode": "offline", "answer": "[{\"word\":\"wait\",\"start\":0.87,\"end\":1.08}]"}
{"utterance_id": "1272-128104-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0002.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Once the audio has finished, identify every instance of a date. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.43, 5.03, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":4.43,\"end\":5.03}]"}
{"utterance_id": "174-168635-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0005.flac", "task_type": "lexical_detection", "query": "cosette", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cosette\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.11, "cosette"]], "response_mode": "offline", "answer": "[{\"word\":\"cosette\",\"start\":0.5,\"end\":1.11}]"}
{"utterance_id": "3853-163249-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0046.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.99, 6.33, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":5.99,\"end\":6.33}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "lexical_detection", "query": "lawn", "prompt": "After the audio ends, extract all occurrences of the word \"lawn\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.99, 10.41, "lawn"]], "response_mode": "offline", "answer": "[{\"word\":\"lawn\",\"start\":9.99,\"end\":10.41}]"}
{"utterance_id": "3081-166546-0089", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0089.flac", "task_type": "lexical_detection", "query": "flew", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"flew\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.43, 4.82, "flew"]], "response_mode": "offline", "answer": "[{\"word\":\"flew\",\"start\":4.43,\"end\":4.82}]"}
{"utterance_id": "2035-147961-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0014.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[2.78, 3.1, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"body\",\"start\":2.78,\"end\":3.1}]"}
{"utterance_id": "5536-43363-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0004.flac", "task_type": "lexical_detection", "query": "finest", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"finest\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.46, "finest"]], "response_mode": "offline", "answer": "[{\"word\":\"finest\",\"start\":2.06,\"end\":2.46}]"}
{"utterance_id": "6313-76958-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0019.flac", "task_type": "lexical_detection", "query": "leg", "prompt": "Process the full audio offline. Find every instance of the word \"leg\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.82, "leg"]], "response_mode": "offline", "answer": "[{\"word\":\"leg\",\"start\":1.54,\"end\":1.82}]"}
{"utterance_id": "1272-135031-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0019.flac", "task_type": "lexical_detection", "query": "remarked", "prompt": "After the audio ends, extract all occurrences of the word \"remarked\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.84, "remarked"]], "response_mode": "offline", "answer": "[{\"word\":\"remarked\",\"start\":1.41,\"end\":1.84}]"}
{"utterance_id": "1988-147956-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0022.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.72, "cottonwood tree"]], "response_mode": "offline", "answer": "[{\"word\":\"cottonwood tree\",\"start\":1.83,\"end\":2.72}]"}
{"utterance_id": "3752-4944-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0011.flac", "task_type": "lexical_detection", "query": "heard", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"heard\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.05, "heard"]], "response_mode": "offline", "answer": "[{\"word\":\"heard\",\"start\":1.84,\"end\":2.05}]"}
{"utterance_id": "5895-34615-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0014.flac", "task_type": "lexical_detection", "query": "exception", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"exception\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.05, 1.77, "exception"]], "response_mode": "offline", "answer": "[{\"word\":\"exception\",\"start\":1.05,\"end\":1.77}]"}
{"utterance_id": "7850-73752-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0013.flac", "task_type": "lexical_detection", "query": "dancing", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"dancing\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.98, 1.43, "dancing"]], "response_mode": "offline", "answer": "[{\"word\":\"dancing\",\"start\":0.98,\"end\":1.43}]"}
{"utterance_id": "3170-137482-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0021.flac", "task_type": "lexical_detection", "query": "rubbed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"rubbed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 0.78, "rubbed"]], "response_mode": "offline", "answer": "[{\"word\":\"rubbed\",\"start\":0.47,\"end\":0.78}]"}
{"utterance_id": "3536-8226-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0006.flac", "task_type": "lexical_detection", "query": "distinct", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"distinct\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.38, 0.86, "distinct"]], "response_mode": "offline", "answer": "[{\"word\":\"distinct\",\"start\":0.38,\"end\":0.86}]"}
{"utterance_id": "2428-83699-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0011.flac", "task_type": "lexical_detection", "query": "night", "prompt": "Once the audio has finished, identify every instance of the word \"night\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 0.88, "night"]], "response_mode": "offline", "answer": "[{\"word\":\"night\",\"start\":0.52,\"end\":0.88}]"}
{"utterance_id": "2412-153948-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0007.flac", "task_type": "lexical_detection", "query": "lamb", "prompt": "Once the audio has finished, identify every instance of the word \"lamb\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.47, 4.74, "lamb"]], "response_mode": "offline", "answer": "[{\"word\":\"lamb\",\"start\":4.47,\"end\":4.74}]"}
{"utterance_id": "5895-34629-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0024.flac", "task_type": "lexical_detection", "query": "success", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"success\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.09, 3.78, "success"]], "response_mode": "offline", "answer": "[{\"word\":\"success\",\"start\":3.09,\"end\":3.78}]"}
{"utterance_id": "5338-284437-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.02, 2.13, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.02,\"end\":2.13}]"}
{"utterance_id": "3081-166546-0078", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0078.flac", "task_type": "lexical_detection", "query": "natural", "prompt": "Once the audio has finished, identify every instance of the word \"natural\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.74, 2.4, "natural"]], "response_mode": "offline", "answer": "[{\"word\":\"natural\",\"start\":1.74,\"end\":2.4}]"}
{"utterance_id": "6345-64257-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0006.flac", "task_type": "lexical_detection", "query": "colors", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"colors\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.55, 8.03, "colors"]], "response_mode": "offline", "answer": "[{\"word\":\"colors\",\"start\":7.55,\"end\":8.03}]"}
{"utterance_id": "6241-61943-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0021.flac", "task_type": "lexical_detection", "query": "glad", "prompt": "After listening to the whole recording, output every instance of the word \"glad\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.17, "glad"]], "response_mode": "offline", "answer": "[{\"word\":\"glad\",\"start\":0.83,\"end\":1.17}]"}
{"utterance_id": "2078-142845-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0029.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.66, 2.34, "toast"]], "response_mode": "offline", "answer": "[{\"word\":\"toast\",\"start\":1.66,\"end\":2.34}]"}
{"utterance_id": "2078-142845-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0021.flac", "task_type": "lexical_detection", "query": "sour", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sour\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.2, 0.66, "sour"]], "response_mode": "offline", "answer": "[{\"word\":\"sour\",\"start\":0.2,\"end\":0.66}]"}
{"utterance_id": "84-121123-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0018.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.13, 0.55, "morrel"]], "response_mode": "offline", "answer": "[{\"word\":\"morrel\",\"start\":0.13,\"end\":0.55}]"}
{"utterance_id": "2035-152373-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0013.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen to the full audio. After it finishes, find every instance of a state or province name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.58, "meath"]], "response_mode": "offline", "answer": "[{\"word\":\"meath\",\"start\":3.1,\"end\":3.58}]"}
{"utterance_id": "2428-83705-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 2.35, "mary ann"]], "response_mode": "offline", "answer": "[{\"word\":\"mary ann\",\"start\":1.71,\"end\":2.35}]"}
{"utterance_id": "1988-147956-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0026.flac", "task_type": "semantic_counting", "query": "color", "prompt": "After the audio ends, tally every instance of a color name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.01, 3.27, "one"], [4.97, 5.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.01,\"end\":3.27},{\"word\":\"two\",\"start\":4.97,\"end\":5.31}]"}
{"utterance_id": "174-50561-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0013.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.32, 9.65, "gold"]], "response_mode": "offline", "answer": "[{\"word\":\"gold\",\"start\":9.32,\"end\":9.65}]"}
{"utterance_id": "6313-66125-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.56, 1.99, "lige"]], "response_mode": "offline", "answer": "[{\"word\":\"lige\",\"start\":1.56,\"end\":1.99}]"}
{"utterance_id": "3576-138058-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0019.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.36, 4.76, "knight"]], "response_mode": "offline", "answer": "[{\"word\":\"knight\",\"start\":4.36,\"end\":4.76}]"}
{"utterance_id": "7850-111771-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.95, 5.43, "grant"], [7.33, 7.73, "lee"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":4.95,\"end\":5.43},{\"word\":\"lee\",\"start\":7.33,\"end\":7.73}]"}
{"utterance_id": "1993-147149-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0019.flac", "task_type": "lexical_detection", "query": "please", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"please\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.2, 1.62, "please"]], "response_mode": "offline", "answer": "[{\"word\":\"please\",\"start\":1.2,\"end\":1.62}]"}
{"utterance_id": "1919-142785-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0010.flac", "task_type": "lexical_counting", "query": "salt", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"salt\" (word, start, end). If none, reply [].", "labels": [[1.2, 1.59, "one"], [10.11, 10.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.2,\"end\":1.59},{\"word\":\"two\",\"start\":10.11,\"end\":10.45}]"}
{"utterance_id": "5536-43363-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0007.flac", "task_type": "lexical_detection", "query": "sex", "prompt": "After listening to the whole recording, output every instance of the word \"sex\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.66, 4.97, "sex"]], "response_mode": "offline", "answer": "[{\"word\":\"sex\",\"start\":4.66,\"end\":4.97}]"}
{"utterance_id": "7850-286674-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0005.flac", "task_type": "lexical_detection", "query": "bodies", "prompt": "Process the full audio offline. Find every instance of the word \"bodies\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.85, 7.42, "bodies"]], "response_mode": "offline", "answer": "[{\"word\":\"bodies\",\"start\":6.85,\"end\":7.42}]"}
{"utterance_id": "2277-149896-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0001.flac", "task_type": "lexical_detection", "query": "regularly", "prompt": "After the audio ends, extract all occurrences of the word \"regularly\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.33, 2.87, "regularly"]], "response_mode": "offline", "answer": "[{\"word\":\"regularly\",\"start\":2.33,\"end\":2.87}]"}
{"utterance_id": "5895-34629-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0016.flac", "task_type": "lexical_detection", "query": "success", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"success\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.8, 4.44, "success"]], "response_mode": "offline", "answer": "[{\"word\":\"success\",\"start\":3.8,\"end\":4.44}]"}
{"utterance_id": "1988-24833-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0011.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.48, 3.88, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":3.48,\"end\":3.88}]"}
{"utterance_id": "3576-138058-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0013.flac", "task_type": "lexical_detection", "query": "men's", "prompt": "Once the audio has finished, identify every instance of the word \"men's\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.05, 3.32, "men's"]], "response_mode": "offline", "answer": "[{\"word\":\"men's\",\"start\":3.05,\"end\":3.32}]"}
{"utterance_id": "6241-66616-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0007.flac", "task_type": "lexical_detection", "query": "minnetaki", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"minnetaki\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.13, 4.83, "minnetaki"]], "response_mode": "offline", "answer": "[{\"word\":\"minnetaki\",\"start\":4.13,\"end\":4.83}]"}
{"utterance_id": "1993-147964-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0006.flac", "task_type": "lexical_detection", "query": "jake", "prompt": "After the audio ends, extract all occurrences of the word \"jake\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.73, 3.15, "jake"]], "response_mode": "offline", "answer": "[{\"word\":\"jake\",\"start\":2.73,\"end\":3.15}]"}
{"utterance_id": "6295-64301-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.39, 4.82, "mary's"]], "response_mode": "offline", "answer": "[{\"word\":\"mary's\",\"start\":4.39,\"end\":4.82}]"}
{"utterance_id": "84-121550-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0017.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the full audio. Count every occurrence of a number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.17, 0.66, "one"], [4.25, 4.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.17,\"end\":0.66},{\"word\":\"two\",\"start\":4.25,\"end\":4.73}]"}
{"utterance_id": "2277-149874-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0009.flac", "task_type": "lexical_detection", "query": "wrapped", "prompt": "After the audio ends, extract all occurrences of the word \"wrapped\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 1.97, "wrapped"]], "response_mode": "offline", "answer": "[{\"word\":\"wrapped\",\"start\":1.65,\"end\":1.97}]"}
{"utterance_id": "8297-275154-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.89, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.51,\"end\":1.89}]"}
{"utterance_id": "6345-93306-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0024.flac", "task_type": "semantic_counting", "query": "color", "prompt": "When the utterance is complete, report how a color name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.8, "one"], [6.67, 7.0, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.36,\"end\":0.8},{\"word\":\"two\",\"start\":6.67,\"end\":7.0}]"}
{"utterance_id": "3752-4944-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.94, 2.39, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":1.94,\"end\":2.39}]"}
{"utterance_id": "5895-34615-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0017.flac", "task_type": "lexical_detection", "query": "brought", "prompt": "After the audio ends, extract all occurrences of the word \"brought\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.36, "brought"]], "response_mode": "offline", "answer": "[{\"word\":\"brought\",\"start\":3.11,\"end\":3.36}]"}
{"utterance_id": "5694-64038-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0024.flac", "task_type": "lexical_counting", "query": "regiment", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"regiment\" with word/start/end. If none, reply [].", "labels": [[0.42, 0.82, "one"], [2.82, 3.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.42,\"end\":0.82},{\"word\":\"two\",\"start\":2.82,\"end\":3.29}]"}
{"utterance_id": "2428-83699-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0027.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[2.7, 2.93, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.7,\"end\":2.93}]"}
{"utterance_id": "3752-4943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.76, 3.4, "mister north"]], "response_mode": "offline", "answer": "[{\"word\":\"mister north\",\"start\":2.76,\"end\":3.4}]"}
{"utterance_id": "7976-105575-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0009.flac", "task_type": "lexical_detection", "query": "wounded", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"wounded\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.21, 2.65, "wounded"]], "response_mode": "offline", "answer": "[{\"word\":\"wounded\",\"start\":2.21,\"end\":2.65}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[19.8, 20.06, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":19.8,\"end\":20.06}]"}
{"utterance_id": "6295-64301-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0008.flac", "task_type": "lexical_detection", "query": "joseph", "prompt": "After listening to the whole recording, output every instance of the word \"joseph\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.46, "joseph"]], "response_mode": "offline", "answer": "[{\"word\":\"joseph\",\"start\":1.82,\"end\":2.46}]"}
{"utterance_id": "2086-149220-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0004.flac", "task_type": "lexical_detection", "query": "wondered", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"wondered\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.26, "wondered"]], "response_mode": "offline", "answer": "[{\"word\":\"wondered\",\"start\":0.82,\"end\":1.26}]"}
{"utterance_id": "652-129742-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0005.flac", "task_type": "lexical_counting", "query": "until", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"until\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.53, 0.8, "one"], [4.04, 4.37, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.53,\"end\":0.8},{\"word\":\"two\",\"start\":4.04,\"end\":4.37}]"}
{"utterance_id": "2803-154328-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0007.flac", "task_type": "lexical_counting", "query": "promised", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"promised\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.95, 2.41, "one"], [3.5, 3.98, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.95,\"end\":2.41},{\"word\":\"two\",\"start\":3.5,\"end\":3.98}]"}
{"utterance_id": "2803-154320-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0002.flac", "task_type": "lexical_detection", "query": "friends", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"friends\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.33, "friends"]], "response_mode": "offline", "answer": "[{\"word\":\"friends\",\"start\":0.7,\"end\":1.33}]"}
{"utterance_id": "5694-64025-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0013.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After listening to the whole recording, output every instance of a date with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.27, "monday"]], "response_mode": "offline", "answer": "[{\"word\":\"monday\",\"start\":0.72,\"end\":1.27}]"}
{"utterance_id": "5536-43359-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an ordinal number with fields word, start, end (seconds). If none, reply [].", "labels": [[5.29, 5.6, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.29,\"end\":5.6}]"}
{"utterance_id": "6241-66616-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0003.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.83, 2.24, "woonga"]], "response_mode": "offline", "answer": "[{\"word\":\"woonga\",\"start\":1.83,\"end\":2.24}]"}
{"utterance_id": "6345-64257-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.19, "ear"], [4.1, 4.63, "brain"]], "response_mode": "offline", "answer": "[{\"word\":\"ear\",\"start\":1.91,\"end\":2.19},{\"word\":\"brain\",\"start\":4.1,\"end\":4.63}]"}
{"utterance_id": "3170-137482-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0033.flac", "task_type": "lexical_detection", "query": "three", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"three\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.65, 0.85, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.65,\"end\":0.85}]"}
{"utterance_id": "1919-142785-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0017.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.63, 1.0, "lemon"]], "response_mode": "offline", "answer": "[{\"word\":\"lemon\",\"start\":0.63,\"end\":1.0}]"}
{"utterance_id": "3576-138058-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0014.flac", "task_type": "lexical_detection", "query": "favourable", "prompt": "When the utterance is complete, report all instances of the word \"favourable\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[15.42, 16.04, "favourable"]], "response_mode": "offline", "answer": "[{\"word\":\"favourable\",\"start\":15.42,\"end\":16.04}]"}
{"utterance_id": "84-121550-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0033.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[4.72, 5.08, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"mouth\",\"start\":4.72,\"end\":5.08}]"}
{"utterance_id": "5536-43363-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0007.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Process the full audio offline. For every occurrence of a food name, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.95, 2.24, "one"], [9.07, 9.46, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.95,\"end\":2.24},{\"word\":\"two\",\"start\":9.07,\"end\":9.46}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 1.01, "board"]], "response_mode": "offline", "answer": "[{\"word\":\"board\",\"start\":0.61,\"end\":1.01}]"}
{"utterance_id": "6313-76958-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0012.flac", "task_type": "lexical_detection", "query": "proved", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"proved\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.63, "proved"]], "response_mode": "offline", "answer": "[{\"word\":\"proved\",\"start\":1.16,\"end\":1.63}]"}
{"utterance_id": "2035-147961-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0013.flac", "task_type": "lexical_detection", "query": "bony", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"bony\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.73, 3.06, "bony"]], "response_mode": "offline", "answer": "[{\"word\":\"bony\",\"start\":2.73,\"end\":3.06}]"}
{"utterance_id": "3752-4944-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0030.flac", "task_type": "lexical_detection", "query": "mean", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mean\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.63, 0.95, "mean"]], "response_mode": "offline", "answer": "[{\"word\":\"mean\",\"start\":0.63,\"end\":0.95}]"}
{"utterance_id": "3081-166546-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0030.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.77, 4.2, "face"], [7.36, 7.69, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":3.77,\"end\":4.2},{\"word\":\"eye\",\"start\":7.36,\"end\":7.69}]"}
{"utterance_id": "7850-286674-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0008.flac", "task_type": "lexical_detection", "query": "upper", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"upper\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.57, "upper"]], "response_mode": "offline", "answer": "[{\"word\":\"upper\",\"start\":0.38,\"end\":0.57}]"}
{"utterance_id": "6241-66616-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.47, 0.71, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":0.47,\"end\":0.71}]"}
{"utterance_id": "6295-244435-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0035.flac", "task_type": "lexical_detection", "query": "bill", "prompt": "When the utterance is complete, report all instances of the word \"bill\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.73, "bill"]], "response_mode": "offline", "answer": "[{\"word\":\"bill\",\"start\":0.51,\"end\":0.73}]"}
{"utterance_id": "3752-4944-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0048.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.66, 1.16, "minister"]], "response_mode": "offline", "answer": "[{\"word\":\"minister\",\"start\":0.66,\"end\":1.16}]"}
{"utterance_id": "1993-147964-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0000.flac", "task_type": "lexical_detection", "query": "sunday", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"sunday\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.97, 3.48, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":2.97,\"end\":3.48}]"}
{"utterance_id": "8842-302203-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0008.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, produce a running count of an ordinal number: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.3, 0.59, "one"], [2.92, 3.2, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.3,\"end\":0.59},{\"word\":\"two\",\"start\":2.92,\"end\":3.2}]"}
{"utterance_id": "174-84280-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0010.flac", "task_type": "lexical_detection", "query": "sister", "prompt": "When the utterance is complete, report all instances of the word \"sister\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.48, 2.03, "sister"]], "response_mode": "offline", "answer": "[{\"word\":\"sister\",\"start\":1.48,\"end\":2.03}]"}
{"utterance_id": "1462-170142-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 1.84, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":1.35,\"end\":1.84}]"}
{"utterance_id": "6295-64301-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0029.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 2.88, "neck"], [4.37, 4.82, "body"]], "response_mode": "offline", "answer": "[{\"word\":\"neck\",\"start\":2.55,\"end\":2.88},{\"word\":\"body\",\"start\":4.37,\"end\":4.82}]"}
{"utterance_id": "6241-61943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0024.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.58, 6.01, "pastor"]], "response_mode": "offline", "answer": "[{\"word\":\"pastor\",\"start\":5.58,\"end\":6.01}]"}
{"utterance_id": "3576-138058-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0011.flac", "task_type": "lexical_detection", "query": "quixote", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"quixote\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.36, 9.8, "quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"quixote\",\"start\":9.36,\"end\":9.8}]"}
{"utterance_id": "3576-138058-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0029.flac", "task_type": "lexical_detection", "query": "claudia", "prompt": "Once the audio has finished, identify every instance of the word \"claudia\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.36, 0.85, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":0.36,\"end\":0.85}]"}
{"utterance_id": "2277-149897-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.44, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":1.03,\"end\":1.44}]"}
{"utterance_id": "5338-284437-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a body part with fields word, start, end (seconds). If none, reply [].", "labels": [[5.89, 6.4, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":5.89,\"end\":6.4}]"}
{"utterance_id": "3853-163249-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0025.flac", "task_type": "lexical_detection", "query": "frenzy", "prompt": "After listening to the whole recording, output every instance of the word \"frenzy\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.29, "frenzy"]], "response_mode": "offline", "answer": "[{\"word\":\"frenzy\",\"start\":0.8,\"end\":1.29}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "lexical_detection", "query": "northern", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"northern\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[17.33, 17.72, "northern"], [31.1, 31.44, "northern"]], "response_mode": "offline", "answer": "[{\"word\":\"northern\",\"start\":17.33,\"end\":17.72},{\"word\":\"northern\",\"start\":31.1,\"end\":31.44}]"}
{"utterance_id": "2078-142845-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0012.flac", "task_type": "lexical_detection", "query": "puddings", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"puddings\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.82, 8.58, "puddings"]], "response_mode": "offline", "answer": "[{\"word\":\"puddings\",\"start\":7.82,\"end\":8.58}]"}
{"utterance_id": "251-118436-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0015.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.77, 2.29, "silver"]], "response_mode": "offline", "answer": "[{\"word\":\"silver\",\"start\":1.77,\"end\":2.29}]"}
{"utterance_id": "251-136532-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0023.flac", "task_type": "lexical_detection", "query": "society", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"society\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.87, "society"]], "response_mode": "offline", "answer": "[{\"word\":\"society\",\"start\":1.26,\"end\":1.87}]"}
{"utterance_id": "6295-244435-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.52, "harry"]], "response_mode": "offline", "answer": "[{\"word\":\"harry\",\"start\":1.11,\"end\":1.52}]"}
{"utterance_id": "6241-66616-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0012.flac", "task_type": "lexical_detection", "query": "veins", "prompt": "When the utterance is complete, report all instances of the word \"veins\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.01, 10.47, "veins"]], "response_mode": "offline", "answer": "[{\"word\":\"veins\",\"start\":10.01,\"end\":10.47}]"}
{"utterance_id": "777-126732-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0013.flac", "task_type": "lexical_detection", "query": "anger", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"anger\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 6.03, "anger"]], "response_mode": "offline", "answer": "[{\"word\":\"anger\",\"start\":5.55,\"end\":6.03}]"}
{"utterance_id": "6241-61946-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0004.flac", "task_type": "lexical_detection", "query": "meadows", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"meadows\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.47, "meadows"]], "response_mode": "offline", "answer": "[{\"word\":\"meadows\",\"start\":2.0,\"end\":2.47}]"}
{"utterance_id": "1988-24833-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0009.flac", "task_type": "lexical_detection", "query": "huh", "prompt": "After the audio ends, extract all occurrences of the word \"huh\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.95, 6.23, "huh"]], "response_mode": "offline", "answer": "[{\"word\":\"huh\",\"start\":5.95,\"end\":6.23}]"}
{"utterance_id": "6241-66616-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0021.flac", "task_type": "lexical_detection", "query": "winter", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"winter\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.46, 1.87, "winter"]], "response_mode": "offline", "answer": "[{\"word\":\"winter\",\"start\":1.46,\"end\":1.87}]"}
{"utterance_id": "1988-148538-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.81, 1.45, "sixteen"]], "response_mode": "offline", "answer": "[{\"word\":\"sixteen\",\"start\":0.81,\"end\":1.45}]"}
{"utterance_id": "7976-110124-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0008.flac", "task_type": "lexical_detection", "query": "decide", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"decide\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.05, "decide"]], "response_mode": "offline", "answer": "[{\"word\":\"decide\",\"start\":0.67,\"end\":1.05}]"}
{"utterance_id": "6241-66616-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0013.flac", "task_type": "lexical_detection", "query": "daring", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"daring\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.48, 3.84, "daring"]], "response_mode": "offline", "answer": "[{\"word\":\"daring\",\"start\":3.48,\"end\":3.84}]"}
{"utterance_id": "1919-142785-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0042.flac", "task_type": "lexical_detection", "query": "pike", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pike\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.85, 3.35, "pike"]], "response_mode": "offline", "answer": "[{\"word\":\"pike\",\"start\":2.85,\"end\":3.35}]"}
{"utterance_id": "6345-93302-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.98, "hands"], [4.44, 4.78, "face"], [7.69, 8.22, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":0.52,\"end\":0.98},{\"word\":\"face\",\"start\":4.44,\"end\":4.78},{\"word\":\"arms\",\"start\":7.69,\"end\":8.22}]"}
{"utterance_id": "2412-153954-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0000.flac", "task_type": "lexical_detection", "query": "shortly", "prompt": "Once the audio has finished, identify every instance of the word \"shortly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.44, 0.9, "shortly"]], "response_mode": "offline", "answer": "[{\"word\":\"shortly\",\"start\":0.44,\"end\":0.9}]"}
{"utterance_id": "3576-138058-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0034.flac", "task_type": "lexical_detection", "query": "thirst", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"thirst\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.72, 3.22, "thirst"]], "response_mode": "offline", "answer": "[{\"word\":\"thirst\",\"start\":2.72,\"end\":3.22}]"}
{"utterance_id": "777-126732-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.3, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":1.91,\"end\":2.3}]"}
{"utterance_id": "84-121123-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0002.flac", "task_type": "lexical_counting", "query": "became", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"became\" with word/start/end. If none, reply [].", "labels": [[4.19, 4.62, "one"], [8.06, 8.39, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.19,\"end\":4.62},{\"word\":\"two\",\"start\":8.06,\"end\":8.39}]"}
{"utterance_id": "5895-34629-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0019.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.48, 2.86, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":2.48,\"end\":2.86}]"}
{"utterance_id": "2277-149896-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0016.flac", "task_type": "lexical_detection", "query": "fortunately", "prompt": "Process the full audio offline. Find every instance of the word \"fortunately\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.25, 0.8, "fortunately"]], "response_mode": "offline", "answer": "[{\"word\":\"fortunately\",\"start\":0.25,\"end\":0.8}]"}
{"utterance_id": "5338-284437-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0013.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.4, 2.92, "one"], [4.25, 4.99, "two"], [8.6, 9.14, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.4,\"end\":2.92},{\"word\":\"two\",\"start\":4.25,\"end\":4.99},{\"word\":\"three\",\"start\":8.6,\"end\":9.14}]"}
{"utterance_id": "7976-110124-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0012.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.51, 1.87, "beggar"]], "response_mode": "offline", "answer": "[{\"word\":\"beggar\",\"start\":1.51,\"end\":1.87}]"}
{"utterance_id": "1988-147956-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0004.flac", "task_type": "lexical_detection", "query": "thatched", "prompt": "After the audio ends, extract all occurrences of the word \"thatched\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.22, 5.79, "thatched"]], "response_mode": "offline", "answer": "[{\"word\":\"thatched\",\"start\":5.22,\"end\":5.79}]"}
{"utterance_id": "174-168635-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0007.flac", "task_type": "lexical_counting", "query": "thought", "prompt": "After the audio ends, tally every instance of the word \"thought\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.81, 6.07, "one"], [8.39, 8.72, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.81,\"end\":6.07},{\"word\":\"two\",\"start\":8.39,\"end\":8.72}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.51, 9.04, "features"], [20.78, 21.25, "features"]], "response_mode": "offline", "answer": "[{\"word\":\"features\",\"start\":8.51,\"end\":9.04},{\"word\":\"features\",\"start\":20.78,\"end\":21.25}]"}
{"utterance_id": "1993-147149-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0009.flac", "task_type": "lexical_counting", "query": "dead", "prompt": "Once the audio has finished, list each occurrence of the word \"dead\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[6.09, 6.44, "one"], [9.31, 9.58, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.09,\"end\":6.44},{\"word\":\"two\",\"start\":9.31,\"end\":9.58}]"}
{"utterance_id": "8297-275155-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0031.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.49, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":2.19,\"end\":2.49}]"}
{"utterance_id": "7850-73752-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0003.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[18.67, 19.47, "charles street"]], "response_mode": "offline", "answer": "[{\"word\":\"charles street\",\"start\":18.67,\"end\":19.47}]"}
{"utterance_id": "1462-170142-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.94, "hilda's"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda's\",\"start\":0.52,\"end\":0.94}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a color name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.28, 1.64, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":1.28,\"end\":1.64}]"}
{"utterance_id": "1673-143397-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0018.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Process the full audio offline. Find every instance of a money or currency term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.4, 3.87, "gold"]], "response_mode": "offline", "answer": "[{\"word\":\"gold\",\"start\":3.4,\"end\":3.87}]"}
{"utterance_id": "5895-34629-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0007.flac", "task_type": "lexical_detection", "query": "hung", "prompt": "Once the audio has finished, identify every instance of the word \"hung\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.12, 6.31, "hung"]], "response_mode": "offline", "answer": "[{\"word\":\"hung\",\"start\":6.12,\"end\":6.31}]"}
{"utterance_id": "6345-93306-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0009.flac", "task_type": "lexical_detection", "query": "cut", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"cut\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.13, "cut"]], "response_mode": "offline", "answer": "[{\"word\":\"cut\",\"start\":0.9,\"end\":1.13}]"}
{"utterance_id": "3752-4943-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.69, "north"]], "response_mode": "offline", "answer": "[{\"word\":\"north\",\"start\":1.44,\"end\":1.69}]"}
{"utterance_id": "3536-23268-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0019.flac", "task_type": "lexical_detection", "query": "approving", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"approving\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.09, 2.61, "approving"]], "response_mode": "offline", "answer": "[{\"word\":\"approving\",\"start\":2.09,\"end\":2.61}]"}
{"utterance_id": "1919-142785-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0017.flac", "task_type": "lexical_detection", "query": "mentioned", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"mentioned\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.15, 4.58, "mentioned"]], "response_mode": "offline", "answer": "[{\"word\":\"mentioned\",\"start\":4.15,\"end\":4.58}]"}
{"utterance_id": "2277-149874-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.39, 2.66, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":2.39,\"end\":2.66}]"}
{"utterance_id": "2803-154328-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0010.flac", "task_type": "lexical_counting", "query": "last", "prompt": "Listen to the entire clip, then report the count sequence for the word \"last\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.26, "one"], [5.15, 5.76, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.72,\"end\":1.26},{\"word\":\"two\",\"start\":5.15,\"end\":5.76}]"}
{"utterance_id": "3081-166546-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0034.flac", "task_type": "lexical_detection", "query": "perhaps", "prompt": "Once the audio has finished, identify every instance of the word \"perhaps\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.64, 4.21, "perhaps"]], "response_mode": "offline", "answer": "[{\"word\":\"perhaps\",\"start\":3.64,\"end\":4.21}]"}
{"utterance_id": "2078-142845-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[21.92, 22.59, "three quarters"]], "response_mode": "offline", "answer": "[{\"word\":\"three quarters\",\"start\":21.92,\"end\":22.59}]"}
{"utterance_id": "3000-15664-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0026.flac", "task_type": "lexical_detection", "query": "mule", "prompt": "Once the audio has finished, identify every instance of the word \"mule\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.54, 0.85, "mule"]], "response_mode": "offline", "answer": "[{\"word\":\"mule\",\"start\":0.54,\"end\":0.85}]"}
{"utterance_id": "3536-8226-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.07, 4.67, "trevelyan"]], "response_mode": "offline", "answer": "[{\"word\":\"trevelyan\",\"start\":4.07,\"end\":4.67}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a number with word/start/end. If none, reply [].", "labels": [[2.26, 2.45, "one"], [4.71, 5.17, "two"], [7.48, 7.78, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.26,\"end\":2.45},{\"word\":\"two\",\"start\":4.71,\"end\":5.17},{\"word\":\"three\",\"start\":7.48,\"end\":7.78}]"}
{"utterance_id": "3752-4944-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0041.flac", "task_type": "lexical_detection", "query": "grass", "prompt": "Process the full audio offline. Find every instance of the word \"grass\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.25, 1.58, "grass"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":1.25,\"end\":1.58}]"}
{"utterance_id": "6345-93306-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0018.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.81, 2.1, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":1.81,\"end\":2.1}]"}
{"utterance_id": "2428-83699-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0017.flac", "task_type": "lexical_detection", "query": "station", "prompt": "After listening to the whole recording, output every instance of the word \"station\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.86, 4.35, "station"]], "response_mode": "offline", "answer": "[{\"word\":\"station\",\"start\":3.86,\"end\":4.35}]"}
