Datasets:
File size: 1,735 Bytes
5a0a123 5045ae6 5a0a123 5045ae6 5a0a123 5045ae6 5a0a123 5045ae6 5a0a123 5045ae6 76101e0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 | {
"source_recordings": 80,
"source_hours": 7.695315520833334,
"all_chunks": 3559,
"accepted": 2603,
"rejected": 956,
"splits": {
"train": {
"samples": 1980,
"hours": 3.875754166666667,
"sources": 62,
"speakers": [
"Speaker_1",
"Speaker_2",
"Speaker_3",
"Speaker_4"
]
},
"validation": {
"samples": 294,
"hours": 0.5720298611111111,
"sources": 9,
"speakers": [
"Speaker_1",
"Speaker_3",
"Speaker_4"
]
},
"test": {
"samples": 329,
"hours": 0.6439156944444445,
"sources": 9,
"speakers": [
"Speaker_1",
"Speaker_3",
"Speaker_4"
]
}
},
"seed": 42,
"split_policy": "source recording + identical source transcript/audio groups; shared speakers allowed; approximately 80/10/10 by group count within speaker strata; one-recording speaker remains train",
"speaker_independent": false,
"alignment_quality_filter": "normalized greedy CTC CER <= 0.45; nonempty supported text; 1–16 seconds; nonzero PCM",
"alignment_checkpoint": "/home/ali/ali/ASR/wav2vec2_base/output_ver_1/checkpoint-547500",
"leakage_audit": "disjoint source IDs, source audio hashes, PCM hashes, and duplicate-transcript groups across all splits",
"human_verification": {
"scope": "source audio and transcript review reported by dataset owner",
"reviewed_speaker_ids": [
"Speaker_1"
],
"reviewed_source_recordings": 68,
"total_source_recordings": 80,
"source_recording_fraction": 0.85,
"accepted_chunks_with_reviewed_source": 2368,
"accepted_chunks_total": 2603,
"independent_chunk_boundary_review": false
}
}
|