{"utterance_id": "1993-147149-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an animal name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 1.05, "ghoul"]], "response_mode": "offline", "answer": "[{\"word\":\"ghoul\",\"start\":0.69,\"end\":1.05}]"}
{"utterance_id": "6319-275224-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0002.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Do not answer until the audio ends. Then list each instance of a time of day as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.45, 4.89, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":4.45,\"end\":4.89}]"}
{"utterance_id": "652-130726-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0026.flac", "task_type": "lexical_detection", "query": "ounces", "prompt": "Once the audio has finished, identify every instance of the word \"ounces\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.01, 5.33, "ounces"]], "response_mode": "offline", "answer": "[{\"word\":\"ounces\",\"start\":5.01,\"end\":5.33}]"}
{"utterance_id": "1272-141231-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0021.flac", "task_type": "lexical_detection", "query": "entered", "prompt": "When the utterance is complete, report all instances of the word \"entered\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.54, "entered"]], "response_mode": "offline", "answer": "[{\"word\":\"entered\",\"start\":1.31,\"end\":1.54}]"}
{"utterance_id": "1993-147964-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0007.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[1.28, 1.64, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.28,\"end\":1.64}]"}
{"utterance_id": "3000-15664-0034", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0034.flac", "task_type": "lexical_counting", "query": "shasta", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"shasta\" (word, start, end). If none, reply [].", "labels": [[0.86, 1.29, "one"], [4.32, 4.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.86,\"end\":1.29},{\"word\":\"two\",\"start\":4.32,\"end\":4.73}]"}
{"utterance_id": "2086-149220-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.23, 8.4, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":8.23,\"end\":8.4}]"}
{"utterance_id": "777-126732-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0040.flac", "task_type": "lexical_detection", "query": "flesh", "prompt": "After listening to the whole recording, output every instance of the word \"flesh\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.05, 2.44, "flesh"]], "response_mode": "offline", "answer": "[{\"word\":\"flesh\",\"start\":2.05,\"end\":2.44}]"}
{"utterance_id": "3000-15664-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0032.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 1.12, "ducks"]], "response_mode": "offline", "answer": "[{\"word\":\"ducks\",\"start\":0.45,\"end\":1.12}]"}
{"utterance_id": "7850-286674-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0003.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.55, 1.99, "food"], [5.35, 5.91, "meals"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":1.55,\"end\":1.99},{\"word\":\"meals\",\"start\":5.35,\"end\":5.91}]"}
{"utterance_id": "5895-34622-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0020.flac", "task_type": "lexical_detection", "query": "poet", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"poet\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.77, "poet"]], "response_mode": "offline", "answer": "[{\"word\":\"poet\",\"start\":1.31,\"end\":1.77}]"}
{"utterance_id": "3536-8226-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.06, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.7,\"end\":2.06}]"}
{"utterance_id": "6345-64257-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0014.flac", "task_type": "lexical_detection", "query": "greeting", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"greeting\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.32, 2.88, "greeting"]], "response_mode": "offline", "answer": "[{\"word\":\"greeting\",\"start\":2.32,\"end\":2.88}]"}
{"utterance_id": "3081-166546-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0001.flac", "task_type": "lexical_detection", "query": "absorbed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"absorbed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.02, "absorbed"]], "response_mode": "offline", "answer": "[{\"word\":\"absorbed\",\"start\":1.44,\"end\":2.02}]"}
{"utterance_id": "3081-166546-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0051.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[6.94, 7.18, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.94,\"end\":7.18}]"}
{"utterance_id": "2412-153947-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0011.flac", "task_type": "lexical_detection", "query": "attribute", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"attribute\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 1.0, "attribute"]], "response_mode": "offline", "answer": "[{\"word\":\"attribute\",\"start\":0.54,\"end\":1.0}]"}
{"utterance_id": "2412-153954-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0014.flac", "task_type": "lexical_detection", "query": "considerable", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"considerable\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.53, 7.16, "considerable"]], "response_mode": "offline", "answer": "[{\"word\":\"considerable\",\"start\":6.53,\"end\":7.16}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "lexical_detection", "query": "meal", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"meal\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.02, 9.27, "meal"]], "response_mode": "offline", "answer": "[{\"word\":\"meal\",\"start\":9.02,\"end\":9.27}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.51, 5.86, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.51,\"end\":5.86}]"}
{"utterance_id": "3536-23268-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0021.flac", "task_type": "lexical_detection", "query": "drank", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"drank\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.45, "drank"]], "response_mode": "offline", "answer": "[{\"word\":\"drank\",\"start\":1.14,\"end\":1.45}]"}
{"utterance_id": "1462-170138-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.54, "peggy"], [3.28, 3.88, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"peggy\",\"start\":1.19,\"end\":1.54},{\"word\":\"alexander\",\"start\":3.28,\"end\":3.88}]"}
{"utterance_id": "7976-110124-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0011.flac", "task_type": "lexical_detection", "query": "eldest", "prompt": "When the utterance is complete, report all instances of the word \"eldest\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.51, 3.84, "eldest"]], "response_mode": "offline", "answer": "[{\"word\":\"eldest\",\"start\":3.51,\"end\":3.84}]"}
{"utterance_id": "6319-57405-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0003.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "You will hear a complete utterance. When it ends, list every instance of a weather term you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.94, 4.29, "rain"]], "response_mode": "offline", "answer": "[{\"word\":\"rain\",\"start\":3.94,\"end\":4.29}]"}
{"utterance_id": "1993-147149-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0003.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen to the full audio. After it finishes, find every instance of a state or province name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.41, 10.15, "buckinghamshire"]], "response_mode": "offline", "answer": "[{\"word\":\"buckinghamshire\",\"start\":9.41,\"end\":10.15}]"}
{"utterance_id": "1673-143397-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0003.flac", "task_type": "lexical_detection", "query": "impunity", "prompt": "After the audio ends, extract all occurrences of the word \"impunity\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.52, 9.12, "impunity"]], "response_mode": "offline", "answer": "[{\"word\":\"impunity\",\"start\":8.52,\"end\":9.12}]"}
{"utterance_id": "6241-66616-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0017.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[0.48, 0.98, "one"], [2.1, 2.54, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.48,\"end\":0.98},{\"word\":\"two\",\"start\":2.1,\"end\":2.54}]"}
{"utterance_id": "777-126732-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0059.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.79, 1.26, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":0.79,\"end\":1.26}]"}
{"utterance_id": "1988-147956-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.19, "brown"], [4.07, 4.44, "dark"]], "response_mode": "offline", "answer": "[{\"word\":\"brown\",\"start\":0.88,\"end\":1.19},{\"word\":\"dark\",\"start\":4.07,\"end\":4.44}]"}
{"utterance_id": "1673-143396-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0018.flac", "task_type": "lexical_detection", "query": "reserve", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"reserve\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.72, 6.12, "reserve"]], "response_mode": "offline", "answer": "[{\"word\":\"reserve\",\"start\":5.72,\"end\":6.12}]"}
{"utterance_id": "3536-23268-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0017.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.63, 4.09, "tea"]], "response_mode": "offline", "answer": "[{\"word\":\"tea\",\"start\":3.63,\"end\":4.09}]"}
{"utterance_id": "1272-135031-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.88, 8.42, "shaggy"]], "response_mode": "offline", "answer": "[{\"word\":\"shaggy\",\"start\":7.88,\"end\":8.42}]"}
{"utterance_id": "1993-147964-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0010.flac", "task_type": "lexical_detection", "query": "working", "prompt": "Once the audio has finished, identify every instance of the word \"working\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.8, 4.2, "working"]], "response_mode": "offline", "answer": "[{\"word\":\"working\",\"start\":3.8,\"end\":4.2}]"}
{"utterance_id": "6319-57405-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0009.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After listening to the whole recording, output timed count words for a body part (one, two, three, ...) as JSON. If none, reply [].", "labels": [[6.07, 6.55, "one"], [7.9, 8.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.07,\"end\":6.55},{\"word\":\"two\",\"start\":7.9,\"end\":8.43}]"}
{"utterance_id": "5338-24640-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0002.flac", "task_type": "lexical_detection", "query": "otherwise", "prompt": "Process the full audio offline. Find every instance of the word \"otherwise\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.27, "otherwise"]], "response_mode": "offline", "answer": "[{\"word\":\"otherwise\",\"start\":1.64,\"end\":2.27}]"}
{"utterance_id": "6241-61946-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0020.flac", "task_type": "lexical_detection", "query": "spreading", "prompt": "Once the audio has finished, identify every instance of the word \"spreading\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.49, 2.91, "spreading"]], "response_mode": "offline", "answer": "[{\"word\":\"spreading\",\"start\":2.49,\"end\":2.91}]"}
{"utterance_id": "1462-170145-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0017.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.59, 3.1, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":2.59,\"end\":3.1}]"}
{"utterance_id": "2428-83699-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0008.flac", "task_type": "lexical_detection", "query": "lively", "prompt": "Once the audio has finished, identify every instance of the word \"lively\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.35, 3.96, "lively"]], "response_mode": "offline", "answer": "[{\"word\":\"lively\",\"start\":3.35,\"end\":3.96}]"}
{"utterance_id": "2035-147961-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0032.flac", "task_type": "lexical_detection", "query": "louder", "prompt": "After listening to the whole recording, output every instance of the word \"louder\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.45, 5.96, "louder"]], "response_mode": "offline", "answer": "[{\"word\":\"louder\",\"start\":5.45,\"end\":5.96}]"}
{"utterance_id": "7850-73752-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0017.flac", "task_type": "lexical_detection", "query": "affairs", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"affairs\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.23, 3.83, "affairs"]], "response_mode": "offline", "answer": "[{\"word\":\"affairs\",\"start\":3.23,\"end\":3.83}]"}
{"utterance_id": "1988-24833-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0013.flac", "task_type": "lexical_detection", "query": "pillows", "prompt": "Once the audio has finished, identify every instance of the word \"pillows\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.6, 0.93, "pillows"]], "response_mode": "offline", "answer": "[{\"word\":\"pillows\",\"start\":0.6,\"end\":0.93}]"}
{"utterance_id": "2078-142845-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0009.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Do not answer until the audio ends. Then list each instance of a country name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.64, 2.44, "italian"]], "response_mode": "offline", "answer": "[{\"word\":\"italian\",\"start\":1.64,\"end\":2.44}]"}
{"utterance_id": "3576-138058-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0019.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.68, 1.01, "horse"]], "response_mode": "offline", "answer": "[{\"word\":\"horse\",\"start\":0.68,\"end\":1.01}]"}
{"utterance_id": "3536-23268-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0016.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.56, 6.87, "one"], [8.37, 8.66, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.56,\"end\":6.87},{\"word\":\"two\",\"start\":8.37,\"end\":8.66}]"}
{"utterance_id": "6319-64726-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0000.flac", "task_type": "lexical_detection", "query": "birth", "prompt": "When the utterance is complete, report all instances of the word \"birth\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.08, 2.43, "birth"]], "response_mode": "offline", "answer": "[{\"word\":\"birth\",\"start\":2.08,\"end\":2.43}]"}
{"utterance_id": "6295-64301-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0012.flac", "task_type": "lexical_detection", "query": "broken", "prompt": "Once the audio has finished, identify every instance of the word \"broken\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.81, 1.42, "broken"]], "response_mode": "offline", "answer": "[{\"word\":\"broken\",\"start\":0.81,\"end\":1.42}]"}
{"utterance_id": "3752-4943-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.2, 1.88, "miss nancy"]], "response_mode": "offline", "answer": "[{\"word\":\"miss nancy\",\"start\":1.2,\"end\":1.88}]"}
{"utterance_id": "2412-153947-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0005.flac", "task_type": "lexical_detection", "query": "hint", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"hint\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.81, 12.18, "hint"]], "response_mode": "offline", "answer": "[{\"word\":\"hint\",\"start\":11.81,\"end\":12.18}]"}
{"utterance_id": "8842-302196-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0004.flac", "task_type": "lexical_detection", "query": "science", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"science\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.6, 2.15, "science"]], "response_mode": "offline", "answer": "[{\"word\":\"science\",\"start\":1.6,\"end\":2.15}]"}
{"utterance_id": "6313-66129-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0017.flac", "task_type": "lexical_detection", "query": "ned", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ned\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.47, 3.79, "ned"]], "response_mode": "offline", "answer": "[{\"word\":\"ned\",\"start\":3.47,\"end\":3.79}]"}
{"utterance_id": "6295-64301-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.87, 4.2, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"mary\",\"start\":3.87,\"end\":4.2}]"}
{"utterance_id": "7850-73752-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0000.flac", "task_type": "lexical_detection", "query": "fortune", "prompt": "Process the full audio offline. Find every instance of the word \"fortune\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.23, 2.93, "fortune"]], "response_mode": "offline", "answer": "[{\"word\":\"fortune\",\"start\":2.23,\"end\":2.93}]"}
{"utterance_id": "6313-76958-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0003.flac", "task_type": "lexical_detection", "query": "followed", "prompt": "Process the full audio offline. Find every instance of the word \"followed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.16, 1.55, "followed"]], "response_mode": "offline", "answer": "[{\"word\":\"followed\",\"start\":1.16,\"end\":1.55}]"}
{"utterance_id": "5895-34629-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 2.18, "ursus"]], "response_mode": "offline", "answer": "[{\"word\":\"ursus\",\"start\":1.64,\"end\":2.18}]"}
{"utterance_id": "1988-148538-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.41, 2.71, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.41,\"end\":2.71}]"}
{"utterance_id": "84-121550-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0033.flac", "task_type": "lexical_detection", "query": "mingled", "prompt": "Process the full audio offline. Find every instance of the word \"mingled\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.22, 2.68, "mingled"]], "response_mode": "offline", "answer": "[{\"word\":\"mingled\",\"start\":2.22,\"end\":2.68}]"}
{"utterance_id": "8297-275156-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0003.flac", "task_type": "lexical_detection", "query": "gossip", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"gossip\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.22, "gossip"], [5.38, 5.85, "gossip"]], "response_mode": "offline", "answer": "[{\"word\":\"gossip\",\"start\":1.75,\"end\":2.22},{\"word\":\"gossip\",\"start\":5.38,\"end\":5.85}]"}
{"utterance_id": "422-122949-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0002.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the full audio. After it finishes, find every instance of a country name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.82, 3.16, "greece"]], "response_mode": "offline", "answer": "[{\"word\":\"greece\",\"start\":2.82,\"end\":3.16}]"}
{"utterance_id": "1993-147149-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0030.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.74, 9.2, "father"], [10.43, 10.9, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":8.74,\"end\":9.2},{\"word\":\"father\",\"start\":10.43,\"end\":10.9}]"}
{"utterance_id": "2277-149874-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0016.flac", "task_type": "lexical_detection", "query": "dollars", "prompt": "Once the audio has finished, identify every instance of the word \"dollars\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.52, 3.88, "dollars"]], "response_mode": "offline", "answer": "[{\"word\":\"dollars\",\"start\":3.52,\"end\":3.88}]"}
{"utterance_id": "7976-110124-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0022.flac", "task_type": "lexical_detection", "query": "sticking", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"sticking\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.05, 2.47, "sticking"]], "response_mode": "offline", "answer": "[{\"word\":\"sticking\",\"start\":2.05,\"end\":2.47}]"}
{"utterance_id": "6319-64726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0002.flac", "task_type": "lexical_detection", "query": "pierce", "prompt": "After listening to the whole recording, output every instance of the word \"pierce\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.34, 8.84, "pierce"]], "response_mode": "offline", "answer": "[{\"word\":\"pierce\",\"start\":8.34,\"end\":8.84}]"}
{"utterance_id": "2428-83705-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0022.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Process the full audio offline. Find every instance of a plant name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.17, 2.61, "bush"]], "response_mode": "offline", "answer": "[{\"word\":\"bush\",\"start\":2.17,\"end\":2.61}]"}
{"utterance_id": "3170-137482-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0038.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.55, 7.78, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":7.55,\"end\":7.78}]"}
{"utterance_id": "6241-61943-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0016.flac", "task_type": "lexical_detection", "query": "icelandic", "prompt": "Once the audio has finished, identify every instance of the word \"icelandic\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.11, 3.77, "icelandic"]], "response_mode": "offline", "answer": "[{\"word\":\"icelandic\",\"start\":3.11,\"end\":3.77}]"}
{"utterance_id": "3853-163249-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0052.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.12, 7.75, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":7.12,\"end\":7.75}]"}
{"utterance_id": "251-137823-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0025.flac", "task_type": "lexical_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of the word \"animal\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.64, "animal"]], "response_mode": "offline", "answer": "[{\"word\":\"animal\",\"start\":2.65,\"end\":3.64}]"}
{"utterance_id": "6313-66129-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0006.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.31, 2.8, "professor"]], "response_mode": "offline", "answer": "[{\"word\":\"professor\",\"start\":2.31,\"end\":2.8}]"}
{"utterance_id": "1988-147956-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0011.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.44, 0.71, "skin"], [2.44, 2.89, "cheeks"]], "response_mode": "offline", "answer": "[{\"word\":\"skin\",\"start\":0.44,\"end\":0.71},{\"word\":\"cheeks\",\"start\":2.44,\"end\":2.89}]"}
{"utterance_id": "5694-64025-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0014.flac", "task_type": "lexical_detection", "query": "soldier", "prompt": "Once the audio has finished, identify every instance of the word \"soldier\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.6, 3.25, "soldier"]], "response_mode": "offline", "answer": "[{\"word\":\"soldier\",\"start\":2.6,\"end\":3.25}]"}
{"utterance_id": "3000-15664-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0028.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a body part with word/start/end. If none, reply [].", "labels": [[2.61, 3.05, "one"], [5.85, 6.22, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.61,\"end\":3.05},{\"word\":\"two\",\"start\":5.85,\"end\":6.22}]"}
{"utterance_id": "1462-170142-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0038.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.31, 0.7, "hilda"]], "response_mode": "offline", "answer": "[{\"word\":\"hilda\",\"start\":0.31,\"end\":0.7}]"}
{"utterance_id": "3536-8226-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.52, 3.3, "lady milborough"]], "response_mode": "offline", "answer": "[{\"word\":\"lady milborough\",\"start\":2.52,\"end\":3.3}]"}
{"utterance_id": "3536-23268-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0010.flac", "task_type": "lexical_detection", "query": "immediately", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"immediately\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.71, "immediately"]], "response_mode": "offline", "answer": "[{\"word\":\"immediately\",\"start\":1.05,\"end\":1.71}]"}
{"utterance_id": "1993-147965-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0003.flac", "task_type": "lexical_detection", "query": "dull", "prompt": "Process the full audio offline. Find every instance of the word \"dull\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.66, 5.96, "dull"]], "response_mode": "offline", "answer": "[{\"word\":\"dull\",\"start\":5.66,\"end\":5.96}]"}
{"utterance_id": "2277-149897-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0029.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.83, 3.27, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":2.83,\"end\":3.27}]"}
{"utterance_id": "5338-284437-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0002.flac", "task_type": "lexical_detection", "query": "cabin", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"cabin\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.5, "cabin"]], "response_mode": "offline", "answer": "[{\"word\":\"cabin\",\"start\":1.92,\"end\":2.5}]"}
{"utterance_id": "84-121550-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0030.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Listen to the full audio. After it finishes, find every instance of a weather term. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.47, 5.79, "rain"]], "response_mode": "offline", "answer": "[{\"word\":\"rain\",\"start\":5.47,\"end\":5.79}]"}
{"utterance_id": "2086-149220-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0027.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After the audio ends, extract all occurrences of a country name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.86, 3.32, "morocco"]], "response_mode": "offline", "answer": "[{\"word\":\"morocco\",\"start\":2.86,\"end\":3.32}]"}
{"utterance_id": "6345-93302-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0004.flac", "task_type": "lexical_detection", "query": "curious", "prompt": "Once the audio has finished, identify every instance of the word \"curious\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.01, 2.44, "curious"]], "response_mode": "offline", "answer": "[{\"word\":\"curious\",\"start\":2.01,\"end\":2.44}]"}
{"utterance_id": "422-122949-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0007.flac", "task_type": "lexical_counting", "query": "according", "prompt": "Once the audio has finished, list each occurrence of the word \"according\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.46, 0.91, "one"], [5.3, 5.82, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.46,\"end\":0.91},{\"word\":\"two\",\"start\":5.3,\"end\":5.82}]"}
{"utterance_id": "3752-4943-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0014.flac", "task_type": "lexical_detection", "query": "unhappy", "prompt": "Once the audio has finished, identify every instance of the word \"unhappy\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.46, 1.86, "unhappy"]], "response_mode": "offline", "answer": "[{\"word\":\"unhappy\",\"start\":1.46,\"end\":1.86}]"}
{"utterance_id": "1462-170142-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0035.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.35, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":0.59,\"end\":1.35}]"}
{"utterance_id": "3081-166546-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0020.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.61, 4.0, "police"]], "response_mode": "offline", "answer": "[{\"word\":\"police\",\"start\":3.61,\"end\":4.0}]"}
{"utterance_id": "2078-142845-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0029.flac", "task_type": "lexical_detection", "query": "buttered", "prompt": "After the audio ends, extract all occurrences of the word \"buttered\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.66, "buttered"]], "response_mode": "offline", "answer": "[{\"word\":\"buttered\",\"start\":1.27,\"end\":1.66}]"}
{"utterance_id": "174-84280-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0015.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.88, 6.37, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":5.88,\"end\":6.37}]"}
{"utterance_id": "3752-4944-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0058.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Do not answer until the audio ends. Then list each instance of an organization name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.18, 1.73, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":1.18,\"end\":1.73}]"}
{"utterance_id": "3576-138058-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0033.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.07, 12.8, "senor"]], "response_mode": "offline", "answer": "[{\"word\":\"senor\",\"start\":12.07,\"end\":12.8}]"}
{"utterance_id": "2428-83699-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0027.flac", "task_type": "lexical_detection", "query": "steps", "prompt": "After the audio ends, extract all occurrences of the word \"steps\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.62, "steps"]], "response_mode": "offline", "answer": "[{\"word\":\"steps\",\"start\":2.93,\"end\":3.62}]"}
{"utterance_id": "8842-302196-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.61, "dante's"]], "response_mode": "offline", "answer": "[{\"word\":\"dante's\",\"start\":2.14,\"end\":2.61}]"}
{"utterance_id": "6345-93302-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0012.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.69, 4.07, "angel"]], "response_mode": "offline", "answer": "[{\"word\":\"angel\",\"start\":3.69,\"end\":4.07}]"}
{"utterance_id": "652-130726-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0007.flac", "task_type": "lexical_detection", "query": "poodle", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"poodle\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.36, 0.67, "poodle"]], "response_mode": "offline", "answer": "[{\"word\":\"poodle\",\"start\":0.36,\"end\":0.67}]"}
{"utterance_id": "7976-110523-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0007.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.53, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":0.8,\"end\":1.53}]"}
{"utterance_id": "652-130737-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0008.flac", "task_type": "lexical_detection", "query": "magnificent", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"magnificent\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.54, "magnificent"]], "response_mode": "offline", "answer": "[{\"word\":\"magnificent\",\"start\":0.83,\"end\":1.54}]"}
{"utterance_id": "3170-137482-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0019.flac", "task_type": "lexical_detection", "query": "picked", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"picked\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.78, "picked"]], "response_mode": "offline", "answer": "[{\"word\":\"picked\",\"start\":0.52,\"end\":0.78}]"}
{"utterance_id": "6319-57405-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0004.flac", "task_type": "lexical_detection", "query": "deucalion", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"deucalion\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.93, "deucalion"]], "response_mode": "offline", "answer": "[{\"word\":\"deucalion\",\"start\":1.2,\"end\":1.93}]"}
{"utterance_id": "8297-275156-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0002.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.34, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.97,\"end\":4.34}]"}
{"utterance_id": "777-126732-0069", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0069.flac", "task_type": "lexical_detection", "query": "seen", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"seen\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.1, 2.41, "seen"]], "response_mode": "offline", "answer": "[{\"word\":\"seen\",\"start\":2.1,\"end\":2.41}]"}
{"utterance_id": "6241-66616-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0006.flac", "task_type": "lexical_detection", "query": "newsome", "prompt": "After the audio ends, extract all occurrences of the word \"newsome\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.2, "newsome"]], "response_mode": "offline", "answer": "[{\"word\":\"newsome\",\"start\":3.75,\"end\":4.2}]"}
{"utterance_id": "7850-286674-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.88, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":0.57,\"end\":0.88}]"}
{"utterance_id": "6345-93302-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0019.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[0.69, 0.97, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":0.69,\"end\":0.97}]"}
{"utterance_id": "6241-66616-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0001.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "After the audio ends, tally every instance of a number. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[0.52, 0.81, "one"], [2.16, 2.47, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.81},{\"word\":\"two\",\"start\":2.16,\"end\":2.47}]"}
{"utterance_id": "2078-142845-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0048.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 1.98, "seventeen thirty four"]], "response_mode": "offline", "answer": "[{\"word\":\"seventeen thirty four\",\"start\":0.37,\"end\":1.98}]"}
{"utterance_id": "6295-64301-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0004.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.77, "earth"]], "response_mode": "offline", "answer": "[{\"word\":\"earth\",\"start\":0.5,\"end\":0.77}]"}
{"utterance_id": "2035-152373-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.07, 10.5, "bel"]], "response_mode": "offline", "answer": "[{\"word\":\"bel\",\"start\":10.07,\"end\":10.5}]"}
{"utterance_id": "1462-170145-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0013.flac", "task_type": "lexical_detection", "query": "feeling", "prompt": "After listening to the whole recording, output every instance of the word \"feeling\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.78, 4.26, "feeling"]], "response_mode": "offline", "answer": "[{\"word\":\"feeling\",\"start\":3.78,\"end\":4.26}]"}
{"utterance_id": "6295-244435-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0016.flac", "task_type": "lexical_detection", "query": "kentucky", "prompt": "When the utterance is complete, report all instances of the word \"kentucky\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.61, 1.1, "kentucky"]], "response_mode": "offline", "answer": "[{\"word\":\"kentucky\",\"start\":0.61,\"end\":1.1}]"}
{"utterance_id": "6295-244435-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0036.flac", "task_type": "lexical_detection", "query": "liveryman", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"liveryman\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.99, 4.75, "liveryman"]], "response_mode": "offline", "answer": "[{\"word\":\"liveryman\",\"start\":3.99,\"end\":4.75}]"}
{"utterance_id": "3752-4943-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.7, "sir north"]], "response_mode": "offline", "answer": "[{\"word\":\"sir north\",\"start\":0.9,\"end\":1.7}]"}
{"utterance_id": "3576-138058-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0030.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.88, "sancho"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho\",\"start\":0.4,\"end\":0.88}]"}
{"utterance_id": "2412-153948-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0000.flac", "task_type": "lexical_detection", "query": "painful", "prompt": "When the utterance is complete, report all instances of the word \"painful\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.02, 10.38, "painful"]], "response_mode": "offline", "answer": "[{\"word\":\"painful\",\"start\":10.02,\"end\":10.38}]"}
{"utterance_id": "2277-149874-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0020.flac", "task_type": "lexical_detection", "query": "offices", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"offices\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.65, 4.23, "offices"]], "response_mode": "offline", "answer": "[{\"word\":\"offices\",\"start\":3.65,\"end\":4.23}]"}
{"utterance_id": "1462-170142-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0025.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.27, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":0.94,\"end\":1.27}]"}
{"utterance_id": "777-126732-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0010.flac", "task_type": "lexical_detection", "query": "together", "prompt": "When the utterance is complete, report all instances of the word \"together\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.7, 3.27, "together"]], "response_mode": "offline", "answer": "[{\"word\":\"together\",\"start\":2.7,\"end\":3.27}]"}
{"utterance_id": "5536-43358-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.77, 1.63, "philosopher"]], "response_mode": "offline", "answer": "[{\"word\":\"philosopher\",\"start\":0.77,\"end\":1.63}]"}
{"utterance_id": "3752-4943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0025.flac", "task_type": "lexical_detection", "query": "arms", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"arms\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.98, 4.28, "arms"]], "response_mode": "offline", "answer": "[{\"word\":\"arms\",\"start\":3.98,\"end\":4.28}]"}
{"utterance_id": "3576-138058-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0036.flac", "task_type": "lexical_counting", "query": "gentlemen", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"gentlemen\" (word, start, end). If none, reply [].", "labels": [[5.69, 6.21, "one"], [17.63, 18.15, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.69,\"end\":6.21},{\"word\":\"two\",\"start\":17.63,\"end\":18.15}]"}
{"utterance_id": "1993-147964-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.01, 6.77, "grandmother"]], "response_mode": "offline", "answer": "[{\"word\":\"grandmother\",\"start\":6.01,\"end\":6.77}]"}
{"utterance_id": "1272-135031-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0009.flac", "task_type": "lexical_detection", "query": "doesn't", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"doesn't\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 0.9, "doesn't"]], "response_mode": "offline", "answer": "[{\"word\":\"doesn't\",\"start\":0.67,\"end\":0.9}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.56, 9.91, "head"], [11.26, 11.69, "hands"], [13.14, 13.75, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":9.56,\"end\":9.91},{\"word\":\"hands\",\"start\":11.26,\"end\":11.69},{\"word\":\"feet\",\"start\":13.14,\"end\":13.75}]"}
{"utterance_id": "84-121550-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0006.flac", "task_type": "lexical_detection", "query": "supreme", "prompt": "Once the audio has finished, identify every instance of the word \"supreme\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.03, 1.7, "supreme"]], "response_mode": "offline", "answer": "[{\"word\":\"supreme\",\"start\":1.03,\"end\":1.7}]"}
{"utterance_id": "6313-76958-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0005.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.29, "guard"]], "response_mode": "offline", "answer": "[{\"word\":\"guard\",\"start\":1.97,\"end\":2.29}]"}
{"utterance_id": "3170-137482-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0024.flac", "task_type": "lexical_detection", "query": "settled", "prompt": "Once the audio has finished, identify every instance of the word \"settled\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.68, 3.03, "settled"]], "response_mode": "offline", "answer": "[{\"word\":\"settled\",\"start\":2.68,\"end\":3.03}]"}
{"utterance_id": "3536-23268-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0008.flac", "task_type": "lexical_detection", "query": "labours", "prompt": "After the audio ends, extract all occurrences of the word \"labours\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.25, "labours"]], "response_mode": "offline", "answer": "[{\"word\":\"labours\",\"start\":1.85,\"end\":2.25}]"}
{"utterance_id": "7976-110124-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0002.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.86, 2.2, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.86,\"end\":2.2}]"}
{"utterance_id": "2902-9006-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.37, 0.93, "julian's"]], "response_mode": "offline", "answer": "[{\"word\":\"julian's\",\"start\":0.37,\"end\":0.93}]"}
{"utterance_id": "3081-166546-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0027.flac", "task_type": "lexical_detection", "query": "chat", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"chat\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.14, 3.65, "chat"]], "response_mode": "offline", "answer": "[{\"word\":\"chat\",\"start\":3.14,\"end\":3.65}]"}
{"utterance_id": "2412-153954-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0016.flac", "task_type": "lexical_detection", "query": "coughed", "prompt": "Once the audio has finished, identify every instance of the word \"coughed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.92, 4.26, "coughed"]], "response_mode": "offline", "answer": "[{\"word\":\"coughed\",\"start\":3.92,\"end\":4.26}]"}
{"utterance_id": "1272-141231-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0002.flac", "task_type": "lexical_detection", "query": "ache", "prompt": "After the audio ends, extract all occurrences of the word \"ache\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.85, 4.11, "ache"]], "response_mode": "offline", "answer": "[{\"word\":\"ache\",\"start\":3.85,\"end\":4.11}]"}
{"utterance_id": "2428-83705-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0006.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[2.9, 3.35, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":2.9,\"end\":3.35}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.95, 5.26, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.95,\"end\":5.26}]"}
{"utterance_id": "8297-275154-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0013.flac", "task_type": "lexical_detection", "query": "neither", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"neither\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.37, "neither"]], "response_mode": "offline", "answer": "[{\"word\":\"neither\",\"start\":0.97,\"end\":1.37}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 1.93, "one"], [3.12, 3.39, "eighty"], [8.81, 9.39, "sixty"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.75,\"end\":1.93},{\"word\":\"eighty\",\"start\":3.12,\"end\":3.39},{\"word\":\"sixty\",\"start\":8.81,\"end\":9.39}]"}
{"utterance_id": "1673-143397-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0002.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.31, 3.96, "magistrate"]], "response_mode": "offline", "answer": "[{\"word\":\"magistrate\",\"start\":3.31,\"end\":3.96}]"}
{"utterance_id": "7976-105575-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0029.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a family relation with fields word, start, end (seconds). If none, reply [].", "labels": [[1.97, 2.37, "brother"]], "response_mode": "offline", "answer": "[{\"word\":\"brother\",\"start\":1.97,\"end\":2.37}]"}
{"utterance_id": "5694-64025-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0016.flac", "task_type": "lexical_counting", "query": "off", "prompt": "Once the audio has finished, list each occurrence of the word \"off\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[5.15, 5.38, "one"], [8.21, 8.56, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.15,\"end\":5.38},{\"word\":\"two\",\"start\":8.21,\"end\":8.56}]"}
{"utterance_id": "2428-83699-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0021.flac", "task_type": "lexical_detection", "query": "luggage", "prompt": "When the utterance is complete, report all instances of the word \"luggage\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.59, "luggage"]], "response_mode": "offline", "answer": "[{\"word\":\"luggage\",\"start\":2.06,\"end\":2.59}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.94, 6.15, "gray"], [9.99, 10.35, "black"]], "response_mode": "offline", "answer": "[{\"word\":\"gray\",\"start\":5.94,\"end\":6.15},{\"word\":\"black\",\"start\":9.99,\"end\":10.35}]"}
{"utterance_id": "7976-110523-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0011.flac", "task_type": "lexical_detection", "query": "saying", "prompt": "After the audio ends, extract all occurrences of the word \"saying\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.61, 5.12, "saying"]], "response_mode": "offline", "answer": "[{\"word\":\"saying\",\"start\":4.61,\"end\":5.12}]"}
{"utterance_id": "1462-170142-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0000.flac", "task_type": "lexical_detection", "query": "voyage", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"voyage\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 2.27, "voyage"]], "response_mode": "offline", "answer": "[{\"word\":\"voyage\",\"start\":1.71,\"end\":2.27}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Do not answer until the audio ends. Then list each instance of a weather term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.2, 2.62, "thaw"]], "response_mode": "offline", "answer": "[{\"word\":\"thaw\",\"start\":2.2,\"end\":2.62}]"}
{"utterance_id": "1272-128104-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0005.flac", "task_type": "lexical_detection", "query": "unnecessary", "prompt": "After the audio ends, extract all occurrences of the word \"unnecessary\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.42, "unnecessary"]], "response_mode": "offline", "answer": "[{\"word\":\"unnecessary\",\"start\":1.55,\"end\":2.42}]"}
{"utterance_id": "2412-153947-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0015.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 3.0, "erewhon"]], "response_mode": "offline", "answer": "[{\"word\":\"erewhon\",\"start\":2.55,\"end\":3.0}]"}
{"utterance_id": "3081-166546-0073", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0073.flac", "task_type": "lexical_detection", "query": "yes", "prompt": "Process the full audio offline. Find every instance of the word \"yes\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.2, "yes"]], "response_mode": "offline", "answer": "[{\"word\":\"yes\",\"start\":0.5,\"end\":1.2}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.69,\"end\":2.08}]"}
{"utterance_id": "2277-149896-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0018.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 0.83, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.55,\"end\":0.83}]"}
{"utterance_id": "6313-66125-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0018.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.31, 2.71, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":2.31,\"end\":2.71}]"}
{"utterance_id": "6313-66129-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0010.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.91, 2.32, "guide"]], "response_mode": "offline", "answer": "[{\"word\":\"guide\",\"start\":1.91,\"end\":2.32}]"}
{"utterance_id": "6319-57405-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.66, 4.37, "deucalion"], [5.99, 6.87, "prometheus"]], "response_mode": "offline", "answer": "[{\"word\":\"deucalion\",\"start\":3.66,\"end\":4.37},{\"word\":\"prometheus\",\"start\":5.99,\"end\":6.87}]"}
{"utterance_id": "6313-66129-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0003.flac", "task_type": "lexical_detection", "query": "tad", "prompt": "When the utterance is complete, report all instances of the word \"tad\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.82, 5.17, "tad"]], "response_mode": "offline", "answer": "[{\"word\":\"tad\",\"start\":4.82,\"end\":5.17}]"}
{"utterance_id": "652-130737-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0002.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 2.86, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":2.65,\"end\":2.86}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.2, 2.52, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.2,\"end\":2.52}]"}
{"utterance_id": "3752-4944-0059", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0059.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the full audio. After it finishes, find every instance of a food name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.12, "brandy"]], "response_mode": "offline", "answer": "[{\"word\":\"brandy\",\"start\":1.74,\"end\":2.12}]"}
{"utterance_id": "777-126732-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0044.flac", "task_type": "lexical_detection", "query": "insight", "prompt": "When the utterance is complete, report all instances of the word \"insight\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.14, "insight"]], "response_mode": "offline", "answer": "[{\"word\":\"insight\",\"start\":0.66,\"end\":1.14}]"}
{"utterance_id": "6319-64726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.79, 1.43, "fifteen"], [5.51, 5.67, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen\",\"start\":0.79,\"end\":1.43},{\"word\":\"one\",\"start\":5.51,\"end\":5.67}]"}
{"utterance_id": "2902-9008-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[18.49, 18.76, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":18.49,\"end\":18.76}]"}
{"utterance_id": "3536-8226-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0006.flac", "task_type": "semantic_detection", "query": "money", "prompt": "Once the audio has finished, identify every instance of a money or currency term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.86, 2.27, "pounds"]], "response_mode": "offline", "answer": "[{\"word\":\"pounds\",\"start\":1.86,\"end\":2.27}]"}
{"utterance_id": "84-121550-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0008.flac", "task_type": "lexical_detection", "query": "doth", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"doth\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 1.97, "doth"]], "response_mode": "offline", "answer": "[{\"word\":\"doth\",\"start\":1.73,\"end\":1.97}]"}
{"utterance_id": "3081-166546-0053", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0053.flac", "task_type": "lexical_detection", "query": "naturally", "prompt": "Process the full audio offline. Find every instance of the word \"naturally\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 0.85, "naturally"]], "response_mode": "offline", "answer": "[{\"word\":\"naturally\",\"start\":0.35,\"end\":0.85}]"}
{"utterance_id": "6295-64301-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0013.flac", "task_type": "lexical_detection", "query": "solitude", "prompt": "When the utterance is complete, report all instances of the word \"solitude\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.76, "solitude"]], "response_mode": "offline", "answer": "[{\"word\":\"solitude\",\"start\":1.93,\"end\":2.76}]"}
{"utterance_id": "7850-281318-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0007.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.32, 1.59, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":1.32,\"end\":1.59}]"}
{"utterance_id": "84-121550-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0018.flac", "task_type": "semantic_counting", "query": "ordinal", "prompt": "Process the full audio offline. For every occurrence of an ordinal number, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.11, "one"], [5.56, 6.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.53,\"end\":1.11},{\"word\":\"two\",\"start\":5.56,\"end\":6.12}]"}
{"utterance_id": "8297-275155-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.74, 3.13, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":2.74,\"end\":3.13}]"}
{"utterance_id": "7976-105575-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.37, 0.72, "grant"]], "response_mode": "offline", "answer": "[{\"word\":\"grant\",\"start\":0.37,\"end\":0.72}]"}
{"utterance_id": "174-50561-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0004.flac", "task_type": "lexical_detection", "query": "emperor's", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"emperor's\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 1.11, "emperor's"]], "response_mode": "offline", "answer": "[{\"word\":\"emperor's\",\"start\":0.66,\"end\":1.11}]"}
{"utterance_id": "7976-110523-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0021.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.96, 4.37, "father's"]], "response_mode": "offline", "answer": "[{\"word\":\"father's\",\"start\":3.96,\"end\":4.37}]"}
{"utterance_id": "2428-83699-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0037.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.01, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":1.61,\"end\":2.01}]"}
{"utterance_id": "6319-64726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.43, 5.79, "head"], [9.0, 9.34, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":5.43,\"end\":5.79},{\"word\":\"hand\",\"start\":9.0,\"end\":9.34}]"}
{"utterance_id": "5694-64025-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0014.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a profession (word, start, end). If none, reply [].", "labels": [[2.21, 3.25, "one"], [7.06, 7.75, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.21,\"end\":3.25},{\"word\":\"two\",\"start\":7.06,\"end\":7.75}]"}
{"utterance_id": "6319-64726-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0009.flac", "task_type": "lexical_detection", "query": "path", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"path\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.29, 9.85, "path"]], "response_mode": "offline", "answer": "[{\"word\":\"path\",\"start\":9.29,\"end\":9.85}]"}
{"utterance_id": "3576-138058-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.04, 2.55, "claudia"]], "response_mode": "offline", "answer": "[{\"word\":\"claudia\",\"start\":2.04,\"end\":2.55}]"}
{"utterance_id": "3000-15664-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0037.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Process the full audio offline. Find every instance of a color name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.25, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"green\",\"start\":0.81,\"end\":1.25}]"}
{"utterance_id": "2035-147960-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0010.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[7.55, 7.96, "buffalo"]], "response_mode": "offline", "answer": "[{\"word\":\"buffalo\",\"start\":7.55,\"end\":7.96}]"}
{"utterance_id": "2035-152373-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0000.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.3, 2.75, "church"]], "response_mode": "offline", "answer": "[{\"word\":\"church\",\"start\":2.3,\"end\":2.75}]"}
{"utterance_id": "6295-244435-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0003.flac", "task_type": "lexical_detection", "query": "shore", "prompt": "After the audio ends, extract all occurrences of the word \"shore\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.42, 2.95, "shore"]], "response_mode": "offline", "answer": "[{\"word\":\"shore\",\"start\":2.42,\"end\":2.95}]"}
{"utterance_id": "3081-166546-0084", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0084.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After the audio ends, extract all occurrences of a city name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.26, "sweetwater"]], "response_mode": "offline", "answer": "[{\"word\":\"sweetwater\",\"start\":0.51,\"end\":1.26}]"}
{"utterance_id": "174-84280-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "When the utterance is complete, report all instances of an ordinal number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.04, 8.48, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":8.04,\"end\":8.48}]"}
{"utterance_id": "652-129742-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0014.flac", "task_type": "lexical_detection", "query": "dish", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"dish\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.18, "dish"]], "response_mode": "offline", "answer": "[{\"word\":\"dish\",\"start\":0.9,\"end\":1.18}]"}
{"utterance_id": "5694-64038-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0007.flac", "task_type": "lexical_detection", "query": "persuaded", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"persuaded\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.72, 1.33, "persuaded"]], "response_mode": "offline", "answer": "[{\"word\":\"persuaded\",\"start\":0.72,\"end\":1.33}]"}
{"utterance_id": "1919-142785-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0006.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.74, 2.28, "cucumber"]], "response_mode": "offline", "answer": "[{\"word\":\"cucumber\",\"start\":1.74,\"end\":2.28}]"}
{"utterance_id": "422-122949-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0022.flac", "task_type": "lexical_detection", "query": "loves", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"loves\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.2, 1.57, "loves"]], "response_mode": "offline", "answer": "[{\"word\":\"loves\",\"start\":1.2,\"end\":1.57}]"}
{"utterance_id": "3853-163249-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0044.flac", "task_type": "lexical_detection", "query": "christie", "prompt": "After the audio ends, extract all occurrences of the word \"christie\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.57, 3.12, "christie"]], "response_mode": "offline", "answer": "[{\"word\":\"christie\",\"start\":2.57,\"end\":3.12}]"}
{"utterance_id": "3576-138058-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.81, 3.72, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":2.81,\"end\":3.72}]"}
{"utterance_id": "3170-137482-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0018.flac", "task_type": "lexical_detection", "query": "traitor", "prompt": "After listening to the whole recording, output every instance of the word \"traitor\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 1.86, "traitor"]], "response_mode": "offline", "answer": "[{\"word\":\"traitor\",\"start\":1.5,\"end\":1.86}]"}
{"utterance_id": "1988-148538-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0007.flac", "task_type": "lexical_detection", "query": "sixteen", "prompt": "After listening to the whole recording, output every instance of the word \"sixteen\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.45, "sixteen"]], "response_mode": "offline", "answer": "[{\"word\":\"sixteen\",\"start\":0.81,\"end\":1.45}]"}
{"utterance_id": "1462-170145-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0015.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.51, "new york"]], "response_mode": "offline", "answer": "[{\"word\":\"new york\",\"start\":4.04,\"end\":4.51}]"}
{"utterance_id": "7976-110523-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0009.flac", "task_type": "lexical_detection", "query": "forest", "prompt": "After listening to the whole recording, output every instance of the word \"forest\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.58, "forest"]], "response_mode": "offline", "answer": "[{\"word\":\"forest\",\"start\":1.14,\"end\":1.58}]"}
{"utterance_id": "777-126732-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0062.flac", "task_type": "lexical_detection", "query": "yes", "prompt": "When the utterance is complete, report all instances of the word \"yes\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.77, "yes"]], "response_mode": "offline", "answer": "[{\"word\":\"yes\",\"start\":0.3,\"end\":0.77}]"}
{"utterance_id": "6241-61946-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0002.flac", "task_type": "lexical_detection", "query": "accord", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"accord\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 3.03, "accord"]], "response_mode": "offline", "answer": "[{\"word\":\"accord\",\"start\":2.52,\"end\":3.03}]"}
{"utterance_id": "2277-149896-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0008.flac", "task_type": "lexical_detection", "query": "suspicious", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"suspicious\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.51, 6.16, "suspicious"]], "response_mode": "offline", "answer": "[{\"word\":\"suspicious\",\"start\":5.51,\"end\":6.16}]"}
{"utterance_id": "652-130737-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0001.flac", "task_type": "semantic_counting", "query": "color", "prompt": "After the audio ends, tally every instance of a color name. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.81, 2.12, "one"], [4.95, 5.21, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.81,\"end\":2.12},{\"word\":\"two\",\"start\":4.95,\"end\":5.21}]"}
{"utterance_id": "777-126732-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0029.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.75, 3.31, "shoulders"]], "response_mode": "offline", "answer": "[{\"word\":\"shoulders\",\"start\":2.75,\"end\":3.31}]"}
{"utterance_id": "3081-166546-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0038.flac", "task_type": "lexical_detection", "query": "companion", "prompt": "After the audio ends, extract all occurrences of the word \"companion\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.88, "companion"]], "response_mode": "offline", "answer": "[{\"word\":\"companion\",\"start\":1.05,\"end\":1.88}]"}
{"utterance_id": "2086-149220-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0011.flac", "task_type": "lexical_detection", "query": "distinguishing", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"distinguishing\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.61, 1.33, "distinguishing"]], "response_mode": "offline", "answer": "[{\"word\":\"distinguishing\",\"start\":0.61,\"end\":1.33}]"}
{"utterance_id": "2277-149896-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0013.flac", "task_type": "lexical_detection", "query": "desk", "prompt": "When the utterance is complete, report all instances of the word \"desk\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 2.97, "desk"]], "response_mode": "offline", "answer": "[{\"word\":\"desk\",\"start\":2.62,\"end\":2.97}]"}
{"utterance_id": "5694-64025-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0022.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the full audio. After it finishes, find every instance of an animal name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.06, 2.44, "mule's"]], "response_mode": "offline", "answer": "[{\"word\":\"mule's\",\"start\":2.06,\"end\":2.44}]"}
{"utterance_id": "1993-147149-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0020.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the entire clip, then report the count sequence for a person's name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.95, "one"], [10.92, 11.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.95},{\"word\":\"two\",\"start\":10.92,\"end\":11.43}]"}
{"utterance_id": "1673-143397-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.31, 1.04, "nestorius"], [5.38, 6.05, "chrysostom"]], "response_mode": "offline", "answer": "[{\"word\":\"nestorius\",\"start\":0.31,\"end\":1.04},{\"word\":\"chrysostom\",\"start\":5.38,\"end\":6.05}]"}
{"utterance_id": "1988-147956-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0016.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.26, 1.6, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.26,\"end\":1.6}]"}
{"utterance_id": "3576-138058-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0031.flac", "task_type": "lexical_detection", "query": "man", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"man\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.29, 2.71, "man"]], "response_mode": "offline", "answer": "[{\"word\":\"man\",\"start\":2.29,\"end\":2.71}]"}
{"utterance_id": "3170-137482-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0015.flac", "task_type": "lexical_counting", "query": "good", "prompt": "After the audio ends, tally every instance of the word \"good\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[5.89, 6.08, "one"], [8.86, 9.13, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.89,\"end\":6.08},{\"word\":\"two\",\"start\":8.86,\"end\":9.13}]"}
{"utterance_id": "777-126732-0058", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0058.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Process the full audio offline. Find every instance of an organization name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 5.02, "police"]], "response_mode": "offline", "answer": "[{\"word\":\"police\",\"start\":4.47,\"end\":5.02}]"}
{"utterance_id": "1462-170145-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[7.47, 8.16, "alexander's"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander's\",\"start\":7.47,\"end\":8.16}]"}
{"utterance_id": "2078-142845-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0016.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 3.14, "maize plant"]], "response_mode": "offline", "answer": "[{\"word\":\"maize plant\",\"start\":1.91,\"end\":3.14}]"}
{"utterance_id": "1462-170145-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0017.flac", "task_type": "lexical_detection", "query": "london", "prompt": "After listening to the whole recording, output every instance of the word \"london\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.59, 3.1, "london"]], "response_mode": "offline", "answer": "[{\"word\":\"london\",\"start\":2.59,\"end\":3.1}]"}
{"utterance_id": "3752-4944-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0038.flac", "task_type": "lexical_detection", "query": "times", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"times\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.56, "times"]], "response_mode": "offline", "answer": "[{\"word\":\"times\",\"start\":2.02,\"end\":2.56}]"}
{"utterance_id": "174-168635-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0012.flac", "task_type": "lexical_detection", "query": "commissioned", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"commissioned\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.43, 3.91, "commissioned"]], "response_mode": "offline", "answer": "[{\"word\":\"commissioned\",\"start\":3.43,\"end\":3.91}]"}
{"utterance_id": "2078-142845-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0026.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 2.2, "toast"]], "response_mode": "offline", "answer": "[{\"word\":\"toast\",\"start\":1.44,\"end\":2.2}]"}
{"utterance_id": "2277-149897-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0016.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.71, 4.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.71,\"end\":4.06}]"}
{"utterance_id": "1673-143397-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0010.flac", "task_type": "lexical_detection", "query": "appeal", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"appeal\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.59, 3.09, "appeal"]], "response_mode": "offline", "answer": "[{\"word\":\"appeal\",\"start\":2.59,\"end\":3.09}]"}
{"utterance_id": "2428-83699-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0023.flac", "task_type": "lexical_detection", "query": "beyond", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"beyond\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.21, "beyond"]], "response_mode": "offline", "answer": "[{\"word\":\"beyond\",\"start\":1.85,\"end\":2.21}]"}
{"utterance_id": "652-130726-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0017.flac", "task_type": "lexical_detection", "query": "dishes", "prompt": "Process the full audio offline. Find every instance of the word \"dishes\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.83, 4.27, "dishes"]], "response_mode": "offline", "answer": "[{\"word\":\"dishes\",\"start\":3.83,\"end\":4.27}]"}
{"utterance_id": "3081-166546-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.25, "george"]], "response_mode": "offline", "answer": "[{\"word\":\"george\",\"start\":0.52,\"end\":1.25}]"}
{"utterance_id": "652-129742-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0004.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[0.55, 0.72, "one"], [2.72, 2.92, "two"], [5.78, 6.06, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.55,\"end\":0.72},{\"word\":\"two\",\"start\":2.72,\"end\":2.92},{\"word\":\"three\",\"start\":5.78,\"end\":6.06}]"}
{"utterance_id": "2803-154328-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0010.flac", "task_type": "lexical_detection", "query": "major", "prompt": "When the utterance is complete, report all instances of the word \"major\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.42, 1.81, "major"]], "response_mode": "offline", "answer": "[{\"word\":\"major\",\"start\":1.42,\"end\":1.81}]"}
{"utterance_id": "84-121123-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0014.flac", "task_type": "lexical_detection", "query": "leave", "prompt": "After listening to the whole recording, output every instance of the word \"leave\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.03, "leave"]], "response_mode": "offline", "answer": "[{\"word\":\"leave\",\"start\":0.85,\"end\":1.03}]"}
{"utterance_id": "1919-142785-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0046.flac", "task_type": "lexical_detection", "query": "illustration", "prompt": "Once the audio has finished, identify every instance of the word \"illustration\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.52, 1.34, "illustration"]], "response_mode": "offline", "answer": "[{\"word\":\"illustration\",\"start\":0.52,\"end\":1.34}]"}
{"utterance_id": "5895-34622-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0023.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After the audio ends, extract all occurrences of an organization name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.09, 3.72, "company"], [4.83, 5.44, "orchestra"]], "response_mode": "offline", "answer": "[{\"word\":\"company\",\"start\":3.09,\"end\":3.72},{\"word\":\"orchestra\",\"start\":4.83,\"end\":5.44}]"}
{"utterance_id": "5694-64025-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0003.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Once the audio has finished, identify every instance of a date. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.38, 1.72, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":1.38,\"end\":1.72}]"}
{"utterance_id": "6241-61943-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0025.flac", "task_type": "lexical_detection", "query": "heat", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"heat\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.05, 1.32, "heat"]], "response_mode": "offline", "answer": "[{\"word\":\"heat\",\"start\":1.05,\"end\":1.32}]"}
{"utterance_id": "6295-244435-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0021.flac", "task_type": "lexical_detection", "query": "late", "prompt": "After listening to the whole recording, output every instance of the word \"late\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.24, "late"]], "response_mode": "offline", "answer": "[{\"word\":\"late\",\"start\":2.03,\"end\":2.24}]"}
{"utterance_id": "7850-281318-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0009.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an animal name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.55, 6.1, "magpie"]], "response_mode": "offline", "answer": "[{\"word\":\"magpie\",\"start\":5.55,\"end\":6.1}]"}
{"utterance_id": "2428-83705-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0025.flac", "task_type": "lexical_counting", "query": "ought", "prompt": "Listen to the entire clip, then report the count sequence for the word \"ought\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.64, 2.82, "one"], [5.92, 6.16, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.64,\"end\":2.82},{\"word\":\"two\",\"start\":5.92,\"end\":6.16}]"}
{"utterance_id": "251-137823-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0026.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.57, 2.01, "eyes"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":1.57,\"end\":2.01}]"}
{"utterance_id": "6313-66125-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0022.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.26, 0.84, "one"], [6.78, 7.14, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.26,\"end\":0.84},{\"word\":\"two\",\"start\":6.78,\"end\":7.14}]"}
{"utterance_id": "1272-135031-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0007.flac", "task_type": "lexical_detection", "query": "eats", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"eats\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.7, 1.04, "eats"]], "response_mode": "offline", "answer": "[{\"word\":\"eats\",\"start\":0.7,\"end\":1.04}]"}
{"utterance_id": "6345-93306-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0022.flac", "task_type": "lexical_detection", "query": "kiss", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"kiss\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.84, 5.05, "kiss"]], "response_mode": "offline", "answer": "[{\"word\":\"kiss\",\"start\":4.84,\"end\":5.05}]"}
{"utterance_id": "6313-66129-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.01, 6.44, "stacy"]], "response_mode": "offline", "answer": "[{\"word\":\"stacy\",\"start\":6.01,\"end\":6.44}]"}
{"utterance_id": "6313-66129-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0009.flac", "task_type": "lexical_detection", "query": "criticism", "prompt": "After the audio ends, extract all occurrences of the word \"criticism\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.96, 2.66, "criticism"]], "response_mode": "offline", "answer": "[{\"word\":\"criticism\",\"start\":1.96,\"end\":2.66}]"}
{"utterance_id": "84-121550-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0009.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.48, 2.99, "hundred"]], "response_mode": "offline", "answer": "[{\"word\":\"hundred\",\"start\":2.48,\"end\":2.99}]"}
{"utterance_id": "1272-141231-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0006.flac", "task_type": "lexical_detection", "query": "triggered", "prompt": "When the utterance is complete, report all instances of the word \"triggered\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.93, 2.29, "triggered"]], "response_mode": "offline", "answer": "[{\"word\":\"triggered\",\"start\":1.93,\"end\":2.29}]"}
{"utterance_id": "2902-9006-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "When the utterance is complete, report all instances of a number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.27, 0.51, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.27,\"end\":0.51}]"}
{"utterance_id": "1272-128104-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0012.flac", "task_type": "lexical_detection", "query": "unfortunately", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"unfortunately\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 2.1, "unfortunately"]], "response_mode": "offline", "answer": "[{\"word\":\"unfortunately\",\"start\":1.14,\"end\":2.1}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[0.47, 1.07, "one"], [2.31, 2.67, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.47,\"end\":1.07},{\"word\":\"two\",\"start\":2.31,\"end\":2.67}]"}
{"utterance_id": "7976-105575-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0025.flac", "task_type": "lexical_detection", "query": "rank", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"rank\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.06, "rank"]], "response_mode": "offline", "answer": "[{\"word\":\"rank\",\"start\":1.75,\"end\":2.06}]"}
{"utterance_id": "1993-147965-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147965/1993-147965-0008.flac", "task_type": "lexical_detection", "query": "cap", "prompt": "Process the full audio offline. Find every instance of the word \"cap\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.64, "cap"]], "response_mode": "offline", "answer": "[{\"word\":\"cap\",\"start\":3.29,\"end\":3.64}]"}
{"utterance_id": "2277-149896-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0024.flac", "task_type": "lexical_detection", "query": "plan", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"plan\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.32, "plan"]], "response_mode": "offline", "answer": "[{\"word\":\"plan\",\"start\":2.98,\"end\":3.32}]"}
{"utterance_id": "3081-166546-0062", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0062.flac", "task_type": "lexical_detection", "query": "manager", "prompt": "Process the full audio offline. Find every instance of the word \"manager\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.35, 2.94, "manager"]], "response_mode": "offline", "answer": "[{\"word\":\"manager\",\"start\":2.35,\"end\":2.94}]"}
{"utterance_id": "5895-34629-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0021.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.81, "cicero"]], "response_mode": "offline", "answer": "[{\"word\":\"cicero\",\"start\":2.19,\"end\":2.81}]"}
{"utterance_id": "3536-23268-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0015.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.23, 0.67, "niece"]], "response_mode": "offline", "answer": "[{\"word\":\"niece\",\"start\":0.23,\"end\":0.67}]"}
{"utterance_id": "3536-23268-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0009.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.44, 2.96, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":2.44,\"end\":2.96}]"}
{"utterance_id": "5694-64029-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0008.flac", "task_type": "lexical_detection", "query": "step", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"step\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.31, "step"]], "response_mode": "offline", "answer": "[{\"word\":\"step\",\"start\":1.92,\"end\":2.31}]"}
{"utterance_id": "1673-143397-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0019.flac", "task_type": "lexical_detection", "query": "dread", "prompt": "Process the full audio offline. Find every instance of the word \"dread\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.98, 5.33, "dread"]], "response_mode": "offline", "answer": "[{\"word\":\"dread\",\"start\":4.98,\"end\":5.33}]"}
{"utterance_id": "7850-73752-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0000.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.25, 0.84, "ferdinand"]], "response_mode": "offline", "answer": "[{\"word\":\"ferdinand\",\"start\":0.25,\"end\":0.84}]"}
{"utterance_id": "2078-142845-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0014.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.71, 2.18, "wheats"]], "response_mode": "offline", "answer": "[{\"word\":\"wheats\",\"start\":1.71,\"end\":2.18}]"}
{"utterance_id": "2086-149220-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0014.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.54, 2.82, "phoebe"]], "response_mode": "offline", "answer": "[{\"word\":\"phoebe\",\"start\":2.54,\"end\":2.82}]"}
{"utterance_id": "2078-142845-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0031.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 6.25, "one quarter"]], "response_mode": "offline", "answer": "[{\"word\":\"one quarter\",\"start\":5.52,\"end\":6.25}]"}
{"utterance_id": "652-130737-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0010.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Once the audio has finished, list each occurrence of a food name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.51, 1.15, "one"], [4.26, 4.65, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.15},{\"word\":\"two\",\"start\":4.26,\"end\":4.65}]"}
{"utterance_id": "3752-4943-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen carefully to the entire clip, then report every instance of a profession as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.28, "clerk"]], "response_mode": "offline", "answer": "[{\"word\":\"clerk\",\"start\":0.89,\"end\":1.28}]"}
{"utterance_id": "2277-149874-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0000.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Once the audio has finished, identify every instance of a number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.52, 1.76, "one"], [13.81, 14.53, "fifty thousand"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.52,\"end\":1.76},{\"word\":\"fifty thousand\",\"start\":13.81,\"end\":14.53}]"}
{"utterance_id": "1673-143396-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0017.flac", "task_type": "lexical_detection", "query": "faint", "prompt": "Once the audio has finished, identify every instance of the word \"faint\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.92, 6.21, "faint"]], "response_mode": "offline", "answer": "[{\"word\":\"faint\",\"start\":5.92,\"end\":6.21}]"}
{"utterance_id": "422-122949-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0030.flac", "task_type": "lexical_detection", "query": "misunderstand", "prompt": "Process the full audio offline. Find every instance of the word \"misunderstand\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.34, 1.31, "misunderstand"]], "response_mode": "offline", "answer": "[{\"word\":\"misunderstand\",\"start\":0.34,\"end\":1.31}]"}
{"utterance_id": "652-129742-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0014.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for a food name with word/start/end. If none, reply [].", "labels": [[1.36, 1.71, "one"], [2.93, 3.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.36,\"end\":1.71},{\"word\":\"two\",\"start\":2.93,\"end\":3.36}]"}
{"utterance_id": "422-122949-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0021.flac", "task_type": "lexical_detection", "query": "misunderstand", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"misunderstand\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.46, 7.43, "misunderstand"]], "response_mode": "offline", "answer": "[{\"word\":\"misunderstand\",\"start\":6.46,\"end\":7.43}]"}
{"utterance_id": "2086-149220-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0007.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an animal name with word/start/end. If none, reply [].", "labels": [[1.6, 2.34, "one"], [4.39, 4.81, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.6,\"end\":2.34},{\"word\":\"two\",\"start\":4.39,\"end\":4.81}]"}
{"utterance_id": "6313-76958-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0024.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.47, 1.88, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":1.47,\"end\":1.88}]"}
{"utterance_id": "5338-24615-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0004.flac", "task_type": "lexical_detection", "query": "remarkably", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"remarkably\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.59, 11.34, "remarkably"]], "response_mode": "offline", "answer": "[{\"word\":\"remarkably\",\"start\":10.59,\"end\":11.34}]"}
{"utterance_id": "7976-110124-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0005.flac", "task_type": "lexical_detection", "query": "roof", "prompt": "After the audio ends, extract all occurrences of the word \"roof\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.17, 4.71, "roof"]], "response_mode": "offline", "answer": "[{\"word\":\"roof\",\"start\":4.17,\"end\":4.71}]"}
{"utterance_id": "1988-147956-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0002.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.77, "badger"]], "response_mode": "offline", "answer": "[{\"word\":\"badger\",\"start\":1.36,\"end\":1.77}]"}
{"utterance_id": "3853-163249-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0024.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "When the utterance is complete, report all instances of a time of day including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.34, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":0.93,\"end\":1.34}]"}
{"utterance_id": "8842-302201-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0003.flac", "task_type": "lexical_counting", "query": "hath", "prompt": "Listen to the audio in full, then emit a JSON array of count events for the word \"hath\" (word, start, end). If none, reply [].", "labels": [[0.79, 1.08, "one"], [4.75, 4.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.79,\"end\":1.08},{\"word\":\"two\",\"start\":4.75,\"end\":4.97}]"}
{"utterance_id": "1993-147964-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.73, 3.15, "jake"]], "response_mode": "offline", "answer": "[{\"word\":\"jake\",\"start\":2.73,\"end\":3.15}]"}
{"utterance_id": "1988-147956-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0028.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the full audio. Count every occurrence of a body part. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.76, 2.17, "one"], [4.17, 4.52, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.76,\"end\":2.17},{\"word\":\"two\",\"start\":4.17,\"end\":4.52}]"}
{"utterance_id": "2035-147960-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.74, "otto"]], "response_mode": "offline", "answer": "[{\"word\":\"otto\",\"start\":0.42,\"end\":0.74}]"}
{"utterance_id": "2086-149220-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0018.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[2.3, 2.89, "one"], [5.9, 6.39, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.3,\"end\":2.89},{\"word\":\"two\",\"start\":5.9,\"end\":6.39}]"}
{"utterance_id": "3576-138058-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0010.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[9.56, 9.91, "one"], [11.26, 11.69, "two"], [13.14, 13.75, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.56,\"end\":9.91},{\"word\":\"two\",\"start\":11.26,\"end\":11.69},{\"word\":\"three\",\"start\":13.14,\"end\":13.75}]"}
{"utterance_id": "2803-161169-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0007.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.82, 4.83, "tree trunks"], [10.13, 10.91, "palm leaves"]], "response_mode": "offline", "answer": "[{\"word\":\"tree trunks\",\"start\":3.82,\"end\":4.83},{\"word\":\"palm leaves\",\"start\":10.13,\"end\":10.91}]"}
{"utterance_id": "1988-147956-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0026.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After listening to the whole recording, output every instance of a plant name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.79, "grass"], [5.31, 5.61, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"grass\",\"start\":1.28,\"end\":1.79},{\"word\":\"tree\",\"start\":5.31,\"end\":5.61}]"}
{"utterance_id": "6241-61946-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.24, 0.77, "hans"]], "response_mode": "offline", "answer": "[{\"word\":\"hans\",\"start\":0.24,\"end\":0.77}]"}
{"utterance_id": "2412-153954-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0020.flac", "task_type": "lexical_detection", "query": "iron", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"iron\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.71, 7.03, "iron"]], "response_mode": "offline", "answer": "[{\"word\":\"iron\",\"start\":6.71,\"end\":7.03}]"}
{"utterance_id": "84-121550-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0027.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.92, 3.64, "beatrice"]], "response_mode": "offline", "answer": "[{\"word\":\"beatrice\",\"start\":2.92,\"end\":3.64}]"}
{"utterance_id": "6241-61943-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0024.flac", "task_type": "lexical_detection", "query": "winds", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"winds\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.33, 2.81, "winds"]], "response_mode": "offline", "answer": "[{\"word\":\"winds\",\"start\":2.33,\"end\":2.81}]"}
{"utterance_id": "84-121550-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.68, 18.2, "delia's"]], "response_mode": "offline", "answer": "[{\"word\":\"delia's\",\"start\":17.68,\"end\":18.2}]"}
{"utterance_id": "1462-170142-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0004.flac", "task_type": "lexical_detection", "query": "lamps", "prompt": "After listening to the whole recording, output every instance of the word \"lamps\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.07, 3.46, "lamps"]], "response_mode": "offline", "answer": "[{\"word\":\"lamps\",\"start\":3.07,\"end\":3.46}]"}
{"utterance_id": "6241-61943-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0009.flac", "task_type": "lexical_detection", "query": "kept", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"kept\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.08, 1.39, "kept"]], "response_mode": "offline", "answer": "[{\"word\":\"kept\",\"start\":1.08,\"end\":1.39}]"}
{"utterance_id": "8297-275154-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0023.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 2.12, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.52,\"end\":2.12}]"}
{"utterance_id": "5694-64025-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0013.flac", "task_type": "lexical_detection", "query": "monday", "prompt": "Process the full audio offline. Find every instance of the word \"monday\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.27, "monday"]], "response_mode": "offline", "answer": "[{\"word\":\"monday\",\"start\":0.72,\"end\":1.27}]"}
{"utterance_id": "5338-284437-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0017.flac", "task_type": "lexical_detection", "query": "majesty", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"majesty\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.42, "majesty"]], "response_mode": "offline", "answer": "[{\"word\":\"majesty\",\"start\":1.78,\"end\":2.42}]"}
{"utterance_id": "2035-152373-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0003.flac", "task_type": "lexical_detection", "query": "poets", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"poets\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.6, 1.14, "poets"]], "response_mode": "offline", "answer": "[{\"word\":\"poets\",\"start\":0.6,\"end\":1.14}]"}
{"utterance_id": "8297-275154-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0012.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.62, 4.09, "sydney"]], "response_mode": "offline", "answer": "[{\"word\":\"sydney\",\"start\":3.62,\"end\":4.09}]"}
{"utterance_id": "7850-111771-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0004.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.58, "general"]], "response_mode": "offline", "answer": "[{\"word\":\"general\",\"start\":1.22,\"end\":1.58}]"}
{"utterance_id": "2277-149896-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0018.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Listen carefully to the entire clip, then report every instance of an animal name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.79, 4.26, "devil"]], "response_mode": "offline", "answer": "[{\"word\":\"devil\",\"start\":3.79,\"end\":4.26}]"}
{"utterance_id": "1993-147964-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0001.flac", "task_type": "lexical_detection", "query": "allow", "prompt": "After the audio ends, extract all occurrences of the word \"allow\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.59, 1.92, "allow"]], "response_mode": "offline", "answer": "[{\"word\":\"allow\",\"start\":1.59,\"end\":1.92}]"}
{"utterance_id": "7976-110523-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0005.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.76, 4.2, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cat\",\"start\":3.76,\"end\":4.2}]"}
{"utterance_id": "1988-147956-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0026.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.01, 3.27, "blue"], [4.97, 5.31, "gold"]], "response_mode": "offline", "answer": "[{\"word\":\"blue\",\"start\":3.01,\"end\":3.27},{\"word\":\"gold\",\"start\":4.97,\"end\":5.31}]"}
{"utterance_id": "6241-61943-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0015.flac", "task_type": "lexical_detection", "query": "professor", "prompt": "After listening to the whole recording, output every instance of the word \"professor\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.9, "professor"]], "response_mode": "offline", "answer": "[{\"word\":\"professor\",\"start\":0.46,\"end\":0.9}]"}
{"utterance_id": "6241-66616-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0008.flac", "task_type": "lexical_detection", "query": "curiously", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"curiously\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.35, 0.9, "curiously"]], "response_mode": "offline", "answer": "[{\"word\":\"curiously\",\"start\":0.35,\"end\":0.9}]"}
{"utterance_id": "2078-142845-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0006.flac", "task_type": "lexical_detection", "query": "risen", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"risen\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.67, 7.08, "risen"]], "response_mode": "offline", "answer": "[{\"word\":\"risen\",\"start\":6.67,\"end\":7.08}]"}
{"utterance_id": "6319-275224-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0011.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.71, 5.48, "convolvulus"], [9.04, 9.5, "lily"], [10.91, 12.13, "carnation flowers"], [14.27, 14.89, "rose tree"], [16.34, 17.09, "sweet peas"]], "response_mode": "offline", "answer": "[{\"word\":\"convolvulus\",\"start\":4.71,\"end\":5.48},{\"word\":\"lily\",\"start\":9.04,\"end\":9.5},{\"word\":\"carnation flowers\",\"start\":10.91,\"end\":12.13},{\"word\":\"rose tree\",\"start\":14.27,\"end\":14.89},{\"word\":\"sweet peas\",\"start\":16.34,\"end\":17.09}]"}
{"utterance_id": "84-121550-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[12.68, 13.38, "sevenfold"]], "response_mode": "offline", "answer": "[{\"word\":\"sevenfold\",\"start\":12.68,\"end\":13.38}]"}
{"utterance_id": "3170-137482-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0000.flac", "task_type": "lexical_detection", "query": "literary", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"literary\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.67, 8.26, "literary"]], "response_mode": "offline", "answer": "[{\"word\":\"literary\",\"start\":7.67,\"end\":8.26}]"}
{"utterance_id": "5694-64029-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0021.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After listening to the whole recording, output every instance of an ordinal number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.06, "fourth"]], "response_mode": "offline", "answer": "[{\"word\":\"fourth\",\"start\":1.74,\"end\":2.06}]"}
{"utterance_id": "2086-149220-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0029.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.67, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":2.18,\"end\":2.67}]"}
{"utterance_id": "6241-66616-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0004.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.64, 3.0, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":2.64,\"end\":3.0}]"}
{"utterance_id": "3081-166546-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0046.flac", "task_type": "lexical_detection", "query": "carefully", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"carefully\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.88, 3.46, "carefully"]], "response_mode": "offline", "answer": "[{\"word\":\"carefully\",\"start\":2.88,\"end\":3.46}]"}
{"utterance_id": "3170-137482-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0046.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.96, 8.41, "sequins"]], "response_mode": "offline", "answer": "[{\"word\":\"sequins\",\"start\":7.96,\"end\":8.41}]"}
{"utterance_id": "5536-43358-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen carefully to the entire clip, then report every instance of a body part as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.53, 6.79, "womb"]], "response_mode": "offline", "answer": "[{\"word\":\"womb\",\"start\":6.53,\"end\":6.79}]"}
{"utterance_id": "5895-34622-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0012.flac", "task_type": "lexical_detection", "query": "curiosity", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"curiosity\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.4, 1.29, "curiosity"]], "response_mode": "offline", "answer": "[{\"word\":\"curiosity\",\"start\":0.4,\"end\":1.29}]"}
{"utterance_id": "3081-166546-0077", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0077.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.48, 4.0, "face"]], "response_mode": "offline", "answer": "[{\"word\":\"face\",\"start\":3.48,\"end\":4.0}]"}
{"utterance_id": "8842-302196-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0007.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.91, 1.14, "twenty"]], "response_mode": "offline", "answer": "[{\"word\":\"twenty\",\"start\":0.91,\"end\":1.14}]"}
{"utterance_id": "8297-275155-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0007.flac", "task_type": "lexical_detection", "query": "nice", "prompt": "After the audio ends, extract all occurrences of the word \"nice\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.76, 1.18, "nice"]], "response_mode": "offline", "answer": "[{\"word\":\"nice\",\"start\":0.76,\"end\":1.18}]"}
{"utterance_id": "6319-275224-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0006.flac", "task_type": "lexical_detection", "query": "lily", "prompt": "Once the audio has finished, identify every instance of the word \"lily\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.56, 5.99, "lily"]], "response_mode": "offline", "answer": "[{\"word\":\"lily\",\"start\":5.56,\"end\":5.99}]"}
{"utterance_id": "5338-24640-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0001.flac", "task_type": "lexical_detection", "query": "claim", "prompt": "Process the full audio offline. Find every instance of the word \"claim\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.51, 3.85, "claim"]], "response_mode": "offline", "answer": "[{\"word\":\"claim\",\"start\":3.51,\"end\":3.85}]"}
{"utterance_id": "652-130726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0018.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a food name with fields word, start, end (seconds). If none, reply [].", "labels": [[1.55, 1.92, "dinners"]], "response_mode": "offline", "answer": "[{\"word\":\"dinners\",\"start\":1.55,\"end\":1.92}]"}
{"utterance_id": "1673-143396-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0005.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.51, 0.81, "son"], [10.47, 10.88, "children"]], "response_mode": "offline", "answer": "[{\"word\":\"son\",\"start\":0.51,\"end\":0.81},{\"word\":\"children\",\"start\":10.47,\"end\":10.88}]"}
{"utterance_id": "5338-284437-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0032.flac", "task_type": "lexical_detection", "query": "puzzled", "prompt": "When the utterance is complete, report all instances of the word \"puzzled\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.48, 8.15, "puzzled"]], "response_mode": "offline", "answer": "[{\"word\":\"puzzled\",\"start\":7.48,\"end\":8.15}]"}
{"utterance_id": "6345-93302-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0026.flac", "task_type": "lexical_detection", "query": "marble", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"marble\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.54, 1.89, "marble"]], "response_mode": "offline", "answer": "[{\"word\":\"marble\",\"start\":1.54,\"end\":1.89}]"}
{"utterance_id": "6319-275224-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0017.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.24, "weeds"]], "response_mode": "offline", "answer": "[{\"word\":\"weeds\",\"start\":0.53,\"end\":1.24}]"}
{"utterance_id": "2803-154328-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0014.flac", "task_type": "lexical_detection", "query": "answered", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"answered\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.73, 2.08, "answered"]], "response_mode": "offline", "answer": "[{\"word\":\"answered\",\"start\":1.73,\"end\":2.08}]"}
{"utterance_id": "1673-143396-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0013.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.33, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.0,\"end\":2.33}]"}
{"utterance_id": "6241-61946-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0007.flac", "task_type": "lexical_detection", "query": "snow", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"snow\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 1.02, "snow"]], "response_mode": "offline", "answer": "[{\"word\":\"snow\",\"start\":0.4,\"end\":1.02}]"}
{"utterance_id": "6241-61943-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 0.9, "professor"]], "response_mode": "offline", "answer": "[{\"word\":\"professor\",\"start\":0.46,\"end\":0.9}]"}
{"utterance_id": "1993-147149-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0020.flac", "task_type": "lexical_counting", "query": "mary", "prompt": "Once the audio has finished, list each occurrence of the word \"mary\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.51, 0.95, "one"], [10.92, 11.43, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.95},{\"word\":\"two\",\"start\":10.92,\"end\":11.43}]"}
{"utterance_id": "7976-110124-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0013.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.79, 2.44, "apples"], [5.37, 5.67, "apple"]], "response_mode": "offline", "answer": "[{\"word\":\"apples\",\"start\":1.79,\"end\":2.44},{\"word\":\"apple\",\"start\":5.37,\"end\":5.67}]"}
{"utterance_id": "1673-143396-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0003.flac", "task_type": "lexical_detection", "query": "stature", "prompt": "Process the full audio offline. Find every instance of the word \"stature\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.15, 5.61, "stature"]], "response_mode": "offline", "answer": "[{\"word\":\"stature\",\"start\":5.15,\"end\":5.61}]"}
{"utterance_id": "6319-64726-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0003.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.29, 3.58, "elder"]], "response_mode": "offline", "answer": "[{\"word\":\"elder\",\"start\":3.29,\"end\":3.58}]"}
{"utterance_id": "6345-93302-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0014.flac", "task_type": "lexical_detection", "query": "seems", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"seems\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.92, 2.17, "seems"]], "response_mode": "offline", "answer": "[{\"word\":\"seems\",\"start\":1.92,\"end\":2.17}]"}
{"utterance_id": "7850-286674-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.47, 3.04, "larvae"]], "response_mode": "offline", "answer": "[{\"word\":\"larvae\",\"start\":2.47,\"end\":3.04}]"}
{"utterance_id": "777-126732-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0027.flac", "task_type": "semantic_detection", "query": "color", "prompt": "After listening to the whole recording, output every instance of a color name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.22, 2.4, "red"]], "response_mode": "offline", "answer": "[{\"word\":\"red\",\"start\":2.22,\"end\":2.4}]"}
{"utterance_id": "5338-284437-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0024.flac", "task_type": "lexical_detection", "query": "laws", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"laws\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.63, 3.22, "laws"]], "response_mode": "offline", "answer": "[{\"word\":\"laws\",\"start\":2.63,\"end\":3.22}]"}
{"utterance_id": "3536-23268-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0023.flac", "task_type": "lexical_detection", "query": "encounter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"encounter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.24, 6.78, "encounter"]], "response_mode": "offline", "answer": "[{\"word\":\"encounter\",\"start\":6.24,\"end\":6.78}]"}
{"utterance_id": "422-122949-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0029.flac", "task_type": "lexical_detection", "query": "offer", "prompt": "Process the full audio offline. Find every instance of the word \"offer\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.1, 2.45, "offer"]], "response_mode": "offline", "answer": "[{\"word\":\"offer\",\"start\":2.1,\"end\":2.45}]"}
{"utterance_id": "1673-143396-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.07, 4.38, "son"]], "response_mode": "offline", "answer": "[{\"word\":\"son\",\"start\":4.07,\"end\":4.38}]"}
{"utterance_id": "5536-43363-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0005.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a food name (word, start, end). If none, reply [].", "labels": [[9.57, 9.91, "one"], [14.29, 14.68, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":9.57,\"end\":9.91},{\"word\":\"two\",\"start\":14.29,\"end\":14.68}]"}
{"utterance_id": "5694-64038-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0009.flac", "task_type": "lexical_detection", "query": "overhead", "prompt": "After the audio ends, extract all occurrences of the word \"overhead\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 2.18, "overhead"]], "response_mode": "offline", "answer": "[{\"word\":\"overhead\",\"start\":1.41,\"end\":2.18}]"}
{"utterance_id": "2803-154328-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 4.08, "mary grant"], [7.49, 8.24, "glenarvan"]], "response_mode": "offline", "answer": "[{\"word\":\"mary grant\",\"start\":3.03,\"end\":4.08},{\"word\":\"glenarvan\",\"start\":7.49,\"end\":8.24}]"}
{"utterance_id": "6241-61946-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0011.flac", "task_type": "lexical_detection", "query": "isolated", "prompt": "Process the full audio offline. Find every instance of the word \"isolated\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.7, "isolated"]], "response_mode": "offline", "answer": "[{\"word\":\"isolated\",\"start\":1.15,\"end\":1.7}]"}
{"utterance_id": "3576-138058-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.25, 0.98, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":0.25,\"end\":0.98}]"}
{"utterance_id": "7976-110523-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0001.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.17, 1.71, "children"]], "response_mode": "offline", "answer": "[{\"word\":\"children\",\"start\":1.17,\"end\":1.71}]"}
{"utterance_id": "5895-34622-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0011.flac", "task_type": "lexical_detection", "query": "journey", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"journey\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.99, 10.48, "journey"]], "response_mode": "offline", "answer": "[{\"word\":\"journey\",\"start\":9.99,\"end\":10.48}]"}
{"utterance_id": "2277-149896-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.4, 2.71, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":2.4,\"end\":2.71}]"}
{"utterance_id": "174-84280-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/84280/174-84280-0014.flac", "task_type": "lexical_detection", "query": "alternative", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"alternative\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.5, "alternative"]], "response_mode": "offline", "answer": "[{\"word\":\"alternative\",\"start\":0.89,\"end\":1.5}]"}
{"utterance_id": "2428-83699-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.5, "madge"]], "response_mode": "offline", "answer": "[{\"word\":\"madge\",\"start\":4.99,\"end\":5.5}]"}
{"utterance_id": "2428-83699-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.69, 10.17, "cousin"]], "response_mode": "offline", "answer": "[{\"word\":\"cousin\",\"start\":9.69,\"end\":10.17}]"}
{"utterance_id": "251-118436-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0008.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.53, "fingernail"]], "response_mode": "offline", "answer": "[{\"word\":\"fingernail\",\"start\":1.82,\"end\":2.53}]"}
{"utterance_id": "5694-64025-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0011.flac", "task_type": "lexical_detection", "query": "charge", "prompt": "Process the full audio offline. Find every instance of the word \"charge\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.93, 5.34, "charge"]], "response_mode": "offline", "answer": "[{\"word\":\"charge\",\"start\":4.93,\"end\":5.34}]"}
{"utterance_id": "3000-15664-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0021.flac", "task_type": "lexical_detection", "query": "belt", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"belt\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.35, 6.63, "belt"]], "response_mode": "offline", "answer": "[{\"word\":\"belt\",\"start\":6.35,\"end\":6.63}]"}
{"utterance_id": "1272-141231-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0020.flac", "task_type": "lexical_detection", "query": "stronger", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"stronger\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.21, 3.82, "stronger"]], "response_mode": "offline", "answer": "[{\"word\":\"stronger\",\"start\":3.21,\"end\":3.82}]"}
{"utterance_id": "5536-43363-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43363/5536-43363-0018.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the entire clip, then report the count sequence for a number as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.57, 0.84, "one"], [2.02, 2.32, "two"], [3.44, 3.86, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.57,\"end\":0.84},{\"word\":\"two\",\"start\":2.02,\"end\":2.32},{\"word\":\"three\",\"start\":3.44,\"end\":3.86}]"}
{"utterance_id": "1462-170142-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0042.flac", "task_type": "lexical_detection", "query": "whispered", "prompt": "After listening to the whole recording, output every instance of the word \"whispered\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.5, 3.04, "whispered"]], "response_mode": "offline", "answer": "[{\"word\":\"whispered\",\"start\":2.5,\"end\":3.04}]"}
{"utterance_id": "251-136532-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0001.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After listening to the whole recording, output every instance of a person's name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.46, 1.3, "bill chandler"]], "response_mode": "offline", "answer": "[{\"word\":\"bill chandler\",\"start\":0.46,\"end\":1.3}]"}
{"utterance_id": "2412-153947-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0014.flac", "task_type": "lexical_detection", "query": "moulds", "prompt": "When the utterance is complete, report all instances of the word \"moulds\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.13, 3.42, "moulds"]], "response_mode": "offline", "answer": "[{\"word\":\"moulds\",\"start\":3.13,\"end\":3.42}]"}
{"utterance_id": "2086-149220-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0017.flac", "task_type": "lexical_counting", "query": "hens", "prompt": "When the utterance is complete, report how the word \"hens\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.78, "one"], [3.5, 3.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.39,\"end\":1.78},{\"word\":\"two\",\"start\":3.5,\"end\":3.73}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "lexical_detection", "query": "southern", "prompt": "After listening to the whole recording, output every instance of the word \"southern\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.02, 2.36, "southern"]], "response_mode": "offline", "answer": "[{\"word\":\"southern\",\"start\":2.02,\"end\":2.36}]"}
{"utterance_id": "6295-244435-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0014.flac", "task_type": "lexical_detection", "query": "wrote", "prompt": "Process the full audio offline. Find every instance of the word \"wrote\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.4, "wrote"]], "response_mode": "offline", "answer": "[{\"word\":\"wrote\",\"start\":1.19,\"end\":1.4}]"}
{"utterance_id": "84-121550-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0023.flac", "task_type": "lexical_detection", "query": "reliance", "prompt": "When the utterance is complete, report all instances of the word \"reliance\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.26, 2.84, "reliance"]], "response_mode": "offline", "answer": "[{\"word\":\"reliance\",\"start\":2.26,\"end\":2.84}]"}
{"utterance_id": "8842-302196-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0007.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Once the audio has finished, identify every instance of a date. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.55, 0.91, "january"]], "response_mode": "offline", "answer": "[{\"word\":\"january\",\"start\":0.55,\"end\":0.91}]"}
{"utterance_id": "2277-149897-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0010.flac", "task_type": "lexical_detection", "query": "beaten", "prompt": "Process the full audio offline. Find every instance of the word \"beaten\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.15, "beaten"]], "response_mode": "offline", "answer": "[{\"word\":\"beaten\",\"start\":0.82,\"end\":1.15}]"}
{"utterance_id": "422-122949-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0001.flac", "task_type": "lexical_detection", "query": "values", "prompt": "After listening to the whole recording, output every instance of the word \"values\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.47, "values"]], "response_mode": "offline", "answer": "[{\"word\":\"values\",\"start\":1.92,\"end\":2.47}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.02, 3.4, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":3.02,\"end\":3.4}]"}
{"utterance_id": "7976-105575-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.47, 1.07, "fifteen"], [2.31, 2.67, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"fifteen\",\"start\":0.47,\"end\":1.07},{\"word\":\"half\",\"start\":2.31,\"end\":2.67}]"}
{"utterance_id": "8297-275155-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.44, 3.82, "kitty"]], "response_mode": "offline", "answer": "[{\"word\":\"kitty\",\"start\":3.44,\"end\":3.82}]"}
{"utterance_id": "2035-147961-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0037.flac", "task_type": "lexical_detection", "query": "kissed", "prompt": "Once the audio has finished, identify every instance of the word \"kissed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.63, 1.92, "kissed"]], "response_mode": "offline", "answer": "[{\"word\":\"kissed\",\"start\":1.63,\"end\":1.92}]"}
{"utterance_id": "6295-244435-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0026.flac", "task_type": "lexical_detection", "query": "buried", "prompt": "After the audio ends, extract all occurrences of the word \"buried\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.82, 1.26, "buried"]], "response_mode": "offline", "answer": "[{\"word\":\"buried\",\"start\":0.82,\"end\":1.26}]"}
{"utterance_id": "1919-142785-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0047.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.03, 1.4, "herbs"]], "response_mode": "offline", "answer": "[{\"word\":\"herbs\",\"start\":1.03,\"end\":1.4}]"}
{"utterance_id": "3853-163249-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0012.flac", "task_type": "lexical_detection", "query": "roses", "prompt": "When the utterance is complete, report all instances of the word \"roses\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.71, "roses"]], "response_mode": "offline", "answer": "[{\"word\":\"roses\",\"start\":3.26,\"end\":3.71}]"}
{"utterance_id": "1462-170142-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0009.flac", "task_type": "lexical_detection", "query": "thrust", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"thrust\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.08, "thrust"]], "response_mode": "offline", "answer": "[{\"word\":\"thrust\",\"start\":3.75,\"end\":4.08}]"}
{"utterance_id": "7976-105575-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0022.flac", "task_type": "lexical_detection", "query": "close", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"close\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.6, 2.05, "close"]], "response_mode": "offline", "answer": "[{\"word\":\"close\",\"start\":1.6,\"end\":2.05}]"}
{"utterance_id": "251-137823-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0003.flac", "task_type": "lexical_detection", "query": "engineer", "prompt": "After the audio ends, extract all occurrences of the word \"engineer\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.78, 1.22, "engineer"]], "response_mode": "offline", "answer": "[{\"word\":\"engineer\",\"start\":0.78,\"end\":1.22}]"}
{"utterance_id": "7850-286674-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0015.flac", "task_type": "lexical_detection", "query": "dinner", "prompt": "After the audio ends, extract all occurrences of the word \"dinner\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.78, 3.21, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":2.78,\"end\":3.21}]"}
{"utterance_id": "251-136532-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0019.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.55, 7.09, "terra"]], "response_mode": "offline", "answer": "[{\"word\":\"terra\",\"start\":6.55,\"end\":7.09}]"}
{"utterance_id": "777-126732-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0002.flac", "task_type": "lexical_detection", "query": "value", "prompt": "When the utterance is complete, report all instances of the word \"value\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.37, 4.84, "value"]], "response_mode": "offline", "answer": "[{\"word\":\"value\",\"start\":4.37,\"end\":4.84}]"}
{"utterance_id": "7976-105575-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0012.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "After listening to the whole recording, output every instance of a state or province name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.92, 4.68, "mississippi"]], "response_mode": "offline", "answer": "[{\"word\":\"mississippi\",\"start\":3.92,\"end\":4.68}]"}
{"utterance_id": "3000-15664-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0012.flac", "task_type": "lexical_detection", "query": "fragrance", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"fragrance\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.05, 4.83, "fragrance"]], "response_mode": "offline", "answer": "[{\"word\":\"fragrance\",\"start\":4.05,\"end\":4.83}]"}
{"utterance_id": "1462-170142-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0010.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.78, 2.3, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":1.78,\"end\":2.3}]"}
{"utterance_id": "2277-149897-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0015.flac", "task_type": "lexical_detection", "query": "quarters", "prompt": "Once the audio has finished, identify every instance of the word \"quarters\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.54, 1.95, "quarters"]], "response_mode": "offline", "answer": "[{\"word\":\"quarters\",\"start\":1.54,\"end\":1.95}]"}
{"utterance_id": "2035-147961-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0029.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen carefully to the entire clip, then report every instance of a number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.98, 3.38, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":2.98,\"end\":3.38}]"}
{"utterance_id": "1993-147149-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0012.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.3, 1.76, "husband"]], "response_mode": "offline", "answer": "[{\"word\":\"husband\",\"start\":1.3,\"end\":1.76}]"}
{"utterance_id": "2277-149897-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0031.flac", "task_type": "lexical_detection", "query": "troubled", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"troubled\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.4, 0.77, "troubled"]], "response_mode": "offline", "answer": "[{\"word\":\"troubled\",\"start\":0.4,\"end\":0.77}]"}
{"utterance_id": "2078-142845-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0036.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.84, 1.62, "buns"]], "response_mode": "offline", "answer": "[{\"word\":\"buns\",\"start\":0.84,\"end\":1.62}]"}
{"utterance_id": "3000-15664-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0035.flac", "task_type": "lexical_detection", "query": "confluence", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"confluence\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.14, 10.89, "confluence"]], "response_mode": "offline", "answer": "[{\"word\":\"confluence\",\"start\":10.14,\"end\":10.89}]"}
{"utterance_id": "84-121550-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0002.flac", "task_type": "lexical_detection", "query": "remember", "prompt": "Process the full audio offline. Find every instance of the word \"remember\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.23, 1.72, "remember"]], "response_mode": "offline", "answer": "[{\"word\":\"remember\",\"start\":1.23,\"end\":1.72}]"}
{"utterance_id": "3170-137482-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0040.flac", "task_type": "lexical_detection", "query": "inspired", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"inspired\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.73, 11.36, "inspired"]], "response_mode": "offline", "answer": "[{\"word\":\"inspired\",\"start\":10.73,\"end\":11.36}]"}
{"utterance_id": "5694-64025-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0005.flac", "task_type": "lexical_detection", "query": "advance", "prompt": "When the utterance is complete, report all instances of the word \"advance\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.69, 6.27, "advance"]], "response_mode": "offline", "answer": "[{\"word\":\"advance\",\"start\":5.69,\"end\":6.27}]"}
{"utterance_id": "5694-64025-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0015.flac", "task_type": "lexical_detection", "query": "desire", "prompt": "Process the full audio offline. Find every instance of the word \"desire\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.26, "desire"]], "response_mode": "offline", "answer": "[{\"word\":\"desire\",\"start\":0.85,\"end\":1.26}]"}
{"utterance_id": "3536-8226-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0021.flac", "task_type": "lexical_detection", "query": "trevelyan", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"trevelyan\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.07, 4.67, "trevelyan"]], "response_mode": "offline", "answer": "[{\"word\":\"trevelyan\",\"start\":4.07,\"end\":4.67}]"}
{"utterance_id": "7976-110124-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0018.flac", "task_type": "lexical_detection", "query": "finally", "prompt": "When the utterance is complete, report all instances of the word \"finally\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.38, 5.85, "finally"]], "response_mode": "offline", "answer": "[{\"word\":\"finally\",\"start\":5.38,\"end\":5.85}]"}
{"utterance_id": "2078-142845-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0017.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Once the audio has finished, identify every instance of a food name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.58, 3.99, "grain"]], "response_mode": "offline", "answer": "[{\"word\":\"grain\",\"start\":3.58,\"end\":3.99}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After the audio ends, extract all occurrences of a profession. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.31, 5.87, "minister"], [6.96, 7.38, "minister"]], "response_mode": "offline", "answer": "[{\"word\":\"minister\",\"start\":5.31,\"end\":5.87},{\"word\":\"minister\",\"start\":6.96,\"end\":7.38}]"}
{"utterance_id": "2277-149897-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0011.flac", "task_type": "lexical_detection", "query": "envelope", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"envelope\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.9, 1.37, "envelope"]], "response_mode": "offline", "answer": "[{\"word\":\"envelope\",\"start\":0.9,\"end\":1.37}]"}
{"utterance_id": "1993-147966-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147966/1993-147966-0000.flac", "task_type": "lexical_detection", "query": "barn", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"barn\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[9.19, 9.5, "barn"]], "response_mode": "offline", "answer": "[{\"word\":\"barn\",\"start\":9.19,\"end\":9.5}]"}
{"utterance_id": "6345-64257-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0018.flac", "task_type": "lexical_detection", "query": "mist", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mist\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.37, 2.86, "mist"]], "response_mode": "offline", "answer": "[{\"word\":\"mist\",\"start\":2.37,\"end\":2.86}]"}
{"utterance_id": "652-130726-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0016.flac", "task_type": "lexical_counting", "query": "near", "prompt": "After the audio ends, tally every instance of the word \"near\". Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[3.96, 4.12, "one"], [5.76, 5.97, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.96,\"end\":4.12},{\"word\":\"two\",\"start\":5.76,\"end\":5.97}]"}
{"utterance_id": "1919-142785-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0006.flac", "task_type": "lexical_detection", "query": "cucumber", "prompt": "After the audio ends, extract all occurrences of the word \"cucumber\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.74, 2.28, "cucumber"]], "response_mode": "offline", "answer": "[{\"word\":\"cucumber\",\"start\":1.74,\"end\":2.28}]"}
{"utterance_id": "3536-8226-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.32, 0.86, "trevelyan"]], "response_mode": "offline", "answer": "[{\"word\":\"trevelyan\",\"start\":0.32,\"end\":0.86}]"}
{"utterance_id": "2086-149220-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0002.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.1, 3.73, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":3.1,\"end\":3.73}]"}
{"utterance_id": "8297-275155-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0022.flac", "task_type": "lexical_detection", "query": "kitty", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"kitty\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.44, 3.82, "kitty"]], "response_mode": "offline", "answer": "[{\"word\":\"kitty\",\"start\":3.44,\"end\":3.82}]"}
{"utterance_id": "2078-142845-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0015.flac", "task_type": "lexical_counting", "query": "water", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"water\" with word/start/end. If none, reply [].", "labels": [[2.49, 2.94, "one"], [5.77, 6.29, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.49,\"end\":2.94},{\"word\":\"two\",\"start\":5.77,\"end\":6.29}]"}
{"utterance_id": "8297-275154-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0023.flac", "task_type": "lexical_detection", "query": "let", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"let\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.65, "let"]], "response_mode": "offline", "answer": "[{\"word\":\"let\",\"start\":0.45,\"end\":0.65}]"}
{"utterance_id": "2035-152373-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0003.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.14, "poets"]], "response_mode": "offline", "answer": "[{\"word\":\"poets\",\"start\":0.6,\"end\":1.14}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.54, 0.71, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.54,\"end\":0.71}]"}
{"utterance_id": "3536-23268-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0012.flac", "task_type": "lexical_detection", "query": "attracted", "prompt": "Once the audio has finished, identify every instance of the word \"attracted\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.95, 6.41, "attracted"]], "response_mode": "offline", "answer": "[{\"word\":\"attracted\",\"start\":5.95,\"end\":6.41}]"}
{"utterance_id": "2035-152373-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0011.flac", "task_type": "lexical_detection", "query": "monastic", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"monastic\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.8, 4.34, "monastic"]], "response_mode": "offline", "answer": "[{\"word\":\"monastic\",\"start\":3.8,\"end\":4.34}]"}
{"utterance_id": "2412-153954-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0022.flac", "task_type": "lexical_detection", "query": "design", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"design\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.65, 1.03, "design"]], "response_mode": "offline", "answer": "[{\"word\":\"design\",\"start\":0.65,\"end\":1.03}]"}
{"utterance_id": "6345-93306-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0013.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the full audio. After it finishes, find every instance of a city name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.47, 7.0, "paris"]], "response_mode": "offline", "answer": "[{\"word\":\"paris\",\"start\":6.47,\"end\":7.0}]"}
{"utterance_id": "777-126732-0073", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0073.flac", "task_type": "lexical_detection", "query": "what's", "prompt": "Process the full audio offline. Find every instance of the word \"what's\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.49, "what's"]], "response_mode": "offline", "answer": "[{\"word\":\"what's\",\"start\":1.28,\"end\":1.49}]"}
{"utterance_id": "2035-152373-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[17.45, 18.06, "patrick"]], "response_mode": "offline", "answer": "[{\"word\":\"patrick\",\"start\":17.45,\"end\":18.06}]"}
{"utterance_id": "2428-83699-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.51, 8.7, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":8.51,\"end\":8.7}]"}
{"utterance_id": "2035-147961-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0029.flac", "task_type": "lexical_detection", "query": "middle", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"middle\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.87, 1.1, "middle"]], "response_mode": "offline", "answer": "[{\"word\":\"middle\",\"start\":0.87,\"end\":1.1}]"}
{"utterance_id": "6345-93306-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0016.flac", "task_type": "lexical_detection", "query": "happened", "prompt": "After the audio ends, extract all occurrences of the word \"happened\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.82, 3.08, "happened"]], "response_mode": "offline", "answer": "[{\"word\":\"happened\",\"start\":2.82,\"end\":3.08}]"}
{"utterance_id": "6241-61943-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.77, 1.14, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.77,\"end\":1.14}]"}
{"utterance_id": "3576-138058-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0027.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.54, 1.06, "servants"]], "response_mode": "offline", "answer": "[{\"word\":\"servants\",\"start\":0.54,\"end\":1.06}]"}
{"utterance_id": "7976-105575-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.45, 6.23, "jimmy king"]], "response_mode": "offline", "answer": "[{\"word\":\"jimmy king\",\"start\":5.45,\"end\":6.23}]"}
{"utterance_id": "3536-23268-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0022.flac", "task_type": "lexical_detection", "query": "lest", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"lest\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.61, 3.91, "lest"]], "response_mode": "offline", "answer": "[{\"word\":\"lest\",\"start\":3.61,\"end\":3.91}]"}
{"utterance_id": "5694-64038-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0003.flac", "task_type": "lexical_detection", "query": "track", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"track\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.87, 5.31, "track"]], "response_mode": "offline", "answer": "[{\"word\":\"track\",\"start\":4.87,\"end\":5.31}]"}
{"utterance_id": "2078-142845-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0040.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.05, "twelve"]], "response_mode": "offline", "answer": "[{\"word\":\"twelve\",\"start\":1.6,\"end\":2.05}]"}
{"utterance_id": "777-126732-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0031.flac", "task_type": "lexical_detection", "query": "sheet", "prompt": "Once the audio has finished, identify every instance of the word \"sheet\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.65, 0.82, "sheet"]], "response_mode": "offline", "answer": "[{\"word\":\"sheet\",\"start\":0.65,\"end\":0.82}]"}
{"utterance_id": "2803-154320-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0010.flac", "task_type": "lexical_detection", "query": "course", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"course\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.45, "course"]], "response_mode": "offline", "answer": "[{\"word\":\"course\",\"start\":1.89,\"end\":2.45}]"}
{"utterance_id": "5694-64025-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0023.flac", "task_type": "lexical_detection", "query": "mighty", "prompt": "Process the full audio offline. Find every instance of the word \"mighty\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.17, 5.5, "mighty"]], "response_mode": "offline", "answer": "[{\"word\":\"mighty\",\"start\":5.17,\"end\":5.5}]"}
{"utterance_id": "251-136532-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0006.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 2.26, "martians"]], "response_mode": "offline", "answer": "[{\"word\":\"martians\",\"start\":1.31,\"end\":2.26}]"}
{"utterance_id": "7976-110124-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0006.flac", "task_type": "lexical_detection", "query": "tiles", "prompt": "When the utterance is complete, report all instances of the word \"tiles\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.94, 6.47, "tiles"]], "response_mode": "offline", "answer": "[{\"word\":\"tiles\",\"start\":5.94,\"end\":6.47}]"}
{"utterance_id": "6319-57405-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0003.flac", "task_type": "lexical_detection", "query": "pouring", "prompt": "After the audio ends, extract all occurrences of the word \"pouring\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.48, 4.93, "pouring"]], "response_mode": "offline", "answer": "[{\"word\":\"pouring\",\"start\":4.48,\"end\":4.93}]"}
{"utterance_id": "5338-24615-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0001.flac", "task_type": "semantic_detection", "query": "country", "prompt": "After listening to the whole recording, output every instance of a country name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.07, 6.48, "english"]], "response_mode": "offline", "answer": "[{\"word\":\"english\",\"start\":6.07,\"end\":6.48}]"}
{"utterance_id": "1988-147956-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0016.flac", "task_type": "lexical_detection", "query": "bank", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bank\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 2.98, "bank"]], "response_mode": "offline", "answer": "[{\"word\":\"bank\",\"start\":2.55,\"end\":2.98}]"}
{"utterance_id": "2035-152373-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0009.flac", "task_type": "semantic_detection", "query": "country", "prompt": "You will hear a complete utterance. When it ends, list every instance of a country name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.19, 5.62, "france"]], "response_mode": "offline", "answer": "[{\"word\":\"france\",\"start\":5.19,\"end\":5.62}]"}
{"utterance_id": "8297-275155-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0019.flac", "task_type": "lexical_detection", "query": "printed", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"printed\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.94, "printed"], [5.24, 5.78, "printed"]], "response_mode": "offline", "answer": "[{\"word\":\"printed\",\"start\":1.52,\"end\":1.94},{\"word\":\"printed\",\"start\":5.24,\"end\":5.78}]"}
{"utterance_id": "3752-4943-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0028.flac", "task_type": "lexical_detection", "query": "tied", "prompt": "When the utterance is complete, report all instances of the word \"tied\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.69, "tied"]], "response_mode": "offline", "answer": "[{\"word\":\"tied\",\"start\":2.37,\"end\":2.69}]"}
{"utterance_id": "2078-142845-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0035.flac", "task_type": "lexical_detection", "query": "sufficient", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"sufficient\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.26, 1.09, "sufficient"]], "response_mode": "offline", "answer": "[{\"word\":\"sufficient\",\"start\":0.26,\"end\":1.09}]"}
{"utterance_id": "251-118436-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0021.flac", "task_type": "lexical_detection", "query": "whisper", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"whisper\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.08, 3.64, "whisper"]], "response_mode": "offline", "answer": "[{\"word\":\"whisper\",\"start\":3.08,\"end\":3.64}]"}
{"utterance_id": "3081-166546-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.38, "side"]], "response_mode": "offline", "answer": "[{\"word\":\"side\",\"start\":2.87,\"end\":3.38}]"}
{"utterance_id": "652-130726-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0006.flac", "task_type": "lexical_detection", "query": "club", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"club\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.11, 4.45, "club"]], "response_mode": "offline", "answer": "[{\"word\":\"club\",\"start\":4.11,\"end\":4.45}]"}
{"utterance_id": "1462-170142-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0026.flac", "task_type": "lexical_detection", "query": "gratitude", "prompt": "When the utterance is complete, report all instances of the word \"gratitude\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.94, 2.59, "gratitude"]], "response_mode": "offline", "answer": "[{\"word\":\"gratitude\",\"start\":1.94,\"end\":2.59}]"}
{"utterance_id": "5338-24640-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0000.flac", "task_type": "lexical_detection", "query": "chapter", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"chapter\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.82, "chapter"]], "response_mode": "offline", "answer": "[{\"word\":\"chapter\",\"start\":0.38,\"end\":0.82}]"}
{"utterance_id": "2902-9006-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0015.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.65, 5.21, "capitol"], [28.33, 29.06, "gibraltar"]], "response_mode": "offline", "answer": "[{\"word\":\"capitol\",\"start\":4.65,\"end\":5.21},{\"word\":\"gibraltar\",\"start\":28.33,\"end\":29.06}]"}
{"utterance_id": "174-50561-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0015.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.58, 8.36, "apple bough"]], "response_mode": "offline", "answer": "[{\"word\":\"apple bough\",\"start\":7.58,\"end\":8.36}]"}
{"utterance_id": "3000-15664-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0043.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen carefully to the entire clip, then report every instance of a city name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.28, "shasta"]], "response_mode": "offline", "answer": "[{\"word\":\"shasta\",\"start\":1.55,\"end\":2.28}]"}
{"utterance_id": "6313-66129-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0022.flac", "task_type": "lexical_detection", "query": "won", "prompt": "After the audio ends, extract all occurrences of the word \"won\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.18, 3.37, "won"]], "response_mode": "offline", "answer": "[{\"word\":\"won\",\"start\":3.18,\"end\":3.37}]"}
{"utterance_id": "251-137823-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0020.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a person's name (word, start, end). If none, reply [].", "labels": [[2.46, 2.84, "one"], [4.28, 4.6, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.46,\"end\":2.84},{\"word\":\"two\",\"start\":4.28,\"end\":4.6}]"}
{"utterance_id": "3170-137482-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0035.flac", "task_type": "lexical_detection", "query": "subject", "prompt": "Process the full audio offline. Find every instance of the word \"subject\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.0, 5.49, "subject"]], "response_mode": "offline", "answer": "[{\"word\":\"subject\",\"start\":5.0,\"end\":5.49}]"}
{"utterance_id": "3752-4944-0066", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0066.flac", "task_type": "lexical_counting", "query": "twice", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"twice\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.26, 0.69, "one"], [3.35, 3.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.26,\"end\":0.69},{\"word\":\"two\",\"start\":3.35,\"end\":3.86}]"}
{"utterance_id": "3752-4944-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0049.flac", "task_type": "lexical_detection", "query": "burst", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"burst\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.52, 1.9, "burst"]], "response_mode": "offline", "answer": "[{\"word\":\"burst\",\"start\":1.52,\"end\":1.9}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.33, 1.23, "one"], [3.58, 4.35, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.33,\"end\":1.23},{\"word\":\"two\",\"start\":3.58,\"end\":4.35}]"}
{"utterance_id": "7976-110523-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.02, 7.32, "grethel"], [12.33, 12.88, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"grethel\",\"start\":7.02,\"end\":7.32},{\"word\":\"hansel\",\"start\":12.33,\"end\":12.88}]"}
{"utterance_id": "2086-149214-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149214/2086-149214-0004.flac", "task_type": "lexical_detection", "query": "laying", "prompt": "Once the audio has finished, identify every instance of the word \"laying\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.75, 4.05, "laying"]], "response_mode": "offline", "answer": "[{\"word\":\"laying\",\"start\":3.75,\"end\":4.05}]"}
{"utterance_id": "5338-284437-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0019.flac", "task_type": "lexical_detection", "query": "surely", "prompt": "When the utterance is complete, report all instances of the word \"surely\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.66, "surely"]], "response_mode": "offline", "answer": "[{\"word\":\"surely\",\"start\":0.3,\"end\":0.66}]"}
{"utterance_id": "84-121550-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0028.flac", "task_type": "lexical_detection", "query": "ages", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"ages\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.84, 6.24, "ages"]], "response_mode": "offline", "answer": "[{\"word\":\"ages\",\"start\":5.84,\"end\":6.24}]"}
{"utterance_id": "3576-138058-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0037.flac", "task_type": "semantic_counting", "query": "profession", "prompt": "Process the full audio offline. For every occurrence of a profession, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.13, "one"], [3.82, 4.3, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.6,\"end\":1.13},{\"word\":\"two\",\"start\":3.82,\"end\":4.3}]"}
{"utterance_id": "8297-275154-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0014.flac", "task_type": "lexical_counting", "query": "always", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"always\" with word/start/end. If none, reply [].", "labels": [[5.04, 5.8, "one"], [6.95, 7.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.04,\"end\":5.8},{\"word\":\"two\",\"start\":6.95,\"end\":7.36}]"}
{"utterance_id": "6319-64726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0018.flac", "task_type": "lexical_counting", "query": "great", "prompt": "Process the full audio offline. For every occurrence of the word \"great\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.88, 5.25, "one"], [9.77, 10.05, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.88,\"end\":5.25},{\"word\":\"two\",\"start\":9.77,\"end\":10.05}]"}
{"utterance_id": "2803-154328-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0021.flac", "task_type": "lexical_detection", "query": "savages", "prompt": "After listening to the whole recording, output every instance of the word \"savages\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.38, "savages"]], "response_mode": "offline", "answer": "[{\"word\":\"savages\",\"start\":0.71,\"end\":1.38}]"}
{"utterance_id": "1673-143396-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0006.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.01, 8.51, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":8.01,\"end\":8.51}]"}
{"utterance_id": "1462-170138-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After the audio ends, extract all occurrences of a family relation. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.62, 4.96, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":4.62,\"end\":4.96}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.21, 3.1, "abalone dishes"], [9.58, 10.03, "abalone"]], "response_mode": "offline", "answer": "[{\"word\":\"abalone dishes\",\"start\":2.21,\"end\":3.1},{\"word\":\"abalone\",\"start\":9.58,\"end\":10.03}]"}
{"utterance_id": "6241-61946-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0016.flac", "task_type": "lexical_detection", "query": "sacred", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"sacred\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.83, "sacred"]], "response_mode": "offline", "answer": "[{\"word\":\"sacred\",\"start\":0.49,\"end\":0.83}]"}
{"utterance_id": "6313-66129-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0020.flac", "task_type": "lexical_detection", "query": "surrounded", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"surrounded\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.55, 2.1, "surrounded"]], "response_mode": "offline", "answer": "[{\"word\":\"surrounded\",\"start\":1.55,\"end\":2.1}]"}
{"utterance_id": "2412-153954-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.35, 4.62, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":4.35,\"end\":4.62}]"}
{"utterance_id": "2277-149897-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0023.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Do not answer until the audio ends. Then list each instance of a date as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.97, 1.41, "sunday"]], "response_mode": "offline", "answer": "[{\"word\":\"sunday\",\"start\":0.97,\"end\":1.41}]"}
{"utterance_id": "1988-147956-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0017.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "After listening to the whole recording, output every instance of a state or province name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.25, 5.8, "virginia"]], "response_mode": "offline", "answer": "[{\"word\":\"virginia\",\"start\":5.25,\"end\":5.8}]"}
{"utterance_id": "6345-64257-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0008.flac", "task_type": "lexical_detection", "query": "tombs", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"tombs\" with fields word, start, end (seconds). If none, reply [].", "labels": [[2.38, 3.11, "tombs"]], "response_mode": "offline", "answer": "[{\"word\":\"tombs\",\"start\":2.38,\"end\":3.11}]"}
{"utterance_id": "7850-286674-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 1.01, "nymphs"]], "response_mode": "offline", "answer": "[{\"word\":\"nymphs\",\"start\":0.48,\"end\":1.01}]"}
{"utterance_id": "6295-64301-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0018.flac", "task_type": "lexical_detection", "query": "garment", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"garment\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[6.17, 6.8, "garment"]], "response_mode": "offline", "answer": "[{\"word\":\"garment\",\"start\":6.17,\"end\":6.8}]"}
{"utterance_id": "3536-8226-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0003.flac", "task_type": "lexical_detection", "query": "expected", "prompt": "After the audio ends, extract all occurrences of the word \"expected\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.91, "expected"]], "response_mode": "offline", "answer": "[{\"word\":\"expected\",\"start\":1.37,\"end\":1.91}]"}
{"utterance_id": "1988-24833-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0020.flac", "task_type": "lexical_detection", "query": "happening", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"happening\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.36, "happening"]], "response_mode": "offline", "answer": "[{\"word\":\"happening\",\"start\":2.9,\"end\":3.36}]"}
{"utterance_id": "6313-76958-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0013.flac", "task_type": "semantic_counting", "query": "clock_time", "prompt": "Process the full audio offline. For every occurrence of a time of day, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.9, 3.15, "one"], [8.45, 8.88, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.9,\"end\":3.15},{\"word\":\"two\",\"start\":8.45,\"end\":8.88}]"}
{"utterance_id": "422-122949-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0031.flac", "task_type": "lexical_detection", "query": "sort", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sort\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.06, "sort"]], "response_mode": "offline", "answer": "[{\"word\":\"sort\",\"start\":0.85,\"end\":1.06}]"}
{"utterance_id": "251-136532-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0020.flac", "task_type": "lexical_detection", "query": "attention", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"attention\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.15, 1.74, "attention"]], "response_mode": "offline", "answer": "[{\"word\":\"attention\",\"start\":1.15,\"end\":1.74}]"}
{"utterance_id": "5338-24615-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0012.flac", "task_type": "lexical_detection", "query": "wonder", "prompt": "Process the full audio offline. Find every instance of the word \"wonder\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.42, "wonder"]], "response_mode": "offline", "answer": "[{\"word\":\"wonder\",\"start\":1.73,\"end\":2.42}]"}
{"utterance_id": "3081-166546-0082", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0082.flac", "task_type": "lexical_detection", "query": "narrowed", "prompt": "When the utterance is complete, report all instances of the word \"narrowed\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.41, "narrowed"]], "response_mode": "offline", "answer": "[{\"word\":\"narrowed\",\"start\":0.97,\"end\":1.41}]"}
{"utterance_id": "1462-170138-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0006.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.96, 3.28, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":2.96,\"end\":3.28}]"}
{"utterance_id": "1993-147149-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0016.flac", "task_type": "lexical_detection", "query": "frame", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"frame\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.11, 3.46, "frame"]], "response_mode": "offline", "answer": "[{\"word\":\"frame\",\"start\":3.11,\"end\":3.46}]"}
{"utterance_id": "3853-163249-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0014.flac", "task_type": "lexical_detection", "query": "grim", "prompt": "After listening to the whole recording, output every instance of the word \"grim\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.61, 3.89, "grim"]], "response_mode": "offline", "answer": "[{\"word\":\"grim\",\"start\":3.61,\"end\":3.89}]"}
{"utterance_id": "7976-110523-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.21, 1.62, "hansel"]], "response_mode": "offline", "answer": "[{\"word\":\"hansel\",\"start\":1.21,\"end\":1.62}]"}
{"utterance_id": "84-121550-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0029.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.37, 7.6, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.37,\"end\":7.6}]"}
{"utterance_id": "1919-142785-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0000.flac", "task_type": "lexical_detection", "query": "pepper", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"pepper\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.84, 2.23, "pepper"]], "response_mode": "offline", "answer": "[{\"word\":\"pepper\",\"start\":1.84,\"end\":2.23}]"}
{"utterance_id": "1272-135031-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0020.flac", "task_type": "lexical_detection", "query": "magic", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"magic\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.1, 2.73, "magic"]], "response_mode": "offline", "answer": "[{\"word\":\"magic\",\"start\":2.1,\"end\":2.73}]"}
{"utterance_id": "6241-66616-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0010.flac", "task_type": "lexical_detection", "query": "bride", "prompt": "Process the full audio offline. Find every instance of the word \"bride\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.69, 4.12, "bride"]], "response_mode": "offline", "answer": "[{\"word\":\"bride\",\"start\":3.69,\"end\":4.12}]"}
{"utterance_id": "174-168635-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.5, 1.11, "cosette"]], "response_mode": "offline", "answer": "[{\"word\":\"cosette\",\"start\":0.5,\"end\":1.11}]"}
{"utterance_id": "3000-15664-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0027.flac", "task_type": "lexical_detection", "query": "striking", "prompt": "Once the audio has finished, identify every instance of the word \"striking\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.65, 3.08, "striking"]], "response_mode": "offline", "answer": "[{\"word\":\"striking\",\"start\":2.65,\"end\":3.08}]"}
{"utterance_id": "7976-110523-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0004.flac", "task_type": "lexical_detection", "query": "coat", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"coat\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.18, 4.63, "coat"]], "response_mode": "offline", "answer": "[{\"word\":\"coat\",\"start\":4.18,\"end\":4.63}]"}
{"utterance_id": "2412-153954-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[3.14, 3.94, "magistrate"]], "response_mode": "offline", "answer": "[{\"word\":\"magistrate\",\"start\":3.14,\"end\":3.94}]"}
{"utterance_id": "3081-166546-0037", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0037.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, produce a running count of a body part: each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[1.55, 2.09, "one"], [4.32, 4.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.55,\"end\":2.09},{\"word\":\"two\",\"start\":4.32,\"end\":4.74}]"}
{"utterance_id": "174-50561-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0010.flac", "task_type": "lexical_detection", "query": "granted", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"granted\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.09, 6.46, "granted"]], "response_mode": "offline", "answer": "[{\"word\":\"granted\",\"start\":6.09,\"end\":6.46}]"}
{"utterance_id": "2086-149220-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0048.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "Listen to the entire clip, then report the count sequence for a body part as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.0, 6.43, "one"], [7.98, 8.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.0,\"end\":6.43},{\"word\":\"two\",\"start\":7.98,\"end\":8.34}]"}
{"utterance_id": "2902-9006-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0006.flac", "task_type": "lexical_detection", "query": "empire", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"empire\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.6, 3.12, "empire"]], "response_mode": "offline", "answer": "[{\"word\":\"empire\",\"start\":2.6,\"end\":3.12}]"}
{"utterance_id": "1673-143396-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0016.flac", "task_type": "lexical_counting", "query": "worthy", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"worthy\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.51, 0.85, "one"], [2.42, 2.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":0.85},{\"word\":\"two\",\"start\":2.42,\"end\":2.73}]"}
{"utterance_id": "6319-64726-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0008.flac", "task_type": "lexical_detection", "query": "prince", "prompt": "Process the full audio offline. Find every instance of the word \"prince\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.34, "prince"]], "response_mode": "offline", "answer": "[{\"word\":\"prince\",\"start\":0.93,\"end\":1.34}]"}
{"utterance_id": "777-126732-0080", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0080.flac", "task_type": "lexical_detection", "query": "notion", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"notion\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.01, 1.46, "notion"]], "response_mode": "offline", "answer": "[{\"word\":\"notion\",\"start\":1.01,\"end\":1.46}]"}
{"utterance_id": "422-122949-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0013.flac", "task_type": "lexical_detection", "query": "entangled", "prompt": "Process the full audio offline. Find every instance of the word \"entangled\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.38, 5.95, "entangled"]], "response_mode": "offline", "answer": "[{\"word\":\"entangled\",\"start\":5.38,\"end\":5.95}]"}
{"utterance_id": "1462-170138-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.59, "mac connell's"]], "response_mode": "offline", "answer": "[{\"word\":\"mac connell's\",\"start\":1.15,\"end\":1.59}]"}
{"utterance_id": "2078-142845-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0004.flac", "task_type": "lexical_detection", "query": "spoon", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"spoon\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.06, 8.64, "spoon"]], "response_mode": "offline", "answer": "[{\"word\":\"spoon\",\"start\":8.06,\"end\":8.64}]"}
{"utterance_id": "6295-64301-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.75, "letty"], [5.09, 5.59, "mary"]], "response_mode": "offline", "answer": "[{\"word\":\"letty\",\"start\":0.33,\"end\":0.75},{\"word\":\"mary\",\"start\":5.09,\"end\":5.59}]"}
{"utterance_id": "652-129742-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0009.flac", "task_type": "lexical_detection", "query": "shredded", "prompt": "After listening to the whole recording, output every instance of the word \"shredded\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.26, 7.66, "shredded"]], "response_mode": "offline", "answer": "[{\"word\":\"shredded\",\"start\":7.26,\"end\":7.66}]"}
{"utterance_id": "2902-9006-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.92, 12.78, "belisarius"]], "response_mode": "offline", "answer": "[{\"word\":\"belisarius\",\"start\":11.92,\"end\":12.78}]"}
{"utterance_id": "8842-302196-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0000.flac", "task_type": "lexical_detection", "query": "poems", "prompt": "When the utterance is complete, report all instances of the word \"poems\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.83, 6.31, "poems"]], "response_mode": "offline", "answer": "[{\"word\":\"poems\",\"start\":5.83,\"end\":6.31}]"}
{"utterance_id": "652-129742-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0002.flac", "task_type": "lexical_detection", "query": "hidden", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"hidden\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.63, 1.98, "hidden"]], "response_mode": "offline", "answer": "[{\"word\":\"hidden\",\"start\":1.63,\"end\":1.98}]"}
{"utterance_id": "2428-83699-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0003.flac", "task_type": "semantic_detection", "query": "money", "prompt": "When the utterance is complete, report all instances of a money or currency term including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.54, "guineas"], [5.09, 5.34, "fee"]], "response_mode": "offline", "answer": "[{\"word\":\"guineas\",\"start\":2.03,\"end\":2.54},{\"word\":\"fee\",\"start\":5.09,\"end\":5.34}]"}
{"utterance_id": "2428-83705-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0003.flac", "task_type": "lexical_detection", "query": "delightful", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"delightful\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.76, "delightful"]], "response_mode": "offline", "answer": "[{\"word\":\"delightful\",\"start\":1.1,\"end\":1.76}]"}
{"utterance_id": "6345-93302-0027", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0027.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.93, 3.36, "tea"]], "response_mode": "offline", "answer": "[{\"word\":\"tea\",\"start\":2.93,\"end\":3.36}]"}
{"utterance_id": "2428-83699-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0032.flac", "task_type": "lexical_detection", "query": "maybe", "prompt": "After the audio ends, extract all occurrences of the word \"maybe\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.2, 0.49, "maybe"]], "response_mode": "offline", "answer": "[{\"word\":\"maybe\",\"start\":0.2,\"end\":0.49}]"}
{"utterance_id": "3576-138058-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0033.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[16.51, 17.31, "barcelona"]], "response_mode": "offline", "answer": "[{\"word\":\"barcelona\",\"start\":16.51,\"end\":17.31}]"}
{"utterance_id": "1462-170138-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0022.flac", "task_type": "lexical_detection", "query": "stage", "prompt": "When the utterance is complete, report all instances of the word \"stage\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.99, 2.33, "stage"]], "response_mode": "offline", "answer": "[{\"word\":\"stage\",\"start\":1.99,\"end\":2.33}]"}
{"utterance_id": "6295-244435-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0005.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.49, 0.73, "white"]], "response_mode": "offline", "answer": "[{\"word\":\"white\",\"start\":0.49,\"end\":0.73}]"}
{"utterance_id": "6241-61943-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0003.flac", "task_type": "lexical_detection", "query": "ghost", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"ghost\" with fields word, start, end (seconds). If none, reply [].", "labels": [[6.49, 6.79, "ghost"]], "response_mode": "offline", "answer": "[{\"word\":\"ghost\",\"start\":6.49,\"end\":6.79}]"}
{"utterance_id": "3170-137482-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0048.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.61, 2.08, "feet"]], "response_mode": "offline", "answer": "[{\"word\":\"feet\",\"start\":1.61,\"end\":2.08}]"}
{"utterance_id": "1993-147149-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0028.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After listening to the whole recording, output every instance of a body part with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.77, 5.21, "bosom"]], "response_mode": "offline", "answer": "[{\"word\":\"bosom\",\"start\":4.77,\"end\":5.21}]"}
{"utterance_id": "3752-4944-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.23, 2.12, "captain burgess"]], "response_mode": "offline", "answer": "[{\"word\":\"captain burgess\",\"start\":1.23,\"end\":2.12}]"}
{"utterance_id": "1272-141231-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.62, 4.01, "brion"]], "response_mode": "offline", "answer": "[{\"word\":\"brion\",\"start\":3.62,\"end\":4.01}]"}
{"utterance_id": "3081-166546-0056", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0056.flac", "task_type": "lexical_detection", "query": "lobby", "prompt": "After listening to the whole recording, output every instance of the word \"lobby\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.04, 4.53, "lobby"]], "response_mode": "offline", "answer": "[{\"word\":\"lobby\",\"start\":4.04,\"end\":4.53}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of count events for a number (word, start, end). If none, reply [].", "labels": [[3.82, 4.46, "one"], [12.85, 13.32, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.82,\"end\":4.46},{\"word\":\"two\",\"start\":12.85,\"end\":13.32}]"}
{"utterance_id": "7976-110523-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0011.flac", "task_type": "lexical_counting", "query": "come", "prompt": "Listen to the full audio. Count every occurrence of the word \"come\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.16, 0.37, "one"], [2.68, 2.96, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.16,\"end\":0.37},{\"word\":\"two\",\"start\":2.68,\"end\":2.96}]"}
{"utterance_id": "8842-302203-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0009.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Listen to the full audio. Count every occurrence of a person's name. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.8, "one"], [7.72, 9.48, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.37,\"end\":1.8},{\"word\":\"two\",\"start\":7.72,\"end\":9.48}]"}
{"utterance_id": "174-168635-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0010.flac", "task_type": "lexical_detection", "query": "perceived", "prompt": "When the utterance is complete, report all instances of the word \"perceived\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.55, "perceived"]], "response_mode": "offline", "answer": "[{\"word\":\"perceived\",\"start\":1.92,\"end\":2.55}]"}
{"utterance_id": "1993-147149-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0030.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Listen to the full audio. Count every occurrence of a family relation. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.74, 9.2, "one"], [10.43, 10.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.74,\"end\":9.2},{\"word\":\"two\",\"start\":10.43,\"end\":10.9}]"}
{"utterance_id": "1988-147956-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0004.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.03, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.89,\"end\":2.03}]"}
{"utterance_id": "2078-142845-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0051.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.28, 6.75, "cheese"], [8.33, 8.8, "sugar"]], "response_mode": "offline", "answer": "[{\"word\":\"cheese\",\"start\":6.28,\"end\":6.75},{\"word\":\"sugar\",\"start\":8.33,\"end\":8.8}]"}
{"utterance_id": "422-122949-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0000.flac", "task_type": "lexical_counting", "query": "society", "prompt": "Process the full audio offline. For every occurrence of the word \"society\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.28, 7.18, "one"], [9.41, 10.08, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.28,\"end\":7.18},{\"word\":\"two\",\"start\":9.41,\"end\":10.08}]"}
{"utterance_id": "2412-153954-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0010.flac", "task_type": "lexical_detection", "query": "exceedingly", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"exceedingly\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.29, 2.86, "exceedingly"]], "response_mode": "offline", "answer": "[{\"word\":\"exceedingly\",\"start\":2.29,\"end\":2.86}]"}
{"utterance_id": "251-137823-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.83, "tom's"]], "response_mode": "offline", "answer": "[{\"word\":\"tom's\",\"start\":0.3,\"end\":0.83}]"}
{"utterance_id": "1673-143396-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0020.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.98, 14.51, "romans"]], "response_mode": "offline", "answer": "[{\"word\":\"romans\",\"start\":13.98,\"end\":14.51}]"}
{"utterance_id": "5895-34622-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0018.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.3, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.03,\"end\":2.3}]"}
{"utterance_id": "2078-142845-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0017.flac", "task_type": "lexical_counting", "query": "maize", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"maize\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.5, 1.25, "one"], [2.88, 3.31, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.5,\"end\":1.25},{\"word\":\"two\",\"start\":2.88,\"end\":3.31}]"}
{"utterance_id": "7976-110523-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0019.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 1.74, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":1.52,\"end\":1.74}]"}
{"utterance_id": "5895-34629-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0000.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.43, 2.06, "southwark"]], "response_mode": "offline", "answer": "[{\"word\":\"southwark\",\"start\":1.43,\"end\":2.06}]"}
{"utterance_id": "7976-110124-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.61, 1.89, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":1.61,\"end\":1.89}]"}
{"utterance_id": "3752-4944-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0039.flac", "task_type": "lexical_detection", "query": "cancer", "prompt": "Process the full audio offline. Find every instance of the word \"cancer\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.86, 1.27, "cancer"], [4.37, 4.81, "cancer"]], "response_mode": "offline", "answer": "[{\"word\":\"cancer\",\"start\":0.86,\"end\":1.27},{\"word\":\"cancer\",\"start\":4.37,\"end\":4.81}]"}
{"utterance_id": "2086-149220-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0041.flac", "task_type": "lexical_detection", "query": "chairs", "prompt": "After listening to the whole recording, output every instance of the word \"chairs\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.35, 6.79, "chairs"]], "response_mode": "offline", "answer": "[{\"word\":\"chairs\",\"start\":6.35,\"end\":6.79}]"}
{"utterance_id": "3170-137482-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0047.flac", "task_type": "lexical_detection", "query": "future", "prompt": "When the utterance is complete, report all instances of the word \"future\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.31, 1.9, "future"]], "response_mode": "offline", "answer": "[{\"word\":\"future\",\"start\":1.31,\"end\":1.9}]"}
{"utterance_id": "1462-170142-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0001.flac", "task_type": "semantic_detection", "query": "city", "prompt": "You will hear a complete utterance. When it ends, list every instance of a city name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.12, 1.73, "euston"], [8.13, 9.1, "bedford square"]], "response_mode": "offline", "answer": "[{\"word\":\"euston\",\"start\":1.12,\"end\":1.73},{\"word\":\"bedford square\",\"start\":8.13,\"end\":9.1}]"}
{"utterance_id": "7850-286674-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0006.flac", "task_type": "lexical_detection", "query": "lips", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lips\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.93, 1.48, "lips"]], "response_mode": "offline", "answer": "[{\"word\":\"lips\",\"start\":0.93,\"end\":1.48}]"}
{"utterance_id": "3752-4944-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0012.flac", "task_type": "lexical_detection", "query": "seeing", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"seeing\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.73, 3.06, "seeing"], [4.8, 5.11, "seeing"]], "response_mode": "offline", "answer": "[{\"word\":\"seeing\",\"start\":2.73,\"end\":3.06},{\"word\":\"seeing\",\"start\":4.8,\"end\":5.11}]"}
{"utterance_id": "8842-302196-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302196/8842-302196-0008.flac", "task_type": "lexical_detection", "query": "useless", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"useless\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.4, 4.97, "useless"]], "response_mode": "offline", "answer": "[{\"word\":\"useless\",\"start\":4.4,\"end\":4.97}]"}
{"utterance_id": "2412-153947-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.05, 7.59, "faces"]], "response_mode": "offline", "answer": "[{\"word\":\"faces\",\"start\":7.05,\"end\":7.59}]"}
{"utterance_id": "3752-4944-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0029.flac", "task_type": "lexical_detection", "query": "flog", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"flog\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.22, 1.56, "flog"]], "response_mode": "offline", "answer": "[{\"word\":\"flog\",\"start\":1.22,\"end\":1.56}]"}
{"utterance_id": "3853-163249-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0032.flac", "task_type": "lexical_detection", "query": "sufferin", "prompt": "When the utterance is complete, report all instances of the word \"sufferin\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.7, 4.22, "sufferin"]], "response_mode": "offline", "answer": "[{\"word\":\"sufferin\",\"start\":3.7,\"end\":4.22}]"}
{"utterance_id": "5694-64029-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64029/5694-64029-0023.flac", "task_type": "lexical_detection", "query": "yards", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"yards\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.15, 2.56, "yards"]], "response_mode": "offline", "answer": "[{\"word\":\"yards\",\"start\":2.15,\"end\":2.56}]"}
{"utterance_id": "5694-64025-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0014.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.21, 3.25, "private soldier"], [7.06, 7.75, "generals"]], "response_mode": "offline", "answer": "[{\"word\":\"private soldier\",\"start\":2.21,\"end\":3.25},{\"word\":\"generals\",\"start\":7.06,\"end\":7.75}]"}
{"utterance_id": "1993-147149-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0030.flac", "task_type": "lexical_detection", "query": "relating", "prompt": "Once the audio has finished, identify every instance of the word \"relating\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.0, 8.47, "relating"]], "response_mode": "offline", "answer": "[{\"word\":\"relating\",\"start\":8.0,\"end\":8.47}]"}
{"utterance_id": "8842-302203-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0006.flac", "task_type": "lexical_detection", "query": "uncertain", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"uncertain\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.2, 3.71, "uncertain"]], "response_mode": "offline", "answer": "[{\"word\":\"uncertain\",\"start\":3.2,\"end\":3.71}]"}
{"utterance_id": "777-126732-0049", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0049.flac", "task_type": "lexical_detection", "query": "survey", "prompt": "Process the full audio offline. Find every instance of the word \"survey\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.58, 0.87, "survey"]], "response_mode": "offline", "answer": "[{\"word\":\"survey\",\"start\":0.58,\"end\":0.87}]"}
{"utterance_id": "84-121123-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0025.flac", "task_type": "lexical_detection", "query": "italian", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"italian\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 2.78, "italian"]], "response_mode": "offline", "answer": "[{\"word\":\"italian\",\"start\":2.29,\"end\":2.78}]"}
{"utterance_id": "5694-64038-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0023.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.37, 4.11, "murfreesboro"]], "response_mode": "offline", "answer": "[{\"word\":\"murfreesboro\",\"start\":3.37,\"end\":4.11}]"}
{"utterance_id": "3000-15664-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0011.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Process the full audio offline. Find every instance of a weather term and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.83, 1.59, "rainstorms"], [9.25, 9.6, "rain"]], "response_mode": "offline", "answer": "[{\"word\":\"rainstorms\",\"start\":0.83,\"end\":1.59},{\"word\":\"rain\",\"start\":9.25,\"end\":9.6}]"}
{"utterance_id": "1919-142785-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0018.flac", "task_type": "semantic_counting", "query": "plant", "prompt": "Listen to the entire clip, then report the count sequence for a plant name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.39, 1.91, "one"], [3.12, 3.51, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.39,\"end\":1.91},{\"word\":\"two\",\"start\":3.12,\"end\":3.51}]"}
{"utterance_id": "2803-154328-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0004.flac", "task_type": "lexical_detection", "query": "terror", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"terror\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.18, 8.51, "terror"]], "response_mode": "offline", "answer": "[{\"word\":\"terror\",\"start\":8.18,\"end\":8.51}]"}
{"utterance_id": "2803-154328-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0018.flac", "task_type": "lexical_detection", "query": "bled", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"bled\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.94, 1.74, "bled"]], "response_mode": "offline", "answer": "[{\"word\":\"bled\",\"start\":0.94,\"end\":1.74}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[11.5, 11.94, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":11.5,\"end\":11.94}]"}
{"utterance_id": "7976-110124-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0021.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "After listening to the whole recording, output every instance of a profession with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.62, 3.11, "robber"]], "response_mode": "offline", "answer": "[{\"word\":\"robber\",\"start\":2.62,\"end\":3.11}]"}
{"utterance_id": "652-130726-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0033.flac", "task_type": "lexical_counting", "query": "abalone", "prompt": "After listening to the whole recording, output timed count words for the word \"abalone\" (one, two, three, ...) as JSON. If none, reply [].", "labels": [[2.21, 2.63, "one"], [9.58, 10.03, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.21,\"end\":2.63},{\"word\":\"two\",\"start\":9.58,\"end\":10.03}]"}
{"utterance_id": "2412-153948-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153948/2412-153948-0009.flac", "task_type": "lexical_detection", "query": "anything", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"anything\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.34, 4.74, "anything"]], "response_mode": "offline", "answer": "[{\"word\":\"anything\",\"start\":4.34,\"end\":4.74}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[2.26, 2.85, "apostle"]], "response_mode": "offline", "answer": "[{\"word\":\"apostle\",\"start\":2.26,\"end\":2.85}]"}
{"utterance_id": "5338-284437-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0031.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen carefully to the entire clip, then report every instance of an ordinal number as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.97, 2.51, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":1.97,\"end\":2.51}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.59, 0.83, "second"], [2.68, 3.07, "one"], [6.19, 6.57, "second"]], "response_mode": "offline", "answer": "[{\"word\":\"second\",\"start\":0.59,\"end\":0.83},{\"word\":\"one\",\"start\":2.68,\"end\":3.07},{\"word\":\"second\",\"start\":6.19,\"end\":6.57}]"}
{"utterance_id": "2412-153954-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0000.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Do not answer until the audio ends. Then list each instance of a weather term as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.08, 5.44, "mist"]], "response_mode": "offline", "answer": "[{\"word\":\"mist\",\"start\":5.08,\"end\":5.44}]"}
{"utterance_id": "652-130726-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0025.flac", "task_type": "lexical_detection", "query": "golden", "prompt": "Process the full audio offline. Find every instance of the word \"golden\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.7, 2.08, "golden"]], "response_mode": "offline", "answer": "[{\"word\":\"golden\",\"start\":1.7,\"end\":2.08}]"}
{"utterance_id": "6295-64301-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0001.flac", "task_type": "lexical_detection", "query": "holiday", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"holiday\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.36, "holiday"]], "response_mode": "offline", "answer": "[{\"word\":\"holiday\",\"start\":1.6,\"end\":2.36}]"}
{"utterance_id": "2035-152373-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.37, 8.84, "moira"]], "response_mode": "offline", "answer": "[{\"word\":\"moira\",\"start\":8.37,\"end\":8.84}]"}
{"utterance_id": "2035-147961-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0024.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.2, 1.56, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":1.2,\"end\":1.56}]"}
{"utterance_id": "5895-34615-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0013.flac", "task_type": "lexical_detection", "query": "modified", "prompt": "When the utterance is complete, report all instances of the word \"modified\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.48, "modified"]], "response_mode": "offline", "answer": "[{\"word\":\"modified\",\"start\":2.87,\"end\":3.48}]"}
{"utterance_id": "6313-66129-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0024.flac", "task_type": "lexical_detection", "query": "rods", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"rods\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.47, 4.96, "rods"]], "response_mode": "offline", "answer": "[{\"word\":\"rods\",\"start\":4.47,\"end\":4.96}]"}
{"utterance_id": "5536-43358-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0015.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After listening to the whole recording, output every instance of an animal name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.15, 3.45, "beast"]], "response_mode": "offline", "answer": "[{\"word\":\"beast\",\"start\":3.15,\"end\":3.45}]"}
{"utterance_id": "1988-147956-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Process the full audio offline. Find every instance of an animal name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.98, 2.56, "horses"]], "response_mode": "offline", "answer": "[{\"word\":\"horses\",\"start\":1.98,\"end\":2.56}]"}
{"utterance_id": "2902-9008-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0012.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.5, "gods"]], "response_mode": "offline", "answer": "[{\"word\":\"gods\",\"start\":5.9,\"end\":6.5}]"}
{"utterance_id": "5536-43358-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0018.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.6, 9.04, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":8.6,\"end\":9.04}]"}
{"utterance_id": "652-129742-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0006.flac", "task_type": "lexical_counting", "query": "cauliflower", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"cauliflower\" with word/start/end. If none, reply [].", "labels": [[0.51, 1.2, "one"], [4.0, 4.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.2},{\"word\":\"two\",\"start\":4.0,\"end\":4.73}]"}
{"utterance_id": "1673-143397-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0011.flac", "task_type": "lexical_detection", "query": "trial", "prompt": "After listening to the whole recording, output every instance of the word \"trial\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.04, 11.52, "trial"]], "response_mode": "offline", "answer": "[{\"word\":\"trial\",\"start\":11.04,\"end\":11.52}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "semantic_detection", "query": "color", "prompt": "When the utterance is complete, report all instances of a color name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.52, 5.95, "pasty"]], "response_mode": "offline", "answer": "[{\"word\":\"pasty\",\"start\":5.52,\"end\":5.95}]"}
{"utterance_id": "6295-244435-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.1, 0.36, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"four\",\"start\":0.1,\"end\":0.36}]"}
{"utterance_id": "1919-142785-0057", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0057.flac", "task_type": "lexical_detection", "query": "degree", "prompt": "After listening to the whole recording, output every instance of the word \"degree\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.92, 7.33, "degree"]], "response_mode": "offline", "answer": "[{\"word\":\"degree\",\"start\":6.92,\"end\":7.33}]"}
{"utterance_id": "3752-4943-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0004.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.31, 2.99, "mister north"]], "response_mode": "offline", "answer": "[{\"word\":\"mister north\",\"start\":2.31,\"end\":2.99}]"}
{"utterance_id": "2035-147961-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0031.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.38, 0.84, "peter"]], "response_mode": "offline", "answer": "[{\"word\":\"peter\",\"start\":0.38,\"end\":0.84}]"}
{"utterance_id": "2078-142845-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0025.flac", "task_type": "lexical_detection", "query": "press", "prompt": "After the audio ends, extract all occurrences of the word \"press\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.07, 9.45, "press"]], "response_mode": "offline", "answer": "[{\"word\":\"press\",\"start\":9.07,\"end\":9.45}]"}
{"utterance_id": "5338-284437-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0025.flac", "task_type": "lexical_detection", "query": "great", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"great\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.96, 1.26, "great"]], "response_mode": "offline", "answer": "[{\"word\":\"great\",\"start\":0.96,\"end\":1.26}]"}
{"utterance_id": "3081-166546-0069", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0069.flac", "task_type": "lexical_detection", "query": "good", "prompt": "Once the audio has finished, identify every instance of the word \"good\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.05, 1.71, "good"]], "response_mode": "offline", "answer": "[{\"word\":\"good\",\"start\":1.05,\"end\":1.71}]"}
{"utterance_id": "5895-34615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0002.flac", "task_type": "lexical_detection", "query": "gwynplaine", "prompt": "Process the full audio offline. Find every instance of the word \"gwynplaine\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.72, 1.4, "gwynplaine"]], "response_mode": "offline", "answer": "[{\"word\":\"gwynplaine\",\"start\":0.72,\"end\":1.4}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Listen to the full audio. After it finishes, find every instance of a color name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.88, 2.44, "brown"], [4.98, 5.35, "green"]], "response_mode": "offline", "answer": "[{\"word\":\"brown\",\"start\":1.88,\"end\":2.44},{\"word\":\"green\",\"start\":4.98,\"end\":5.35}]"}
{"utterance_id": "1988-147956-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0025.flac", "task_type": "lexical_detection", "query": "eager", "prompt": "When the utterance is complete, report all instances of the word \"eager\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 2.12, "eager"]], "response_mode": "offline", "answer": "[{\"word\":\"eager\",\"start\":1.65,\"end\":2.12}]"}
{"utterance_id": "84-121550-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0004.flac", "task_type": "lexical_detection", "query": "human", "prompt": "When the utterance is complete, report all instances of the word \"human\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.26, 6.63, "human"]], "response_mode": "offline", "answer": "[{\"word\":\"human\",\"start\":6.26,\"end\":6.63}]"}
{"utterance_id": "777-126732-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0045.flac", "task_type": "lexical_detection", "query": "intolerable", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"intolerable\" with fields word, start, end (seconds). If none, reply [].", "labels": [[9.33, 10.05, "intolerable"]], "response_mode": "offline", "answer": "[{\"word\":\"intolerable\",\"start\":9.33,\"end\":10.05}]"}
{"utterance_id": "7850-286674-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0010.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.75, 2.39, "dragon fly"]], "response_mode": "offline", "answer": "[{\"word\":\"dragon fly\",\"start\":1.75,\"end\":2.39}]"}
{"utterance_id": "5895-34629-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0018.flac", "task_type": "lexical_detection", "query": "france", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"france\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.99, 5.58, "france"]], "response_mode": "offline", "answer": "[{\"word\":\"france\",\"start\":4.99,\"end\":5.58}]"}
{"utterance_id": "5338-284437-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0016.flac", "task_type": "lexical_detection", "query": "nothing", "prompt": "After listening to the whole recording, output every instance of the word \"nothing\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 0.99, "nothing"]], "response_mode": "offline", "answer": "[{\"word\":\"nothing\",\"start\":0.66,\"end\":0.99}]"}
{"utterance_id": "1673-143397-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0016.flac", "task_type": "lexical_detection", "query": "superstition", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"superstition\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.8, 7.58, "superstition"]], "response_mode": "offline", "answer": "[{\"word\":\"superstition\",\"start\":6.8,\"end\":7.58}]"}
{"utterance_id": "2078-142845-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0040.flac", "task_type": "lexical_detection", "query": "twelve", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"twelve\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.05, "twelve"]], "response_mode": "offline", "answer": "[{\"word\":\"twelve\",\"start\":1.6,\"end\":2.05}]"}
{"utterance_id": "84-121550-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0025.flac", "task_type": "lexical_detection", "query": "dante", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"dante\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 1.12, "dante"]], "response_mode": "offline", "answer": "[{\"word\":\"dante\",\"start\":0.5,\"end\":1.12}]"}
{"utterance_id": "6313-66129-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0026.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "After the audio ends, tally every instance of a body part. Reply with JSON only — one object per occurrence, word=\"one\"/\"two\"/... plus start and end in seconds. If none, reply [].", "labels": [[1.21, 1.58, "one"], [3.73, 4.06, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.21,\"end\":1.58},{\"word\":\"two\",\"start\":3.73,\"end\":4.06}]"}
{"utterance_id": "3576-138058-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0017.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.17, 5.67, "father"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":5.17,\"end\":5.67}]"}
{"utterance_id": "2078-142845-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0030.flac", "task_type": "semantic_detection", "query": "food", "prompt": "When the utterance is complete, report all instances of a food name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.54, 1.92, "bread"], [4.44, 5.07, "toast"], [6.43, 6.89, "bread"]], "response_mode": "offline", "answer": "[{\"word\":\"bread\",\"start\":1.54,\"end\":1.92},{\"word\":\"toast\",\"start\":4.44,\"end\":5.07},{\"word\":\"bread\",\"start\":6.43,\"end\":6.89}]"}
{"utterance_id": "652-130737-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0007.flac", "task_type": "lexical_detection", "query": "pleasing", "prompt": "When the utterance is complete, report all instances of the word \"pleasing\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.14, 2.61, "pleasing"]], "response_mode": "offline", "answer": "[{\"word\":\"pleasing\",\"start\":2.14,\"end\":2.61}]"}
{"utterance_id": "6345-93306-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0006.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen to the full audio. After it finishes, find every instance of a plant name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.32, 4.22, "laurel bushes"]], "response_mode": "offline", "answer": "[{\"word\":\"laurel bushes\",\"start\":3.32,\"end\":4.22}]"}
{"utterance_id": "3536-8226-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0026.flac", "task_type": "lexical_detection", "query": "gait", "prompt": "After the audio ends, extract all occurrences of the word \"gait\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.27, 5.6, "gait"]], "response_mode": "offline", "answer": "[{\"word\":\"gait\",\"start\":5.27,\"end\":5.6}]"}
{"utterance_id": "1673-143396-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0009.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of an ordinal number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.45, 0.78, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":0.45,\"end\":0.78}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "lexical_detection", "query": "mischief", "prompt": "Process the full audio offline. Find every instance of the word \"mischief\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.05, 5.56, "mischief"]], "response_mode": "offline", "answer": "[{\"word\":\"mischief\",\"start\":5.05,\"end\":5.56}]"}
{"utterance_id": "3170-137482-0047", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0047.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Once the audio has finished, identify every instance of a profession. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.21, 6.96, "adviser"]], "response_mode": "offline", "answer": "[{\"word\":\"adviser\",\"start\":6.21,\"end\":6.96}]"}
{"utterance_id": "174-168635-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0008.flac", "task_type": "lexical_counting", "query": "gulf", "prompt": "Process the full audio offline. For every occurrence of the word \"gulf\", emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.21, 4.67, "one"], [9.32, 9.73, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.21,\"end\":4.67},{\"word\":\"two\",\"start\":9.32,\"end\":9.73}]"}
{"utterance_id": "5338-24615-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "After the audio ends, extract all occurrences of a body part. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[26.72, 27.1, "lungs"]], "response_mode": "offline", "answer": "[{\"word\":\"lungs\",\"start\":26.72,\"end\":27.1}]"}
{"utterance_id": "8297-275155-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0009.flac", "task_type": "lexical_detection", "query": "sydenham", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"sydenham\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.0, 4.59, "sydenham"]], "response_mode": "offline", "answer": "[{\"word\":\"sydenham\",\"start\":4.0,\"end\":4.59}]"}
{"utterance_id": "8297-275155-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0019.flac", "task_type": "lexical_counting", "query": "stories", "prompt": "Once the audio has finished, list each occurrence of the word \"stories\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.78, 1.22, "one"], [4.2, 4.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.78,\"end\":1.22},{\"word\":\"two\",\"start\":4.2,\"end\":4.74}]"}
{"utterance_id": "1272-141231-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.44, 4.89, "chest"]], "response_mode": "offline", "answer": "[{\"word\":\"chest\",\"start\":4.44,\"end\":4.89}]"}
{"utterance_id": "7976-105575-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0000.flac", "task_type": "lexical_detection", "query": "fought", "prompt": "Process the full audio offline. Find every instance of the word \"fought\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.01, 6.27, "fought"]], "response_mode": "offline", "answer": "[{\"word\":\"fought\",\"start\":6.01,\"end\":6.27}]"}
{"utterance_id": "251-136532-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0002.flac", "task_type": "lexical_detection", "query": "archaeology", "prompt": "After the audio ends, extract all occurrences of the word \"archaeology\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.55, 3.36, "archaeology"]], "response_mode": "offline", "answer": "[{\"word\":\"archaeology\",\"start\":2.55,\"end\":3.36}]"}
{"utterance_id": "2035-147960-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0005.flac", "task_type": "lexical_detection", "query": "snake", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"snake\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.45, 4.85, "snake"]], "response_mode": "offline", "answer": "[{\"word\":\"snake\",\"start\":4.45,\"end\":4.85}]"}
{"utterance_id": "5694-64025-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0021.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.39, 3.78, "mule's"]], "response_mode": "offline", "answer": "[{\"word\":\"mule's\",\"start\":3.39,\"end\":3.78}]"}
{"utterance_id": "5895-34615-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0012.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the full audio. After it finishes, find every instance of an organization name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.48, 1.33, "manichaeans"]], "response_mode": "offline", "answer": "[{\"word\":\"manichaeans\",\"start\":0.48,\"end\":1.33}]"}
{"utterance_id": "3752-4944-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0026.flac", "task_type": "lexical_detection", "query": "roads", "prompt": "Process the full audio offline. Find every instance of the word \"roads\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.55, 1.04, "roads"]], "response_mode": "offline", "answer": "[{\"word\":\"roads\",\"start\":0.55,\"end\":1.04}]"}
{"utterance_id": "3576-138058-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0038.flac", "task_type": "lexical_detection", "query": "regent's", "prompt": "After the audio ends, extract all occurrences of the word \"regent's\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.62, 1.06, "regent's"]], "response_mode": "offline", "answer": "[{\"word\":\"regent's\",\"start\":0.62,\"end\":1.06}]"}
{"utterance_id": "174-168635-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0016.flac", "task_type": "lexical_detection", "query": "listening", "prompt": "Process the full audio offline. Find every instance of the word \"listening\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.24, 5.71, "listening"]], "response_mode": "offline", "answer": "[{\"word\":\"listening\",\"start\":5.24,\"end\":5.71}]"}
{"utterance_id": "1919-142785-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0041.flac", "task_type": "lexical_counting", "query": "together", "prompt": "When the utterance is complete, report how the word \"together\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.3, 3.83, "one"], [11.11, 11.63, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.3,\"end\":3.83},{\"word\":\"two\",\"start\":11.11,\"end\":11.63}]"}
{"utterance_id": "422-122949-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0035.flac", "task_type": "lexical_detection", "query": "choose", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"choose\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.67, 1.06, "choose"]], "response_mode": "offline", "answer": "[{\"word\":\"choose\",\"start\":0.67,\"end\":1.06}]"}
{"utterance_id": "8297-275156-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.74, 4.17, "randal"]], "response_mode": "offline", "answer": "[{\"word\":\"randal\",\"start\":3.74,\"end\":4.17}]"}
{"utterance_id": "1462-170145-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0022.flac", "task_type": "lexical_detection", "query": "truly", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"truly\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.41, 3.76, "truly"]], "response_mode": "offline", "answer": "[{\"word\":\"truly\",\"start\":3.41,\"end\":3.76}]"}
{"utterance_id": "8297-275156-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[10.31, 11.05, "catherine's"]], "response_mode": "offline", "answer": "[{\"word\":\"catherine's\",\"start\":10.31,\"end\":11.05}]"}
{"utterance_id": "7850-286674-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0010.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.92, 12.44, "food"]], "response_mode": "offline", "answer": "[{\"word\":\"food\",\"start\":11.92,\"end\":12.44}]"}
{"utterance_id": "1988-24833-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0002.flac", "task_type": "lexical_detection", "query": "sure", "prompt": "When the utterance is complete, report all instances of the word \"sure\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.91, 3.09, "sure"]], "response_mode": "offline", "answer": "[{\"word\":\"sure\",\"start\":2.91,\"end\":3.09}]"}
{"utterance_id": "777-126732-0050", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0050.flac", "task_type": "semantic_detection", "query": "money", "prompt": "After listening to the whole recording, output every instance of a money or currency term with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.8, 1.14, "cash"]], "response_mode": "offline", "answer": "[{\"word\":\"cash\",\"start\":0.8,\"end\":1.14}]"}
{"utterance_id": "6313-66125-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0000.flac", "task_type": "lexical_detection", "query": "you'll", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"you'll\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.22, 0.55, "you'll"]], "response_mode": "offline", "answer": "[{\"word\":\"you'll\",\"start\":0.22,\"end\":0.55}]"}
{"utterance_id": "2277-149874-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149874/2277-149874-0009.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Once the audio has finished, identify every instance of a family relation. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.43, 3.07, "offspring"]], "response_mode": "offline", "answer": "[{\"word\":\"offspring\",\"start\":2.43,\"end\":3.07}]"}
{"utterance_id": "6295-64301-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0009.flac", "task_type": "lexical_detection", "query": "mood", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"mood\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.65, 0.93, "mood"]], "response_mode": "offline", "answer": "[{\"word\":\"mood\",\"start\":0.65,\"end\":0.93}]"}
{"utterance_id": "84-121550-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121550/84-121550-0000.flac", "task_type": "lexical_detection", "query": "bore", "prompt": "Process the full audio offline. Find every instance of the word \"bore\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.73, 7.0, "bore"]], "response_mode": "offline", "answer": "[{\"word\":\"bore\",\"start\":6.73,\"end\":7.0}]"}
{"utterance_id": "1919-142785-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0038.flac", "task_type": "lexical_detection", "query": "pound", "prompt": "When the utterance is complete, report all instances of the word \"pound\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.37, 0.73, "pound"]], "response_mode": "offline", "answer": "[{\"word\":\"pound\",\"start\":0.37,\"end\":0.73}]"}
{"utterance_id": "2428-83705-0029", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0029.flac", "task_type": "lexical_detection", "query": "offerings", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"offerings\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.0, 5.56, "offerings"]], "response_mode": "offline", "answer": "[{\"word\":\"offerings\",\"start\":5.0,\"end\":5.56}]"}
{"utterance_id": "777-126732-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0011.flac", "task_type": "lexical_detection", "query": "property", "prompt": "When the utterance is complete, report all instances of the word \"property\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.1, 1.56, "property"]], "response_mode": "offline", "answer": "[{\"word\":\"property\",\"start\":1.1,\"end\":1.56}]"}
{"utterance_id": "5536-43359-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0000.flac", "task_type": "lexical_detection", "query": "national", "prompt": "After the audio ends, extract all occurrences of the word \"national\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.41, "national"]], "response_mode": "offline", "answer": "[{\"word\":\"national\",\"start\":1.02,\"end\":1.41}]"}
{"utterance_id": "777-126732-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0022.flac", "task_type": "lexical_detection", "query": "extraordinary", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"extraordinary\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.48, 1.02, "extraordinary"]], "response_mode": "offline", "answer": "[{\"word\":\"extraordinary\",\"start\":0.48,\"end\":1.02}]"}
{"utterance_id": "1988-148538-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0001.flac", "task_type": "lexical_detection", "query": "resort", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"resort\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.42, 4.84, "resort"]], "response_mode": "offline", "answer": "[{\"word\":\"resort\",\"start\":4.42,\"end\":4.84}]"}
{"utterance_id": "84-121123-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0013.flac", "task_type": "lexical_detection", "query": "morrel", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"morrel\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.38, 0.72, "morrel"]], "response_mode": "offline", "answer": "[{\"word\":\"morrel\",\"start\":0.38,\"end\":0.72}]"}
{"utterance_id": "2902-9006-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0013.flac", "task_type": "lexical_detection", "query": "learned", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"learned\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.26, 12.58, "learned"]], "response_mode": "offline", "answer": "[{\"word\":\"learned\",\"start\":12.26,\"end\":12.58}]"}
{"utterance_id": "251-137823-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0006.flac", "task_type": "lexical_detection", "query": "protect", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"protect\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.66, 2.02, "protect"]], "response_mode": "offline", "answer": "[{\"word\":\"protect\",\"start\":1.66,\"end\":2.02}]"}
{"utterance_id": "2803-161169-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0006.flac", "task_type": "lexical_detection", "query": "lamp", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lamp\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.23, 4.79, "lamp"]], "response_mode": "offline", "answer": "[{\"word\":\"lamp\",\"start\":4.23,\"end\":4.79}]"}
{"utterance_id": "777-126732-0079", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0079.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.73, 2.03, "one"], [5.27, 5.64, "half"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.73,\"end\":2.03},{\"word\":\"half\",\"start\":5.27,\"end\":5.64}]"}
{"utterance_id": "8842-302201-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0002.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.44, 1.7, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.44,\"end\":1.7}]"}
{"utterance_id": "3000-15664-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0005.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.28, 0.61, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.28,\"end\":0.61}]"}
{"utterance_id": "2035-152373-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0012.flac", "task_type": "semantic_detection", "query": "city", "prompt": "When the utterance is complete, report all instances of a city name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.65, 3.19, "iona"]], "response_mode": "offline", "answer": "[{\"word\":\"iona\",\"start\":2.65,\"end\":3.19}]"}
{"utterance_id": "2078-142845-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0051.flac", "task_type": "semantic_counting", "query": "food", "prompt": "When the utterance is complete, report how a food name accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.28, 6.75, "one"], [8.33, 8.8, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":6.28,\"end\":6.75},{\"word\":\"two\",\"start\":8.33,\"end\":8.8}]"}
{"utterance_id": "3752-4944-0067", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0067.flac", "task_type": "lexical_detection", "query": "aroma", "prompt": "When the utterance is complete, report all instances of the word \"aroma\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.41, 1.78, "aroma"]], "response_mode": "offline", "answer": "[{\"word\":\"aroma\",\"start\":1.41,\"end\":1.78}]"}
{"utterance_id": "6241-66616-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0001.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.01, "head"]], "response_mode": "offline", "answer": "[{\"word\":\"head\",\"start\":1.85,\"end\":2.01}]"}
{"utterance_id": "1673-143396-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0000.flac", "task_type": "lexical_detection", "query": "distinguished", "prompt": "Process the full audio offline. Find every instance of the word \"distinguished\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[10.04, 10.67, "distinguished"]], "response_mode": "offline", "answer": "[{\"word\":\"distinguished\",\"start\":10.04,\"end\":10.67}]"}
{"utterance_id": "1988-148538-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0015.flac", "task_type": "lexical_detection", "query": "precisely", "prompt": "Once the audio has finished, identify every instance of the word \"precisely\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.22, 3.78, "precisely"]], "response_mode": "offline", "answer": "[{\"word\":\"precisely\",\"start\":3.22,\"end\":3.78}]"}
{"utterance_id": "2078-142845-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0045.flac", "task_type": "lexical_detection", "query": "preserve", "prompt": "When the utterance is complete, report all instances of the word \"preserve\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.03, 5.58, "preserve"]], "response_mode": "offline", "answer": "[{\"word\":\"preserve\",\"start\":5.03,\"end\":5.58}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "lexical_detection", "query": "apostle", "prompt": "Once the audio has finished, identify every instance of the word \"apostle\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.26, 2.85, "apostle"]], "response_mode": "offline", "answer": "[{\"word\":\"apostle\",\"start\":2.26,\"end\":2.85}]"}
{"utterance_id": "777-126732-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0023.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.79, 5.25, "finger"]], "response_mode": "offline", "answer": "[{\"word\":\"finger\",\"start\":4.79,\"end\":5.25}]"}
{"utterance_id": "652-130726-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0012.flac", "task_type": "lexical_detection", "query": "restaurants", "prompt": "Process the full audio offline. Find every instance of the word \"restaurants\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.02, 1.62, "restaurants"]], "response_mode": "offline", "answer": "[{\"word\":\"restaurants\",\"start\":1.02,\"end\":1.62}]"}
{"utterance_id": "6313-76958-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0022.flac", "task_type": "lexical_counting", "query": "voice", "prompt": "Once the audio has finished, list each occurrence of the word \"voice\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[1.85, 2.16, "one"], [3.5, 3.86, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.85,\"end\":2.16},{\"word\":\"two\",\"start\":3.5,\"end\":3.86}]"}
{"utterance_id": "652-130726-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0002.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Do not answer until the audio ends. Then list each instance of a city name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[8.52, 8.83, "powell"]], "response_mode": "offline", "answer": "[{\"word\":\"powell\",\"start\":8.52,\"end\":8.83}]"}
{"utterance_id": "3000-15664-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0043.flac", "task_type": "lexical_detection", "query": "icy", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"icy\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.21, 1.55, "icy"]], "response_mode": "offline", "answer": "[{\"word\":\"icy\",\"start\":1.21,\"end\":1.55}]"}
{"utterance_id": "1462-170138-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0017.flac", "task_type": "semantic_detection", "query": "city", "prompt": "After listening to the whole recording, output every instance of a city name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.11, 5.61, "tokyo"]], "response_mode": "offline", "answer": "[{\"word\":\"tokyo\",\"start\":5.11,\"end\":5.61}]"}
{"utterance_id": "3536-8226-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "After the audio ends, extract all occurrences of an ordinal number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.42, 0.83, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.42,\"end\":0.83}]"}
{"utterance_id": "2428-83705-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0018.flac", "task_type": "semantic_counting", "query": "money", "prompt": "Listen to the full audio. Count every occurrence of a money or currency term. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 2.92, "one"], [4.91, 5.34, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.66,\"end\":2.92},{\"word\":\"two\",\"start\":4.91,\"end\":5.34}]"}
{"utterance_id": "251-118436-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0006.flac", "task_type": "semantic_counting", "query": "body_part", "prompt": "When the utterance is complete, report how a body part accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.87, 3.11, "one"], [6.33, 6.76, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":2.87,\"end\":3.11},{\"word\":\"two\",\"start\":6.33,\"end\":6.76}]"}
{"utterance_id": "3170-137482-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0025.flac", "task_type": "lexical_detection", "query": "modest", "prompt": "Process the full audio offline. Find every instance of the word \"modest\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.79, 6.27, "modest"]], "response_mode": "offline", "answer": "[{\"word\":\"modest\",\"start\":5.79,\"end\":6.27}]"}
{"utterance_id": "1272-135031-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0011.flac", "task_type": "lexical_detection", "query": "exactly", "prompt": "Once the audio has finished, identify every instance of the word \"exactly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.73, 1.41, "exactly"]], "response_mode": "offline", "answer": "[{\"word\":\"exactly\",\"start\":0.73,\"end\":1.41}]"}
{"utterance_id": "3170-137482-0032", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0032.flac", "task_type": "lexical_detection", "query": "obeyed", "prompt": "Process the full audio offline. Find every instance of the word \"obeyed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.6, 1.02, "obeyed"]], "response_mode": "offline", "answer": "[{\"word\":\"obeyed\",\"start\":0.6,\"end\":1.02}]"}
{"utterance_id": "2035-152373-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0002.flac", "task_type": "lexical_detection", "query": "christianity", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"christianity\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.55, 5.51, "christianity"]], "response_mode": "offline", "answer": "[{\"word\":\"christianity\",\"start\":4.55,\"end\":5.51}]"}
{"utterance_id": "6319-275224-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0012.flac", "task_type": "lexical_detection", "query": "flowers", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"flowers\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.71, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":2.95,\"end\":3.71}]"}
{"utterance_id": "1272-141231-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0025.flac", "task_type": "lexical_detection", "query": "defeat", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"defeat\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.65, 8.2, "defeat"]], "response_mode": "offline", "answer": "[{\"word\":\"defeat\",\"start\":7.65,\"end\":8.2}]"}
{"utterance_id": "7850-286674-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0013.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.75, 2.01, "snapping turtle"]], "response_mode": "offline", "answer": "[{\"word\":\"snapping turtle\",\"start\":0.75,\"end\":2.01}]"}
{"utterance_id": "3081-166546-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0039.flac", "task_type": "lexical_detection", "query": "sudden", "prompt": "When the utterance is complete, report all instances of the word \"sudden\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.03, 3.45, "sudden"]], "response_mode": "offline", "answer": "[{\"word\":\"sudden\",\"start\":3.03,\"end\":3.45}]"}
{"utterance_id": "1462-170145-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0008.flac", "task_type": "lexical_detection", "query": "paused", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"paused\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.58, 1.11, "paused"]], "response_mode": "offline", "answer": "[{\"word\":\"paused\",\"start\":0.58,\"end\":1.11}]"}
{"utterance_id": "1272-128104-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0000.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "When the utterance is complete, report all instances of a profession including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.52, 2.15, "apostle"]], "response_mode": "offline", "answer": "[{\"word\":\"apostle\",\"start\":1.52,\"end\":2.15}]"}
{"utterance_id": "5536-43358-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43358/5536-43358-0007.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.34, 8.94, "tobacco"]], "response_mode": "offline", "answer": "[{\"word\":\"tobacco\",\"start\":8.34,\"end\":8.94}]"}
{"utterance_id": "5895-34629-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0026.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "You will hear a complete utterance. When it ends, list every instance of a time of day you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.6, 2.04, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":1.6,\"end\":2.04}]"}
{"utterance_id": "6241-61943-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Listen to the full audio. After it finishes, find every instance of a body part. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.47, 2.94, "hands"]], "response_mode": "offline", "answer": "[{\"word\":\"hands\",\"start\":2.47,\"end\":2.94}]"}
{"utterance_id": "3853-163249-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0014.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Do not answer until the audio ends. Then list each instance of a number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.02, 7.27, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":7.02,\"end\":7.27}]"}
{"utterance_id": "2902-9006-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.84, 7.14, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":6.84,\"end\":7.14}]"}
{"utterance_id": "652-129742-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0014.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Process the full audio offline. Find every instance of a food name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.36, 1.71, "dressing"], [2.93, 3.36, "garlic"]], "response_mode": "offline", "answer": "[{\"word\":\"dressing\",\"start\":1.36,\"end\":1.71},{\"word\":\"garlic\",\"start\":2.93,\"end\":3.36}]"}
{"utterance_id": "422-122949-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0007.flac", "task_type": "lexical_detection", "query": "arouses", "prompt": "When the utterance is complete, report all instances of the word \"arouses\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.8, 4.39, "arouses"], [8.9, 9.41, "arouses"]], "response_mode": "offline", "answer": "[{\"word\":\"arouses\",\"start\":3.8,\"end\":4.39},{\"word\":\"arouses\",\"start\":8.9,\"end\":9.41}]"}
{"utterance_id": "2428-83705-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0013.flac", "task_type": "lexical_detection", "query": "suspicion", "prompt": "When the utterance is complete, report all instances of the word \"suspicion\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.73, "suspicion"]], "response_mode": "offline", "answer": "[{\"word\":\"suspicion\",\"start\":3.11,\"end\":3.73}]"}
{"utterance_id": "652-129742-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0001.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "When the utterance is complete, report all instances of an animal name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 0.96, "birds"]], "response_mode": "offline", "answer": "[{\"word\":\"birds\",\"start\":0.51,\"end\":0.96}]"}
{"utterance_id": "652-130737-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130737/652-130737-0003.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen to the full audio. After it finishes, find every instance of a country name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.05, "austrian"]], "response_mode": "offline", "answer": "[{\"word\":\"austrian\",\"start\":0.51,\"end\":1.05}]"}
{"utterance_id": "2803-154320-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0011.flac", "task_type": "lexical_counting", "query": "think", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"think\" with word/start/end. If none, reply [].", "labels": [[0.21, 0.61, "one"], [2.21, 2.45, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.21,\"end\":0.61},{\"word\":\"two\",\"start\":2.21,\"end\":2.45}]"}
{"utterance_id": "7976-105575-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0024.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Listen carefully to the entire clip, then report every instance of a plant name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.52, 3.01, "oak tree"]], "response_mode": "offline", "answer": "[{\"word\":\"oak tree\",\"start\":2.52,\"end\":3.01}]"}
{"utterance_id": "3576-138058-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 1.36, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":0.52,\"end\":1.36}]"}
{"utterance_id": "5694-64038-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0024.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.45, 2.82, "jim's"]], "response_mode": "offline", "answer": "[{\"word\":\"jim's\",\"start\":2.45,\"end\":2.82}]"}
{"utterance_id": "8297-275155-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0017.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "Process the full audio offline. For every occurrence of a family relation, emit a count-word event with timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.67, 6.16, "one"], [7.22, 7.82, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.67,\"end\":6.16},{\"word\":\"two\",\"start\":7.22,\"end\":7.82}]"}
{"utterance_id": "5694-64025-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0009.flac", "task_type": "lexical_detection", "query": "fighting", "prompt": "After the audio ends, extract all occurrences of the word \"fighting\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.51, 2.9, "fighting"]], "response_mode": "offline", "answer": "[{\"word\":\"fighting\",\"start\":2.51,\"end\":2.9}]"}
{"utterance_id": "3853-163249-0052", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0052.flac", "task_type": "lexical_detection", "query": "folded", "prompt": "When the utterance is complete, report all instances of the word \"folded\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.78, 10.16, "folded"]], "response_mode": "offline", "answer": "[{\"word\":\"folded\",\"start\":9.78,\"end\":10.16}]"}
{"utterance_id": "3576-138058-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0011.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "You will hear a complete utterance. When it ends, list every instance of a plant name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.16, 3.51, "tree"]], "response_mode": "offline", "answer": "[{\"word\":\"tree\",\"start\":3.16,\"end\":3.51}]"}
{"utterance_id": "652-129742-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0013.flac", "task_type": "lexical_detection", "query": "mustard", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"mustard\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.41, 3.92, "mustard"]], "response_mode": "offline", "answer": "[{\"word\":\"mustard\",\"start\":3.41,\"end\":3.92}]"}
{"utterance_id": "3536-8226-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0002.flac", "task_type": "semantic_detection", "query": "date", "prompt": "After the audio ends, extract all occurrences of a date. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.58, 5.03, "march"]], "response_mode": "offline", "answer": "[{\"word\":\"march\",\"start\":4.58,\"end\":5.03}]"}
{"utterance_id": "3752-4944-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0051.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.74, 3.39, "commandant"]], "response_mode": "offline", "answer": "[{\"word\":\"commandant\",\"start\":2.74,\"end\":3.39}]"}
{"utterance_id": "5895-34629-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0005.flac", "task_type": "lexical_detection", "query": "theatre", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"theatre\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.73, 1.3, "theatre"]], "response_mode": "offline", "answer": "[{\"word\":\"theatre\",\"start\":0.73,\"end\":1.3}]"}
{"utterance_id": "7850-73752-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0010.flac", "task_type": "lexical_detection", "query": "merited", "prompt": "Once the audio has finished, identify every instance of the word \"merited\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.64, 3.15, "merited"]], "response_mode": "offline", "answer": "[{\"word\":\"merited\",\"start\":2.64,\"end\":3.15}]"}
{"utterance_id": "6345-93302-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0001.flac", "task_type": "lexical_detection", "query": "sprang", "prompt": "After listening to the whole recording, output every instance of the word \"sprang\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.11, 6.56, "sprang"]], "response_mode": "offline", "answer": "[{\"word\":\"sprang\",\"start\":6.11,\"end\":6.56}]"}
{"utterance_id": "2086-149220-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0002.flac", "task_type": "lexical_detection", "query": "species", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"species\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.37, 1.87, "species"]], "response_mode": "offline", "answer": "[{\"word\":\"species\",\"start\":1.37,\"end\":1.87}]"}
{"utterance_id": "3752-4943-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0007.flac", "task_type": "lexical_detection", "query": "bleeding", "prompt": "Process the full audio offline. Find every instance of the word \"bleeding\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.53, 3.11, "bleeding"]], "response_mode": "offline", "answer": "[{\"word\":\"bleeding\",\"start\":2.53,\"end\":3.11}]"}
{"utterance_id": "6313-66129-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0033.flac", "task_type": "lexical_detection", "query": "saddle", "prompt": "After listening to the whole recording, output every instance of the word \"saddle\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.86, 6.23, "saddle"]], "response_mode": "offline", "answer": "[{\"word\":\"saddle\",\"start\":5.86,\"end\":6.23}]"}
{"utterance_id": "5536-43359-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5536/43359/5536-43359-0010.flac", "task_type": "lexical_counting", "query": "daily", "prompt": "Once the audio has finished, list each occurrence of the word \"daily\" as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.42, 0.72, "one"], [2.81, 3.12, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.42,\"end\":0.72},{\"word\":\"two\",\"start\":2.81,\"end\":3.12}]"}
{"utterance_id": "652-129742-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0013.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Do not answer until the audio ends. Then list each instance of a plant name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.9, 1.5, "lettuce leaf"]], "response_mode": "offline", "answer": "[{\"word\":\"lettuce leaf\",\"start\":0.9,\"end\":1.5}]"}
{"utterance_id": "2277-149896-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0021.flac", "task_type": "lexical_detection", "query": "confounded", "prompt": "Once the audio has finished, identify every instance of the word \"confounded\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.01, 2.64, "confounded"]], "response_mode": "offline", "answer": "[{\"word\":\"confounded\",\"start\":2.01,\"end\":2.64}]"}
{"utterance_id": "2086-149220-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0031.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.83, 8.21, "eye"]], "response_mode": "offline", "answer": "[{\"word\":\"eye\",\"start\":7.83,\"end\":8.21}]"}
{"utterance_id": "1993-147149-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0005.flac", "task_type": "lexical_detection", "query": "latch", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"latch\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.94, 2.31, "latch"]], "response_mode": "offline", "answer": "[{\"word\":\"latch\",\"start\":1.94,\"end\":2.31}]"}
{"utterance_id": "5694-64038-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64038/5694-64038-0005.flac", "task_type": "lexical_detection", "query": "boys", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"boys\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.96, 2.55, "boys"]], "response_mode": "offline", "answer": "[{\"word\":\"boys\",\"start\":1.96,\"end\":2.55}]"}
{"utterance_id": "6241-61946-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61946/6241-61946-0019.flac", "task_type": "lexical_detection", "query": "uncle", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"uncle\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.66, 0.95, "uncle"]], "response_mode": "offline", "answer": "[{\"word\":\"uncle\",\"start\":0.66,\"end\":0.95}]"}
{"utterance_id": "3853-163249-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0006.flac", "task_type": "lexical_detection", "query": "acquainted", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"acquainted\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.37, 7.87, "acquainted"]], "response_mode": "offline", "answer": "[{\"word\":\"acquainted\",\"start\":7.37,\"end\":7.87}]"}
{"utterance_id": "777-126732-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0051.flac", "task_type": "lexical_detection", "query": "law", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"law\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.79, 2.1, "law"]], "response_mode": "offline", "answer": "[{\"word\":\"law\",\"start\":1.79,\"end\":2.1}]"}
{"utterance_id": "8297-275156-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0011.flac", "task_type": "lexical_detection", "query": "early", "prompt": "When the utterance is complete, report all instances of the word \"early\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.37, 2.63, "early"]], "response_mode": "offline", "answer": "[{\"word\":\"early\",\"start\":2.37,\"end\":2.63}]"}
{"utterance_id": "1993-147149-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0009.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Do not answer until the audio ends. Then list each instance of a family relation as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.37, 1.72, "father"], [9.58, 9.93, "son"]], "response_mode": "offline", "answer": "[{\"word\":\"father\",\"start\":1.37,\"end\":1.72},{\"word\":\"son\",\"start\":9.58,\"end\":9.93}]"}
{"utterance_id": "5694-64025-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0018.flac", "task_type": "lexical_detection", "query": "wise", "prompt": "Once the audio has finished, identify every instance of the word \"wise\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.61, 1.11, "wise"]], "response_mode": "offline", "answer": "[{\"word\":\"wise\",\"start\":0.61,\"end\":1.11}]"}
{"utterance_id": "652-130726-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0020.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Once the audio has finished, identify every instance of an organization name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.61, 3.09, "odeon"]], "response_mode": "offline", "answer": "[{\"word\":\"odeon\",\"start\":2.61,\"end\":3.09}]"}
{"utterance_id": "8842-302201-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302201/8842-302201-0015.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "You will hear a complete utterance. When it ends, list every instance of a body part you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.01, 4.37, "eyes"], [7.07, 7.47, "mouth"]], "response_mode": "offline", "answer": "[{\"word\":\"eyes\",\"start\":4.01,\"end\":4.37},{\"word\":\"mouth\",\"start\":7.07,\"end\":7.47}]"}
{"utterance_id": "2277-149896-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0016.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "When the utterance is complete, report all instances of a family relation including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.68, 2.03, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":1.68,\"end\":2.03}]"}
{"utterance_id": "8297-275154-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0007.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.5, 4.0, "herbert"]], "response_mode": "offline", "answer": "[{\"word\":\"herbert\",\"start\":3.5,\"end\":4.0}]"}
{"utterance_id": "777-126732-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.9, "lombroso"]], "response_mode": "offline", "answer": "[{\"word\":\"lombroso\",\"start\":0.33,\"end\":0.9}]"}
{"utterance_id": "8842-302203-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0005.flac", "task_type": "lexical_detection", "query": "bewilderment", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"bewilderment\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.66, "bewilderment"]], "response_mode": "offline", "answer": "[{\"word\":\"bewilderment\",\"start\":1.14,\"end\":1.66}]"}
{"utterance_id": "6319-64726-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0006.flac", "task_type": "lexical_detection", "query": "awake", "prompt": "After the audio ends, extract all occurrences of the word \"awake\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.69, 2.13, "awake"]], "response_mode": "offline", "answer": "[{\"word\":\"awake\",\"start\":1.69,\"end\":2.13}]"}
{"utterance_id": "2412-153947-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0013.flac", "task_type": "lexical_detection", "query": "edition", "prompt": "After the audio ends, extract all occurrences of the word \"edition\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.98, 1.36, "edition"]], "response_mode": "offline", "answer": "[{\"word\":\"edition\",\"start\":0.98,\"end\":1.36}]"}
{"utterance_id": "2412-153947-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0002.flac", "task_type": "lexical_detection", "query": "avoided", "prompt": "Process the full audio offline. Find every instance of the word \"avoided\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.03, 2.43, "avoided"]], "response_mode": "offline", "answer": "[{\"word\":\"avoided\",\"start\":2.03,\"end\":2.43}]"}
{"utterance_id": "5338-24615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0006.flac", "task_type": "lexical_detection", "query": "painter", "prompt": "Once the audio has finished, identify every instance of the word \"painter\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.91, 2.46, "painter"]], "response_mode": "offline", "answer": "[{\"word\":\"painter\",\"start\":1.91,\"end\":2.46}]"}
{"utterance_id": "6345-64257-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the full audio. After it finishes, find every instance of a profession. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.29, 1.98, "servants"]], "response_mode": "offline", "answer": "[{\"word\":\"servants\",\"start\":1.29,\"end\":1.98}]"}
{"utterance_id": "251-136532-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/136532/251-136532-0019.flac", "task_type": "lexical_detection", "query": "interested", "prompt": "After listening to the whole recording, output every instance of the word \"interested\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.15, 1.72, "interested"]], "response_mode": "offline", "answer": "[{\"word\":\"interested\",\"start\":1.15,\"end\":1.72}]"}
{"utterance_id": "1988-24833-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0025.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.07, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.82,\"end\":2.07}]"}
{"utterance_id": "1988-24833-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0009.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of an organization name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.73, 1.07, "schools"]], "response_mode": "offline", "answer": "[{\"word\":\"schools\",\"start\":0.73,\"end\":1.07}]"}
{"utterance_id": "6319-275224-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/275224/6319-275224-0020.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "After the audio ends, extract all occurrences of a plant name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.57, 9.17, "flowers"]], "response_mode": "offline", "answer": "[{\"word\":\"flowers\",\"start\":8.57,\"end\":9.17}]"}
{"utterance_id": "2428-83705-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0022.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.82, 2.04, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.82,\"end\":2.04}]"}
{"utterance_id": "1919-142785-0040", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0040.flac", "task_type": "lexical_detection", "query": "flavour", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"flavour\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.21, 7.68, "flavour"]], "response_mode": "offline", "answer": "[{\"word\":\"flavour\",\"start\":7.21,\"end\":7.68}]"}
{"utterance_id": "3170-137482-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0043.flac", "task_type": "lexical_detection", "query": "intimate", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"intimate\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[7.02, 7.48, "intimate"]], "response_mode": "offline", "answer": "[{\"word\":\"intimate\",\"start\":7.02,\"end\":7.48}]"}
{"utterance_id": "1919-142785-0014", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0014.flac", "task_type": "lexical_detection", "query": "stirring", "prompt": "When the utterance is complete, report all instances of the word \"stirring\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.43, 3.81, "stirring"]], "response_mode": "offline", "answer": "[{\"word\":\"stirring\",\"start\":3.43,\"end\":3.81}]"}
{"utterance_id": "1673-143396-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0008.flac", "task_type": "lexical_detection", "query": "portion", "prompt": "After the audio ends, extract all occurrences of the word \"portion\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.91, 6.28, "portion"]], "response_mode": "offline", "answer": "[{\"word\":\"portion\",\"start\":5.91,\"end\":6.28}]"}
{"utterance_id": "6319-64726-0018", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0018.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.66, 3.14, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":2.66,\"end\":3.14}]"}
{"utterance_id": "3000-15664-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0009.flac", "task_type": "lexical_detection", "query": "weary", "prompt": "After listening to the whole recording, output every instance of the word \"weary\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.42, 5.94, "weary"]], "response_mode": "offline", "answer": "[{\"word\":\"weary\",\"start\":5.42,\"end\":5.94}]"}
{"utterance_id": "7976-110523-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0000.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "After listening to the whole recording, output every instance of a family relation with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.57, 13.02, "wife"]], "response_mode": "offline", "answer": "[{\"word\":\"wife\",\"start\":12.57,\"end\":13.02}]"}
{"utterance_id": "174-50561-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0005.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "Once the audio has finished, identify every instance of a plant name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.89, 2.26, "rose"]], "response_mode": "offline", "answer": "[{\"word\":\"rose\",\"start\":1.89,\"end\":2.26}]"}
{"utterance_id": "3853-163249-0045", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0045.flac", "task_type": "lexical_detection", "query": "tongue", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"tongue\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.2, 6.51, "tongue"]], "response_mode": "offline", "answer": "[{\"word\":\"tongue\",\"start\":6.2,\"end\":6.51}]"}
{"utterance_id": "5694-64025-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0020.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "After the audio ends, extract all occurrences of an animal name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.41, 0.74, "mule"]], "response_mode": "offline", "answer": "[{\"word\":\"mule\",\"start\":0.41,\"end\":0.74}]"}
{"utterance_id": "652-130726-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0019.flac", "task_type": "lexical_detection", "query": "prices", "prompt": "After listening to the whole recording, output every instance of the word \"prices\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.88, "prices"]], "response_mode": "offline", "answer": "[{\"word\":\"prices\",\"start\":0.5,\"end\":0.88}]"}
{"utterance_id": "3853-163249-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0017.flac", "task_type": "lexical_detection", "query": "lads", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"lads\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.03, 7.54, "lads"]], "response_mode": "offline", "answer": "[{\"word\":\"lads\",\"start\":7.03,\"end\":7.54}]"}
{"utterance_id": "174-50561-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0011.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Once the audio has finished, identify every instance of a color name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[1.45, 1.74, "yellow"]], "response_mode": "offline", "answer": "[{\"word\":\"yellow\",\"start\":1.45,\"end\":1.74}]"}
{"utterance_id": "8297-275156-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0012.flac", "task_type": "lexical_detection", "query": "postscript", "prompt": "Process the full audio offline. Find every instance of the word \"postscript\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.16, 2.96, "postscript"]], "response_mode": "offline", "answer": "[{\"word\":\"postscript\",\"start\":2.16,\"end\":2.96}]"}
{"utterance_id": "6241-66616-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0021.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Once the audio has finished, identify every instance of a city name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.6, 3.08, "detroit"]], "response_mode": "offline", "answer": "[{\"word\":\"detroit\",\"start\":2.6,\"end\":3.08}]"}
{"utterance_id": "1919-142785-0048", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0048.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen carefully to the entire clip, then report every instance of a country name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.99, "french"]], "response_mode": "offline", "answer": "[{\"word\":\"french\",\"start\":0.52,\"end\":0.99}]"}
{"utterance_id": "6313-66125-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0013.flac", "task_type": "lexical_detection", "query": "tilted", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"tilted\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.54, 0.87, "tilted"]], "response_mode": "offline", "answer": "[{\"word\":\"tilted\",\"start\":0.54,\"end\":0.87}]"}
{"utterance_id": "2428-83699-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0030.flac", "task_type": "lexical_detection", "query": "interval", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"interval\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.87, 3.35, "interval"]], "response_mode": "offline", "answer": "[{\"word\":\"interval\",\"start\":2.87,\"end\":3.35}]"}
{"utterance_id": "1272-141231-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0010.flac", "task_type": "lexical_detection", "query": "particularly", "prompt": "Once the audio has finished, identify every instance of the word \"particularly\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.5, 1.26, "particularly"]], "response_mode": "offline", "answer": "[{\"word\":\"particularly\",\"start\":0.5,\"end\":1.26}]"}
{"utterance_id": "1919-142785-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[4.73, 5.25, "virgil"]], "response_mode": "offline", "answer": "[{\"word\":\"virgil\",\"start\":4.73,\"end\":5.25}]"}
{"utterance_id": "7976-105575-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/105575/7976-105575-0013.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the full audio. After it finishes, find every instance of a time of day. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.64, 1.06, "morning"]], "response_mode": "offline", "answer": "[{\"word\":\"morning\",\"start\":0.64,\"end\":1.06}]"}
{"utterance_id": "251-118436-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/118436/251-118436-0001.flac", "task_type": "lexical_detection", "query": "cried", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"cried\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.9, 3.49, "cried"]], "response_mode": "offline", "answer": "[{\"word\":\"cried\",\"start\":2.9,\"end\":3.49}]"}
{"utterance_id": "3853-163249-0036", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0036.flac", "task_type": "semantic_counting", "query": "kinship", "prompt": "After listening to the whole recording, output timed count words for a family relation (one, two, three, ...) as JSON. If none, reply [].", "labels": [[0.42, 0.66, "one"], [4.47, 4.9, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.42,\"end\":0.66},{\"word\":\"two\",\"start\":4.47,\"end\":4.9}]"}
{"utterance_id": "1673-143397-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143397/1673-143397-0011.flac", "task_type": "semantic_counting", "query": "person_name", "prompt": "Once the audio has finished, list each occurrence of a person's name as {\"word\": <count>, \"start\", \"end\"} in order. If none, reply [].", "labels": [[0.31, 1.04, "one"], [5.38, 6.05, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.31,\"end\":1.04},{\"word\":\"two\",\"start\":5.38,\"end\":6.05}]"}
{"utterance_id": "2902-9006-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0013.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Once the audio has finished, identify every instance of a country name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.44, 6.94, "italy"]], "response_mode": "offline", "answer": "[{\"word\":\"italy\",\"start\":6.44,\"end\":6.94}]"}
{"utterance_id": "7850-281318-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/281318/7850-281318-0012.flac", "task_type": "lexical_detection", "query": "build", "prompt": "After the audio ends, extract all occurrences of the word \"build\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.94, 3.23, "build"]], "response_mode": "offline", "answer": "[{\"word\":\"build\",\"start\":2.94,\"end\":3.23}]"}
{"utterance_id": "8297-275154-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.11, 1.65, "master"]], "response_mode": "offline", "answer": "[{\"word\":\"master\",\"start\":1.11,\"end\":1.65}]"}
{"utterance_id": "6319-64726-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0005.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "When the utterance is complete, report all instances of a body part including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.29, 6.73, "hand"]], "response_mode": "offline", "answer": "[{\"word\":\"hand\",\"start\":6.29,\"end\":6.73}]"}
{"utterance_id": "6313-66125-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0007.flac", "task_type": "lexical_detection", "query": "thomas", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"thomas\" with fields word, start, end (seconds). If none, reply [].", "labels": [[4.17, 4.7, "thomas"]], "response_mode": "offline", "answer": "[{\"word\":\"thomas\",\"start\":4.17,\"end\":4.7}]"}
{"utterance_id": "5338-24640-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0008.flac", "task_type": "semantic_detection", "query": "organization", "prompt": "After listening to the whole recording, output every instance of an organization name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.6, 7.29, "government"]], "response_mode": "offline", "answer": "[{\"word\":\"government\",\"start\":6.6,\"end\":7.29}]"}
{"utterance_id": "84-121123-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0012.flac", "task_type": "lexical_detection", "query": "man's", "prompt": "Once the audio has finished, identify every instance of the word \"man's\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.56, 0.84, "man's"]], "response_mode": "offline", "answer": "[{\"word\":\"man's\",\"start\":0.56,\"end\":0.84}]"}
{"utterance_id": "777-126732-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0025.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.53, 1.33, "michaelis"]], "response_mode": "offline", "answer": "[{\"word\":\"michaelis\",\"start\":0.53,\"end\":1.33}]"}
{"utterance_id": "1462-170142-0038", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0038.flac", "task_type": "lexical_detection", "query": "arm", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"arm\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.42, 1.67, "arm"]], "response_mode": "offline", "answer": "[{\"word\":\"arm\",\"start\":1.42,\"end\":1.67}]"}
{"utterance_id": "1988-24833-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0016.flac", "task_type": "lexical_detection", "query": "fault", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"fault\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.75, 1.18, "fault"]], "response_mode": "offline", "answer": "[{\"word\":\"fault\",\"start\":0.75,\"end\":1.18}]"}
{"utterance_id": "2412-153947-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0010.flac", "task_type": "lexical_counting", "query": "seventy", "prompt": "When the utterance is complete, report how the word \"seventy\" accumulated over time using count words and spans. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 5.89, "one"], [16.02, 16.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.55,\"end\":5.89},{\"word\":\"two\",\"start\":16.02,\"end\":16.36}]"}
{"utterance_id": "6295-64301-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0022.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.48, 10.17, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":9.48,\"end\":10.17}]"}
{"utterance_id": "1993-147149-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0013.flac", "task_type": "lexical_detection", "query": "alice", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"alice\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.99, 1.5, "alice"]], "response_mode": "offline", "answer": "[{\"word\":\"alice\",\"start\":0.99,\"end\":1.5}]"}
{"utterance_id": "2412-153954-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0021.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Listen to the full audio. After it finishes, find every instance of an ordinal number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.18, 2.42, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":2.18,\"end\":2.42}]"}
{"utterance_id": "1462-170142-0035", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170142/1462-170142-0035.flac", "task_type": "lexical_detection", "query": "bartley", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"bartley\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.35, "bartley"]], "response_mode": "offline", "answer": "[{\"word\":\"bartley\",\"start\":0.59,\"end\":1.35}]"}
{"utterance_id": "2277-149896-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149896/2277-149896-0015.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.95, 3.41, "carrie"]], "response_mode": "offline", "answer": "[{\"word\":\"carrie\",\"start\":2.95,\"end\":3.41}]"}
{"utterance_id": "2035-147961-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0023.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Do not answer until the audio ends. Then list each instance of an animal name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.52, 3.05, "blacks"]], "response_mode": "offline", "answer": "[{\"word\":\"blacks\",\"start\":2.52,\"end\":3.05}]"}
{"utterance_id": "1673-143396-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0017.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "You will hear a complete utterance. When it ends, list every instance of a profession you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.5, 1.88, "doctor"]], "response_mode": "offline", "answer": "[{\"word\":\"doctor\",\"start\":1.5,\"end\":1.88}]"}
{"utterance_id": "1919-142785-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0002.flac", "task_type": "lexical_detection", "query": "originally", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"originally\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[0.27, 0.93, "originally"]], "response_mode": "offline", "answer": "[{\"word\":\"originally\",\"start\":0.27,\"end\":0.93}]"}
{"utterance_id": "6319-57405-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0012.flac", "task_type": "lexical_detection", "query": "noble", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"noble\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.97, 4.43, "noble"]], "response_mode": "offline", "answer": "[{\"word\":\"noble\",\"start\":3.97,\"end\":4.43}]"}
{"utterance_id": "1988-24833-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/24833/1988-24833-0004.flac", "task_type": "lexical_detection", "query": "program", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"program\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.7, 2.28, "program"]], "response_mode": "offline", "answer": "[{\"word\":\"program\",\"start\":1.7,\"end\":2.28}]"}
{"utterance_id": "1993-147149-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0024.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "You will hear a complete utterance. When it ends, list every instance of a family relation you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.79, 10.06, "aunt's"]], "response_mode": "offline", "answer": "[{\"word\":\"aunt's\",\"start\":9.79,\"end\":10.06}]"}
{"utterance_id": "3576-138058-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0002.flac", "task_type": "lexical_counting", "query": "dulcinea", "prompt": "Listen to the full audio. Count every occurrence of the word \"dulcinea\". After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.42, 6.3, "one"], [21.05, 21.85, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.42,\"end\":6.3},{\"word\":\"two\",\"start\":21.05,\"end\":21.85}]"}
{"utterance_id": "84-121123-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.18, 2.77, "villefort"]], "response_mode": "offline", "answer": "[{\"word\":\"villefort\",\"start\":2.18,\"end\":2.77}]"}
{"utterance_id": "5694-64025-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5694/64025/5694-64025-0002.flac", "task_type": "lexical_detection", "query": "heroes", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"heroes\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.63, 4.34, "heroes"]], "response_mode": "offline", "answer": "[{\"word\":\"heroes\",\"start\":3.63,\"end\":4.34}]"}
{"utterance_id": "3576-138058-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Do not answer until the audio ends. Then list each instance of a person's name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.1, 1.53, "sancho"]], "response_mode": "offline", "answer": "[{\"word\":\"sancho\",\"start\":1.1,\"end\":1.53}]"}
{"utterance_id": "3081-166546-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3081/166546/3081-166546-0033.flac", "task_type": "lexical_detection", "query": "none", "prompt": "When the utterance is complete, report all instances of the word \"none\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.13, 2.63, "none"]], "response_mode": "offline", "answer": "[{\"word\":\"none\",\"start\":2.13,\"end\":2.63}]"}
{"utterance_id": "7976-110124-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110124/7976-110124-0010.flac", "task_type": "lexical_detection", "query": "elder", "prompt": "When the utterance is complete, report all instances of the word \"elder\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.49, 3.81, "elder"]], "response_mode": "offline", "answer": "[{\"word\":\"elder\",\"start\":3.49,\"end\":3.81}]"}
{"utterance_id": "2078-142845-0043", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0043.flac", "task_type": "lexical_detection", "query": "currants", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"currants\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[11.26, 11.96, "currants"]], "response_mode": "offline", "answer": "[{\"word\":\"currants\",\"start\":11.26,\"end\":11.96}]"}
{"utterance_id": "652-129742-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0017.flac", "task_type": "semantic_detection", "query": "plant", "prompt": "When the utterance is complete, report all instances of a plant name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.89, 1.47, "watercress"]], "response_mode": "offline", "answer": "[{\"word\":\"watercress\",\"start\":0.89,\"end\":1.47}]"}
{"utterance_id": "6313-66129-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0009.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the full audio. After it finishes, find every instance of a person's name. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.29, 1.24, "president brown"]], "response_mode": "offline", "answer": "[{\"word\":\"president brown\",\"start\":0.29,\"end\":1.24}]"}
{"utterance_id": "6319-64726-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/64726/6319-64726-0001.flac", "task_type": "semantic_counting", "query": "animal", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for an animal name with word/start/end. If none, reply [].", "labels": [[1.18, 1.74, "one"], [13.97, 14.36, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.18,\"end\":1.74},{\"word\":\"two\",\"start\":13.97,\"end\":14.36}]"}
{"utterance_id": "2412-153947-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0003.flac", "task_type": "lexical_detection", "query": "books", "prompt": "Process the full audio offline. Find every instance of the word \"books\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.55, 5.93, "books"]], "response_mode": "offline", "answer": "[{\"word\":\"books\",\"start\":5.55,\"end\":5.93}]"}
{"utterance_id": "5338-24640-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24640/5338-24640-0002.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[2.56, 3.27, "mister morton"]], "response_mode": "offline", "answer": "[{\"word\":\"mister morton\",\"start\":2.56,\"end\":3.27}]"}
{"utterance_id": "6345-93306-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0000.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Once the audio has finished, identify every instance of an ordinal number. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.56, 1.05, "last"]], "response_mode": "offline", "answer": "[{\"word\":\"last\",\"start\":0.56,\"end\":1.05}]"}
{"utterance_id": "174-50561-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0017.flac", "task_type": "lexical_detection", "query": "becomes", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"becomes\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.14, 1.58, "becomes"]], "response_mode": "offline", "answer": "[{\"word\":\"becomes\",\"start\":1.14,\"end\":1.58}]"}
{"utterance_id": "3576-138058-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0000.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After listening to the whole recording, output every instance of a food name with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.02, 9.27, "meal"]], "response_mode": "offline", "answer": "[{\"word\":\"meal\",\"start\":9.02,\"end\":9.27}]"}
{"utterance_id": "2803-154320-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154320/2803-154320-0003.flac", "task_type": "lexical_detection", "query": "route", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"route\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.5, 4.0, "route"]], "response_mode": "offline", "answer": "[{\"word\":\"route\",\"start\":3.5,\"end\":4.0}]"}
{"utterance_id": "7976-110523-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0009.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen carefully to the entire clip, then report every instance of a time of day as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.93, 3.5, "evening"]], "response_mode": "offline", "answer": "[{\"word\":\"evening\",\"start\":2.93,\"end\":3.5}]"}
{"utterance_id": "3853-163249-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0028.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Once the audio has finished, identify every instance of a body part. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[6.87, 7.36, "stommick"]], "response_mode": "offline", "answer": "[{\"word\":\"stommick\",\"start\":6.87,\"end\":7.36}]"}
{"utterance_id": "7976-110523-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0008.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen to the full audio. After it finishes, find every instance of a family relation. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.65, 2.16, "stepmother"]], "response_mode": "offline", "answer": "[{\"word\":\"stepmother\",\"start\":1.65,\"end\":2.16}]"}
{"utterance_id": "1462-170138-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.27, 0.88, "alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"alexander\",\"start\":0.27,\"end\":0.88}]"}
{"utterance_id": "2277-149897-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0019.flac", "task_type": "semantic_detection", "query": "clock_time", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a time of day with fields word, start, end (seconds). If none, reply [].", "labels": [[0.44, 1.08, "three o'clock"]], "response_mode": "offline", "answer": "[{\"word\":\"three o'clock\",\"start\":0.44,\"end\":1.08}]"}
{"utterance_id": "422-122949-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0023.flac", "task_type": "lexical_detection", "query": "awaken", "prompt": "Once the audio has finished, identify every instance of the word \"awaken\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.89, 3.37, "awaken"]], "response_mode": "offline", "answer": "[{\"word\":\"awaken\",\"start\":2.89,\"end\":3.37}]"}
{"utterance_id": "6313-76958-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/76958/6313-76958-0003.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.67, 2.08, "chunky's"]], "response_mode": "offline", "answer": "[{\"word\":\"chunky's\",\"start\":1.67,\"end\":2.08}]"}
{"utterance_id": "8297-275156-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0008.flac", "task_type": "lexical_detection", "query": "results", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"results\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.85, 1.49, "results"]], "response_mode": "offline", "answer": "[{\"word\":\"results\",\"start\":0.85,\"end\":1.49}]"}
{"utterance_id": "1272-128104-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/128104/1272-128104-0002.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[11.54, 12.0, "mind"]], "response_mode": "offline", "answer": "[{\"word\":\"mind\",\"start\":11.54,\"end\":12.0}]"}
{"utterance_id": "6241-61943-0026", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0026.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Do not answer until the audio ends. Then list each instance of a food name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.93, 6.6, "codfish"]], "response_mode": "offline", "answer": "[{\"word\":\"codfish\",\"start\":5.93,\"end\":6.6}]"}
{"utterance_id": "3853-163249-0039", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0039.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.01, 2.84, "missus sterling"]], "response_mode": "offline", "answer": "[{\"word\":\"missus sterling\",\"start\":2.01,\"end\":2.84}]"}
{"utterance_id": "6345-93306-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0013.flac", "task_type": "lexical_detection", "query": "packed", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"packed\" with fields word, start, end (seconds). If none, reply [].", "labels": [[7.31, 7.64, "packed"]], "response_mode": "offline", "answer": "[{\"word\":\"packed\",\"start\":7.31,\"end\":7.64}]"}
{"utterance_id": "2902-9008-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9008/2902-9008-0015.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Process the full audio offline. Find every instance of a profession and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.53, "prefect"]], "response_mode": "offline", "answer": "[{\"word\":\"prefect\",\"start\":2.0,\"end\":2.53}]"}
{"utterance_id": "3170-137482-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0003.flac", "task_type": "semantic_detection", "query": "money", "prompt": "After the audio ends, extract all occurrences of a money or currency term. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.25, 6.57, "money"], [9.68, 9.93, "pay"], [12.74, 13.59, "expenses"]], "response_mode": "offline", "answer": "[{\"word\":\"money\",\"start\":6.25,\"end\":6.57},{\"word\":\"pay\",\"start\":9.68,\"end\":9.93},{\"word\":\"expenses\",\"start\":12.74,\"end\":13.59}]"}
{"utterance_id": "6241-61943-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/61943/6241-61943-0001.flac", "task_type": "lexical_detection", "query": "suffer", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"suffer\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[2.28, 2.73, "suffer"]], "response_mode": "offline", "answer": "[{\"word\":\"suffer\",\"start\":2.28,\"end\":2.73}]"}
{"utterance_id": "8842-302203-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0008.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "You will hear a complete utterance. When it ends, list every instance of a number you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.89, 4.23, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":3.89,\"end\":4.23}]"}
{"utterance_id": "6345-64257-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/64257/6345-64257-0005.flac", "task_type": "lexical_counting", "query": "large", "prompt": "Listen to the entire clip, then report the count sequence for the word \"large\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.7, "one"], [5.92, 6.33, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.27,\"end\":1.7},{\"word\":\"two\",\"start\":5.92,\"end\":6.33}]"}
{"utterance_id": "1919-142785-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1919/142785/1919-142785-0008.flac", "task_type": "lexical_counting", "query": "fire", "prompt": "Listen to the entire clip, then report the count sequence for the word \"fire\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.52, 4.92, "one"], [15.44, 15.83, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":4.52,\"end\":4.92},{\"word\":\"two\",\"start\":15.44,\"end\":15.83}]"}
{"utterance_id": "2086-149220-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0009.flac", "task_type": "lexical_detection", "query": "existed", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"existed\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.49, 2.02, "existed"]], "response_mode": "offline", "answer": "[{\"word\":\"existed\",\"start\":1.49,\"end\":2.02}]"}
{"utterance_id": "652-130726-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0004.flac", "task_type": "lexical_counting", "query": "always", "prompt": "Do not answer until the audio ends. Then return a JSON list of count events for the word \"always\" with word/start/end. If none, reply [].", "labels": [[0.99, 1.33, "one"], [4.34, 4.74, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.99,\"end\":1.33},{\"word\":\"two\",\"start\":4.34,\"end\":4.74}]"}
{"utterance_id": "7850-73752-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/73752/7850-73752-0004.flac", "task_type": "lexical_detection", "query": "happiness", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"happiness\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.96, 3.61, "happiness"]], "response_mode": "offline", "answer": "[{\"word\":\"happiness\",\"start\":2.96,\"end\":3.61}]"}
{"utterance_id": "3000-15664-0041", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0041.flac", "task_type": "lexical_detection", "query": "thence", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"thence\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.63, 8.07, "thence"]], "response_mode": "offline", "answer": "[{\"word\":\"thence\",\"start\":7.63,\"end\":8.07}]"}
{"utterance_id": "6241-66616-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0022.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Do not answer until the audio ends. Then list each instance of a body part as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[5.26, 5.58, "scalp"]], "response_mode": "offline", "answer": "[{\"word\":\"scalp\",\"start\":5.26,\"end\":5.58}]"}
{"utterance_id": "3576-138058-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0009.flac", "task_type": "lexical_detection", "query": "entirely", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"entirely\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.01, 9.8, "entirely"]], "response_mode": "offline", "answer": "[{\"word\":\"entirely\",\"start\":9.01,\"end\":9.8}]"}
{"utterance_id": "3752-4944-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0024.flac", "task_type": "lexical_detection", "query": "mister", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"mister\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.14, 1.4, "mister"]], "response_mode": "offline", "answer": "[{\"word\":\"mister\",\"start\":1.14,\"end\":1.4}]"}
{"utterance_id": "6345-93306-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0023.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Process the full audio offline. Find every instance of a family relation and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.53, 4.87, "mother"]], "response_mode": "offline", "answer": "[{\"word\":\"mother\",\"start\":4.53,\"end\":4.87}]"}
{"utterance_id": "2277-149897-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2277/149897/2277-149897-0023.flac", "task_type": "lexical_detection", "query": "spent", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"spent\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.77, 2.09, "spent"]], "response_mode": "offline", "answer": "[{\"word\":\"spent\",\"start\":1.77,\"end\":2.09}]"}
{"utterance_id": "777-126732-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0001.flac", "task_type": "lexical_detection", "query": "prophetic", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"prophetic\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.04, 1.48, "prophetic"]], "response_mode": "offline", "answer": "[{\"word\":\"prophetic\",\"start\":1.04,\"end\":1.48}]"}
{"utterance_id": "2412-153947-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0003.flac", "task_type": "semantic_counting", "query": "cardinal", "prompt": "Listen to the full audio. Count every occurrence of a number. After it finishes, for each occurrence in order output a JSON object whose word is the count word (\"one\", \"two\", ...) with that occurrence's start/end times. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.38, 5.55, "one"], [8.45, 8.6, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":5.38,\"end\":5.55},{\"word\":\"two\",\"start\":8.45,\"end\":8.6}]"}
{"utterance_id": "6313-66125-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66125/6313-66125-0008.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Process the full audio offline. Find every instance of a person's name and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.33, 0.99, "master tad"]], "response_mode": "offline", "answer": "[{\"word\":\"master tad\",\"start\":0.33,\"end\":0.99}]"}
{"utterance_id": "6345-93306-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93306/6345-93306-0019.flac", "task_type": "semantic_detection", "query": "kinship", "prompt": "Listen carefully to the entire clip, then report every instance of a family relation as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.88, 1.28, "papa's"]], "response_mode": "offline", "answer": "[{\"word\":\"papa's\",\"start\":0.88,\"end\":1.28}]"}
{"utterance_id": "3170-137482-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0001.flac", "task_type": "lexical_detection", "query": "musical", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"musical\" with fields word, start, end (seconds). If none, reply [].", "labels": [[3.11, 3.58, "musical"]], "response_mode": "offline", "answer": "[{\"word\":\"musical\",\"start\":3.11,\"end\":3.58}]"}
{"utterance_id": "652-130726-0024", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0024.flac", "task_type": "lexical_detection", "query": "carrots", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"carrots\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.26, 4.76, "carrots"]], "response_mode": "offline", "answer": "[{\"word\":\"carrots\",\"start\":4.26,\"end\":4.76}]"}
{"utterance_id": "5895-34622-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0004.flac", "task_type": "lexical_counting", "query": "wheels", "prompt": "Listen to the entire clip, then report the count sequence for the word \"wheels\" as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.52, 0.87, "one"], [3.5, 3.96, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.52,\"end\":0.87},{\"word\":\"two\",\"start\":3.5,\"end\":3.96}]"}
{"utterance_id": "6295-64301-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0021.flac", "task_type": "lexical_detection", "query": "worship", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"worship\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[4.56, 5.11, "worship"]], "response_mode": "offline", "answer": "[{\"word\":\"worship\",\"start\":4.56,\"end\":5.11}]"}
{"utterance_id": "1673-143396-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0005.flac", "task_type": "lexical_detection", "query": "virgin", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"virgin\" with fields word, start, end (seconds). If none, reply [].", "labels": [[0.99, 1.53, "virgin"]], "response_mode": "offline", "answer": "[{\"word\":\"virgin\",\"start\":0.99,\"end\":1.53}]"}
{"utterance_id": "2803-161169-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0008.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the entire clip, then report the count sequence for a food name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.14, "one"], [9.6, 10.06, "two"], [12.23, 12.94, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":3.75,\"end\":4.14},{\"word\":\"two\",\"start\":9.6,\"end\":10.06},{\"word\":\"three\",\"start\":12.23,\"end\":12.94}]"}
{"utterance_id": "2035-152373-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/152373/2035-152373-0000.flac", "task_type": "lexical_detection", "query": "important", "prompt": "After listening to the whole recording, output every instance of the word \"important\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[6.13, 6.65, "important"]], "response_mode": "offline", "answer": "[{\"word\":\"important\",\"start\":6.13,\"end\":6.65}]"}
{"utterance_id": "3853-163249-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0007.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "Do not answer until the audio ends. Then list each instance of an ordinal number as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.43, 3.83, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":3.43,\"end\":3.83}]"}
{"utterance_id": "422-122949-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/422/122949/422-122949-0016.flac", "task_type": "lexical_detection", "query": "philosophers", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"philosophers\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.35, 2.11, "philosophers"]], "response_mode": "offline", "answer": "[{\"word\":\"philosophers\",\"start\":1.35,\"end\":2.11}]"}
{"utterance_id": "1462-170138-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170138/1462-170138-0017.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a person's name with fields word, start, end (seconds). If none, reply [].", "labels": [[0.47, 1.49, "sir harry towne"], [3.32, 4.29, "mister alexander"]], "response_mode": "offline", "answer": "[{\"word\":\"sir harry towne\",\"start\":0.47,\"end\":1.49},{\"word\":\"mister alexander\",\"start\":3.32,\"end\":4.29}]"}
{"utterance_id": "2412-153947-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153947/2412-153947-0013.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the full audio. After it finishes, find every instance of a number. For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.71, 2.99, "three"]], "response_mode": "offline", "answer": "[{\"word\":\"three\",\"start\":2.71,\"end\":2.99}]"}
{"utterance_id": "5895-34622-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34622/5895-34622-0008.flac", "task_type": "lexical_detection", "query": "theatre", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"theatre\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.95, 8.53, "theatre"]], "response_mode": "offline", "answer": "[{\"word\":\"theatre\",\"start\":7.95,\"end\":8.53}]"}
{"utterance_id": "3752-4944-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0020.flac", "task_type": "lexical_counting", "query": "reverend", "prompt": "You will hear a complete utterance. When it ends, produce a running count of the word \"reverend\": each event is {\"word\": \"one\"|\"two\"|..., \"start\", \"end\"}. If none, reply [].", "labels": [[0.33, 0.7, "one"], [3.58, 3.98, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.33,\"end\":0.7},{\"word\":\"two\",\"start\":3.58,\"end\":3.98}]"}
{"utterance_id": "8842-302203-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0000.flac", "task_type": "lexical_detection", "query": "unknown", "prompt": "After the audio ends, extract all occurrences of the word \"unknown\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.63, 8.9, "unknown"]], "response_mode": "offline", "answer": "[{\"word\":\"unknown\",\"start\":8.63,\"end\":8.9}]"}
{"utterance_id": "652-130726-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/130726/652-130726-0022.flac", "task_type": "semantic_detection", "query": "city", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a city name with fields word, start, end (seconds). If none, reply [].", "labels": [[8.06, 8.62, "heidelberg"]], "response_mode": "offline", "answer": "[{\"word\":\"heidelberg\",\"start\":8.06,\"end\":8.62}]"}
{"utterance_id": "3536-8226-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/8226/3536-8226-0009.flac", "task_type": "lexical_detection", "query": "can't", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"can't\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.77, "can't"]], "response_mode": "offline", "answer": "[{\"word\":\"can't\",\"start\":0.49,\"end\":0.77}]"}
{"utterance_id": "251-137823-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0006.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.49, 0.84, "bud"]], "response_mode": "offline", "answer": "[{\"word\":\"bud\",\"start\":0.49,\"end\":0.84}]"}
{"utterance_id": "6295-64301-0030", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0030.flac", "task_type": "lexical_detection", "query": "violin", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"violin\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.89, "violin"]], "response_mode": "offline", "answer": "[{\"word\":\"violin\",\"start\":0.5,\"end\":0.89}]"}
{"utterance_id": "2035-147961-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147961/2035-147961-0008.flac", "task_type": "lexical_detection", "query": "docile", "prompt": "Process the full audio offline. Find every instance of the word \"docile\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.11, 3.74, "docile"]], "response_mode": "offline", "answer": "[{\"word\":\"docile\",\"start\":3.11,\"end\":3.74}]"}
{"utterance_id": "1993-147149-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147149/1993-147149-0015.flac", "task_type": "semantic_detection", "query": "date", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a date with fields word, start, end (seconds). If none, reply [].", "labels": [[0.7, 1.15, "christmas"]], "response_mode": "offline", "answer": "[{\"word\":\"christmas\",\"start\":0.7,\"end\":1.15}]"}
{"utterance_id": "6345-93302-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6345/93302/6345-93302-0015.flac", "task_type": "lexical_detection", "query": "shock", "prompt": "Once the audio has finished, identify every instance of the word \"shock\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.48, 0.79, "shock"]], "response_mode": "offline", "answer": "[{\"word\":\"shock\",\"start\":0.48,\"end\":0.79}]"}
{"utterance_id": "2078-142845-0042", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0042.flac", "task_type": "semantic_detection", "query": "food", "prompt": "You will hear a complete utterance. When it ends, list every instance of a food name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.28, 1.95, "buns"]], "response_mode": "offline", "answer": "[{\"word\":\"buns\",\"start\":1.28,\"end\":1.95}]"}
{"utterance_id": "84-121123-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/84/121123/84-121123-0025.flac", "task_type": "semantic_detection", "query": "country", "prompt": "Listen carefully to the entire clip, then report every instance of a country name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.29, 2.78, "italian"]], "response_mode": "offline", "answer": "[{\"word\":\"italian\",\"start\":2.29,\"end\":2.78}]"}
{"utterance_id": "3752-4944-0067", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4944/3752-4944-0067.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.0, 2.55, "spirit"]], "response_mode": "offline", "answer": "[{\"word\":\"spirit\",\"start\":2.0,\"end\":2.55}]"}
{"utterance_id": "174-50561-0009", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0009.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[1.09, 1.67, "singer"]], "response_mode": "offline", "answer": "[{\"word\":\"singer\",\"start\":1.09,\"end\":1.67}]"}
{"utterance_id": "174-50561-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/50561/174-50561-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[7.79, 8.03, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":7.79,\"end\":8.03}]"}
{"utterance_id": "8297-275154-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0006.flac", "task_type": "lexical_detection", "query": "accept", "prompt": "After listening to the whole recording, output every instance of the word \"accept\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.4, 1.86, "accept"]], "response_mode": "offline", "answer": "[{\"word\":\"accept\",\"start\":1.4,\"end\":1.86}]"}
{"utterance_id": "3752-4943-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3752/4943/3752-4943-0011.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a profession with fields word, start, end (seconds). If none, reply [].", "labels": [[0.71, 1.32, "ministers"]], "response_mode": "offline", "answer": "[{\"word\":\"ministers\",\"start\":0.71,\"end\":1.32}]"}
{"utterance_id": "7976-110523-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7976/110523/7976-110523-0005.flac", "task_type": "lexical_detection", "query": "cat", "prompt": "Do not answer until the audio ends. Then list each instance of the word \"cat\" as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.76, 4.2, "cat"]], "response_mode": "offline", "answer": "[{\"word\":\"cat\",\"start\":3.76,\"end\":4.2}]"}
{"utterance_id": "1988-147956-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0028.flac", "task_type": "lexical_detection", "query": "touched", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"touched\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.32, 1.63, "touched"]], "response_mode": "offline", "answer": "[{\"word\":\"touched\",\"start\":1.32,\"end\":1.63}]"}
{"utterance_id": "6313-66129-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6313/66129/6313-66129-0017.flac", "task_type": "semantic_detection", "query": "animal", "prompt": "Once the audio has finished, identify every instance of an animal name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.74, 1.22, "animals"]], "response_mode": "offline", "answer": "[{\"word\":\"animals\",\"start\":0.74,\"end\":1.22}]"}
{"utterance_id": "2086-149220-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0001.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[13.82, 14.26, "fruit"]], "response_mode": "offline", "answer": "[{\"word\":\"fruit\",\"start\":13.82,\"end\":14.26}]"}
{"utterance_id": "2803-161169-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0001.flac", "task_type": "lexical_detection", "query": "thousands", "prompt": "Process the full audio offline. Find every instance of the word \"thousands\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.33, 4.96, "thousands"]], "response_mode": "offline", "answer": "[{\"word\":\"thousands\",\"start\":4.33,\"end\":4.96}]"}
{"utterance_id": "1988-147956-0022", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/147956/1988-147956-0022.flac", "task_type": "lexical_detection", "query": "pointed", "prompt": "Once the audio has finished, identify every instance of the word \"pointed\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.6, 1.04, "pointed"]], "response_mode": "offline", "answer": "[{\"word\":\"pointed\",\"start\":0.6,\"end\":1.04}]"}
{"utterance_id": "3853-163249-0051", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0051.flac", "task_type": "lexical_detection", "query": "folks", "prompt": "Process the full audio offline. Find every instance of the word \"folks\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.71, 1.07, "folks"]], "response_mode": "offline", "answer": "[{\"word\":\"folks\",\"start\":0.71,\"end\":1.07}]"}
{"utterance_id": "2078-142845-0046", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2078/142845/2078-142845-0046.flac", "task_type": "lexical_detection", "query": "principally", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"principally\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.96, 3.58, "principally"]], "response_mode": "offline", "answer": "[{\"word\":\"principally\",\"start\":2.96,\"end\":3.58}]"}
{"utterance_id": "7850-286674-0015", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/286674/7850-286674-0015.flac", "task_type": "semantic_detection", "query": "food", "prompt": "Listen carefully to the entire clip, then report every instance of a food name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.78, 3.21, "dinner"]], "response_mode": "offline", "answer": "[{\"word\":\"dinner\",\"start\":2.78,\"end\":3.21}]"}
{"utterance_id": "6295-244435-0019", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0019.flac", "task_type": "lexical_detection", "query": "regret", "prompt": "When the utterance is complete, report all instances of the word \"regret\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.26, 3.88, "regret"]], "response_mode": "offline", "answer": "[{\"word\":\"regret\",\"start\":3.26,\"end\":3.88}]"}
{"utterance_id": "777-126732-0074", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/777/126732/777-126732-0074.flac", "task_type": "lexical_detection", "query": "gets", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of the word \"gets\" with fields word, start, end (seconds). If none, reply [].", "labels": [[1.53, 1.72, "gets"]], "response_mode": "offline", "answer": "[{\"word\":\"gets\",\"start\":1.53,\"end\":1.72}]"}
{"utterance_id": "1673-143396-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0012.flac", "task_type": "lexical_detection", "query": "selected", "prompt": "Once the audio has finished, identify every instance of the word \"selected\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[8.77, 9.31, "selected"]], "response_mode": "offline", "answer": "[{\"word\":\"selected\",\"start\":8.77,\"end\":9.31}]"}
{"utterance_id": "1272-135031-0023", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/135031/1272-135031-0023.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Listen carefully to the entire clip, then report every instance of a person's name as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.5, 0.97, "kaliko"], [3.7, 4.09, "ruggedo"]], "response_mode": "offline", "answer": "[{\"word\":\"kaliko\",\"start\":0.5,\"end\":0.97},{\"word\":\"ruggedo\",\"start\":3.7,\"end\":4.09}]"}
{"utterance_id": "2803-154328-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0006.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.14, 1.72, "voice"]], "response_mode": "offline", "answer": "[{\"word\":\"voice\",\"start\":1.14,\"end\":1.72}]"}
{"utterance_id": "5338-284437-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0033.flac", "task_type": "lexical_detection", "query": "i'll", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"i'll\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.3, 0.59, "i'll"]], "response_mode": "offline", "answer": "[{\"word\":\"i'll\",\"start\":0.3,\"end\":0.59}]"}
{"utterance_id": "2035-147960-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0003.flac", "task_type": "semantic_detection", "query": "food", "prompt": "After the audio ends, extract all occurrences of a food name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.8, 5.33, "wine"]], "response_mode": "offline", "answer": "[{\"word\":\"wine\",\"start\":4.8,\"end\":5.33}]"}
{"utterance_id": "7850-111771-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/7850/111771/7850-111771-0002.flac", "task_type": "semantic_detection", "query": "state_or_province", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a state or province name with fields word, start, end (seconds). If none, reply [].", "labels": [[5.86, 6.29, "illinois"]], "response_mode": "offline", "answer": "[{\"word\":\"illinois\",\"start\":5.86,\"end\":6.29}]"}
{"utterance_id": "3000-15664-0044", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3000/15664/3000-15664-0044.flac", "task_type": "lexical_detection", "query": "spacious", "prompt": "After the audio ends, extract all occurrences of the word \"spacious\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.24, 5.85, "spacious"]], "response_mode": "offline", "answer": "[{\"word\":\"spacious\",\"start\":5.24,\"end\":5.85}]"}
{"utterance_id": "2086-149220-0033", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0033.flac", "task_type": "lexical_detection", "query": "turning", "prompt": "Once the audio has finished, identify every instance of the word \"turning\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[3.32, 3.65, "turning"]], "response_mode": "offline", "answer": "[{\"word\":\"turning\",\"start\":3.32,\"end\":3.65}]"}
{"utterance_id": "1272-141231-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1272/141231/1272-141231-0021.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.64, 2.38, "twenties"]], "response_mode": "offline", "answer": "[{\"word\":\"twenties\",\"start\":1.64,\"end\":2.38}]"}
{"utterance_id": "3576-138058-0011", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0011.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "You will hear a complete utterance. When it ends, list every instance of a person's name you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[9.13, 9.8, "don quixote"]], "response_mode": "offline", "answer": "[{\"word\":\"don quixote\",\"start\":9.13,\"end\":9.8}]"}
{"utterance_id": "5895-34615-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34615/5895-34615-0012.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.21, 5.6, "god"]], "response_mode": "offline", "answer": "[{\"word\":\"god\",\"start\":5.21,\"end\":5.6}]"}
{"utterance_id": "8297-275155-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275155/8297-275155-0016.flac", "task_type": "lexical_detection", "query": "charmed", "prompt": "Process the full audio offline. Find every instance of the word \"charmed\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[7.03, 7.6, "charmed"]], "response_mode": "offline", "answer": "[{\"word\":\"charmed\",\"start\":7.03,\"end\":7.6}]"}
{"utterance_id": "251-137823-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/251/137823/251-137823-0020.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "After the audio ends, extract all occurrences of a person's name. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 2.84, "tom's"], [4.28, 4.6, "tom"]], "response_mode": "offline", "answer": "[{\"word\":\"tom's\",\"start\":2.46,\"end\":2.84},{\"word\":\"tom\",\"start\":4.28,\"end\":4.6}]"}
{"utterance_id": "1993-147964-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1993/147964/1993-147964-0001.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After listening to the whole recording, output every instance of a number with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.92, 2.06, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.92,\"end\":2.06}]"}
{"utterance_id": "2428-83705-0031", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83705/2428-83705-0031.flac", "task_type": "lexical_detection", "query": "nature", "prompt": "After the audio ends, extract all occurrences of the word \"nature\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.89, 2.37, "nature"]], "response_mode": "offline", "answer": "[{\"word\":\"nature\",\"start\":1.89,\"end\":2.37}]"}
{"utterance_id": "8842-302203-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0010.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Process the full audio offline. Find every instance of a number and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[12.18, 12.59, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":12.18,\"end\":12.59}]"}
{"utterance_id": "174-168635-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/174/168635/174-168635-0001.flac", "task_type": "lexical_detection", "query": "convict", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"convict\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.85, 2.53, "convict"]], "response_mode": "offline", "answer": "[{\"word\":\"convict\",\"start\":1.85,\"end\":2.53}]"}
{"utterance_id": "3853-163249-0016", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3853/163249/3853-163249-0016.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.25, 0.57, "david"]], "response_mode": "offline", "answer": "[{\"word\":\"david\",\"start\":0.25,\"end\":0.57}]"}
{"utterance_id": "2412-153954-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0003.flac", "task_type": "lexical_detection", "query": "painted", "prompt": "Once the audio has finished, identify every instance of the word \"painted\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.79, 5.11, "painted"]], "response_mode": "offline", "answer": "[{\"word\":\"painted\",\"start\":4.79,\"end\":5.11}]"}
{"utterance_id": "3170-137482-0017", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3170/137482/3170-137482-0017.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "Listen to the audio in full, then emit a JSON array of every instance of a number with fields word, start, end (seconds). If none, reply [].", "labels": [[1.44, 1.61, "two"], [8.09, 8.69, "six"], [10.69, 10.96, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"two\",\"start\":1.44,\"end\":1.61},{\"word\":\"six\",\"start\":8.09,\"end\":8.69},{\"word\":\"two\",\"start\":10.69,\"end\":10.96}]"}
{"utterance_id": "8297-275156-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275156/8297-275156-0010.flac", "task_type": "lexical_detection", "query": "happier", "prompt": "After the audio ends, extract all occurrences of the word \"happier\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.97, 1.6, "happier"]], "response_mode": "offline", "answer": "[{\"word\":\"happier\",\"start\":0.97,\"end\":1.6}]"}
{"utterance_id": "6295-64301-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/64301/6295-64301-0003.flac", "task_type": "semantic_detection", "query": "body_part", "prompt": "Process the full audio offline. Find every instance of a body part and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.19, 2.66, "heart"]], "response_mode": "offline", "answer": "[{\"word\":\"heart\",\"start\":2.19,\"end\":2.66}]"}
{"utterance_id": "2902-9006-0001", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2902/9006/2902-9006-0001.flac", "task_type": "semantic_detection", "query": "profession", "prompt": "Do not answer until the audio ends. Then list each instance of a profession as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[15.19, 15.96, "apologist"]], "response_mode": "offline", "answer": "[{\"word\":\"apologist\",\"start\":15.19,\"end\":15.96}]"}
{"utterance_id": "2035-147960-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2035/147960/2035-147960-0010.flac", "task_type": "semantic_detection", "query": "ordinal", "prompt": "When the utterance is complete, report all instances of an ordinal number including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.9, 6.2, "first"]], "response_mode": "offline", "answer": "[{\"word\":\"first\",\"start\":5.9,\"end\":6.2}]"}
{"utterance_id": "6319-57405-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0004.flac", "task_type": "semantic_detection", "query": "weather", "prompt": "Once the audio has finished, identify every instance of a weather term. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[4.77, 5.3, "storm"]], "response_mode": "offline", "answer": "[{\"word\":\"storm\",\"start\":4.77,\"end\":5.3}]"}
{"utterance_id": "652-129742-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/652/129742/652-129742-0012.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the entire clip, then report the count sequence for a food name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.51, 1.11, "one"], [2.87, 3.34, "two"], [5.4, 5.89, "three"], [8.39, 8.99, "four"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.51,\"end\":1.11},{\"word\":\"two\",\"start\":2.87,\"end\":3.34},{\"word\":\"three\",\"start\":5.4,\"end\":5.89},{\"word\":\"four\",\"start\":8.39,\"end\":8.99}]"}
{"utterance_id": "1673-143396-0013", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1673/143396/1673-143396-0013.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "Once the audio has finished, identify every instance of a person's name. Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.73, 1.25, "messiah"], [3.32, 3.87, "christ"], [11.97, 12.34, "jesus"]], "response_mode": "offline", "answer": "[{\"word\":\"messiah\",\"start\":0.73,\"end\":1.25},{\"word\":\"christ\",\"start\":3.32,\"end\":3.87},{\"word\":\"jesus\",\"start\":11.97,\"end\":12.34}]"}
{"utterance_id": "8297-275154-0025", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8297/275154/8297-275154-0025.flac", "task_type": "lexical_detection", "query": "done", "prompt": "When the utterance is complete, report all instances of the word \"done\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[4.03, 4.37, "done"]], "response_mode": "offline", "answer": "[{\"word\":\"done\",\"start\":4.03,\"end\":4.37}]"}
{"utterance_id": "2086-149220-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2086/149220/2086-149220-0003.flac", "task_type": "lexical_detection", "query": "tendency", "prompt": "Once the audio has finished, identify every instance of the word \"tendency\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[5.11, 5.64, "tendency"]], "response_mode": "offline", "answer": "[{\"word\":\"tendency\",\"start\":5.11,\"end\":5.64}]"}
{"utterance_id": "3576-138058-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3576/138058/3576-138058-0007.flac", "task_type": "lexical_detection", "query": "dost", "prompt": "Once the audio has finished, identify every instance of the word \"dost\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[0.21, 0.5, "dost"]], "response_mode": "offline", "answer": "[{\"word\":\"dost\",\"start\":0.21,\"end\":0.5}]"}
{"utterance_id": "3536-23268-0021", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/3536/23268/3536-23268-0021.flac", "task_type": "semantic_counting", "query": "food", "prompt": "Listen to the entire clip, then report the count sequence for a food name as timed JSON events (word is the count word, not the surface form). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.64, 1.98, "one"], [13.16, 13.55, "two"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":1.64,\"end\":1.98},{\"word\":\"two\",\"start\":13.16,\"end\":13.55}]"}
{"utterance_id": "5338-284437-0000", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/284437/5338-284437-0000.flac", "task_type": "lexical_detection", "query": "appearance", "prompt": "Process the full audio offline. Find every instance of the word \"appearance\" and give the surface word plus start/end timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[3.75, 4.39, "appearance"]], "response_mode": "offline", "answer": "[{\"word\":\"appearance\",\"start\":3.75,\"end\":4.39}]"}
{"utterance_id": "6295-244435-0010", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6295/244435/6295-244435-0010.flac", "task_type": "lexical_detection", "query": "president", "prompt": "After the audio ends, extract all occurrences of the word \"president\". Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.27, 1.78, "president"]], "response_mode": "offline", "answer": "[{\"word\":\"president\",\"start\":1.27,\"end\":1.78}]"}
{"utterance_id": "2428-83699-0028", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2428/83699/2428-83699-0028.flac", "task_type": "lexical_detection", "query": "knocker", "prompt": "After listening to the whole recording, output every instance of the word \"knocker\" with precise start and end times (seconds). Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.59, 1.32, "knocker"]], "response_mode": "offline", "answer": "[{\"word\":\"knocker\",\"start\":0.59,\"end\":1.32}]"}
{"utterance_id": "8842-302203-0008", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/8842/302203/8842-302203-0008.flac", "task_type": "lexical_detection", "query": "begins", "prompt": "Listen to the full audio. After it finishes, find every instance of the word \"begins\". For each match, report the spoken word and its start/end times in seconds. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.81, 1.15, "begins"]], "response_mode": "offline", "answer": "[{\"word\":\"begins\",\"start\":0.81,\"end\":1.15}]"}
{"utterance_id": "6319-57405-0005", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6319/57405/6319-57405-0005.flac", "task_type": "semantic_detection", "query": "person_name", "prompt": "When the utterance is complete, report all instances of a person's name including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.29, "prometheus"], [4.3, 4.89, "jupiter"]], "response_mode": "offline", "answer": "[{\"word\":\"prometheus\",\"start\":2.46,\"end\":3.29},{\"word\":\"jupiter\",\"start\":4.3,\"end\":4.89}]"}
{"utterance_id": "1988-148538-0012", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1988/148538/1988-148538-0012.flac", "task_type": "lexical_detection", "query": "superiority", "prompt": "Listen carefully to the entire clip, then report every instance of the word \"superiority\" as JSON objects with word, start, and end. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[2.46, 3.3, "superiority"]], "response_mode": "offline", "answer": "[{\"word\":\"superiority\",\"start\":2.46,\"end\":3.3}]"}
{"utterance_id": "6241-66616-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/6241/66616/6241-66616-0002.flac", "task_type": "lexical_detection", "query": "main", "prompt": "Once the audio has finished, identify every instance of the word \"main\". Return a JSON list of {\"word\", \"start\", \"end\"} in time order. If none, reply [].", "labels": [[2.79, 3.05, "main"]], "response_mode": "offline", "answer": "[{\"word\":\"main\",\"start\":2.79,\"end\":3.05}]"}
{"utterance_id": "2412-153954-0007", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2412/153954/2412-153954-0007.flac", "task_type": "semantic_detection", "query": "color", "prompt": "Do not answer until the audio ends. Then list each instance of a color name as {\"word\", \"start\", \"end\"}. If none, reply [].", "labels": [[3.26, 3.44, "grey"]], "response_mode": "offline", "answer": "[{\"word\":\"grey\",\"start\":3.26,\"end\":3.44}]"}
{"utterance_id": "1462-170145-0002", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/1462/170145/1462-170145-0002.flac", "task_type": "lexical_detection", "query": "crossed", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"crossed\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.19, 1.54, "crossed"]], "response_mode": "offline", "answer": "[{\"word\":\"crossed\",\"start\":1.19,\"end\":1.54}]"}
{"utterance_id": "2803-154328-0020", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/154328/2803-154328-0020.flac", "task_type": "lexical_detection", "query": "hut", "prompt": "When the utterance is complete, report all instances of the word \"hut\" including the spoken word and time span. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[5.7, 6.1, "hut"]], "response_mode": "offline", "answer": "[{\"word\":\"hut\",\"start\":5.7,\"end\":6.1}]"}
{"utterance_id": "5895-34629-0004", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5895/34629/5895-34629-0004.flac", "task_type": "lexical_detection", "query": "box", "prompt": "You will hear a complete utterance. When it ends, list every instance of the word \"box\" you heard with word-level timestamps. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[1.91, 2.37, "box"]], "response_mode": "offline", "answer": "[{\"word\":\"box\",\"start\":1.91,\"end\":2.37}]"}
{"utterance_id": "5338-24615-0006", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/5338/24615/5338-24615-0006.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[0.69, 0.86, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":0.69,\"end\":0.86}]"}
{"utterance_id": "2803-161169-0003", "split": "dev-clean", "audio_path": "LibriSpeech/dev-clean/2803/161169/2803-161169-0003.flac", "task_type": "semantic_detection", "query": "cardinal", "prompt": "After the audio ends, extract all occurrences of a number. Output each spoken word with its start and end time. Reply with JSON only: a list of objects {\"word\": <string>, \"start\": <seconds>, \"end\": <seconds>} in chronological order. If none, reply [].", "labels": [[8.32, 8.81, "one"]], "response_mode": "offline", "answer": "[{\"word\":\"one\",\"start\":8.32,\"end\":8.81}]"}