Catalogue snapshot 2026-11-09

data/benchmarks.jsonl

@@ -0,0 +1,54 @@
+{"slug":"turnbench","name":"TurnBench","maintainer":"Sesame AI Labs / Mundo AI / CMU / oto","tagline":"Scores end-of-turn and interruption detection on 30 hours of hand-labelled dyadic conversation, with conversation type as a controlled variable.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["end-of-turn recall","interruption recall","false-positive rate","median detection latency"],"models_scored":1,"updated":"2026-10","url":"https://turnbench.sesame.com/","paper_url":"https://arxiv.org/abs/2608.25218","leaderboard_url":"https://turnbench.sesame.com/","page_url":"https://www.fullduplex.ai/benchmarks/turnbench"}
+{"slug":"fdb-v1","name":"Full-Duplex-Bench v1","maintainer":"NTU / UC Berkeley / UW / MIT","tagline":"Original Full-Duplex-Bench — automatic scoring of the four canonical turn-taking events: when-to-speak, back-channel, interruption success, pause handling.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["turn-taking latency","back-channel timing","interruption success","pause handling"],"models_scored":4,"updated":"2025-03","url":"https://github.com/DanielLin94144/Full-Duplex-Bench","paper_url":"https://arxiv.org/abs/2503.04721","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fdb-v1"}
+{"slug":"fdb-v15","name":"Full-Duplex-Bench v1.5","maintainer":"NTU / UC Berkeley / UW / MIT","tagline":"Mid-2025 refresh of FDB v1 — adds the simultaneous-speech / overlap axis (user interruption, listener back-channel, side conversation, ambient speech).","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["overlap duration","mutual-silence rate","interruption recovery"],"models_scored":3,"updated":"2025-08","url":"https://github.com/DanielLin94144/Full-Duplex-Bench","paper_url":"https://arxiv.org/abs/2507.23159","leaderboard_url":"https://artificialanalysis.ai/speech-to-speech","page_url":"https://www.fullduplex.ai/benchmarks/fdb-v15"}
+{"slug":"fdb-v2","name":"Full-Duplex-Bench v2","maintainer":"NTU / UC Berkeley / UW / MIT","tagline":"Late-2025 WebRTC-based live harness — first FDB release that runs dynamic evaluation against streaming endpoints, plus multi-turn stimuli.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex","speech-lm"],"metrics":["streaming latency","multi-turn coherence","overlap handling"],"models_scored":0,"updated":"2026-02","url":"https://github.com/DanielLin94144/Full-Duplex-Bench","paper_url":"https://arxiv.org/abs/2510.07838","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fdb-v2"}
+{"slug":"fdb-v3","name":"Full-Duplex-Bench v3","maintainer":"NTU (NVIDIA advisory)","tagline":"Multi-step tool calling for voice agents, driven by real human speech with natural disfluencies.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex","vertical-agent"],"metrics":["Pass@1","tool-call F1","response latency","turn-take rate"],"models_scored":1,"updated":"2026-05","url":"https://github.com/DanielLin94144/Full-Duplex-Bench","paper_url":"https://arxiv.org/abs/2604.04847","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fdb-v3"}
+{"slug":"videofdb","name":"VideoFDB","maintainer":"NVIDIA / David AI","tagline":"Full-duplex benchmark for video calls: does the agent read the user's nonverbal cues, and do its own voice and face respond the way a person's would?","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex"],"metrics":["perception rubric (0 to 5)","generation rubric (0 to 5)","TOR-Alignment","median response latency"],"models_scored":4,"updated":"2026-10","url":"https://research.nvidia.com/labs/amri/projects/video-fdb/","paper_url":"https://arxiv.org/abs/2605.30256","leaderboard_url":"https://research.nvidia.com/labs/amri/projects/video-fdb/","page_url":"https://www.fullduplex.ai/benchmarks/videofdb"}
+{"slug":"duplexact","name":"DuplexAct","maintainer":"Dong et al.","tagline":"Scores whether full-duplex speech models take the right interaction action when a conversation calls for it, and whether they follow through.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["immediate execution success","follow-through success"],"models_scored":0,"updated":"2026-10","url":"https://github.com/JiahengDong/DuplexAct","paper_url":"https://arxiv.org/abs/2610.17738","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplexact"}
+{"slug":"multipartyturnbench","name":"MultiPartyTurnBench","maintainer":"Mathur and Manocha","tagline":"Floor control with two to five speakers: can a detector or full-duplex agent tell a real turn change from a hold, a backchannel, or speech that does not take the floor?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["end-of-turn recall at fixed FPR","interruption recall at fixed FPR","signed latency","selective interruption balanced accuracy"],"models_scored":0,"updated":"2026-10","url":"https://multipartyturnbench.github.io/","paper_url":"https://arxiv.org/abs/2610.20172","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/multipartyturnbench"}
+{"slug":"multitalkbench","name":"MultiTalkBench","maintainer":"Wang et al. (MultiTalk)","tagline":"Long, multi-party, English and Chinese full-duplex dialogue, with conversations that average 32.6 minutes.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["long-range entity tracking","topic coherence","addressee selection","turn-taking"],"models_scored":0,"updated":"2026-09","url":"https://huggingface.co/datasets/MultiTalk/MultiTalkBench","paper_url":"https://arxiv.org/abs/2609.36903","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/multitalkbench"}
+{"slug":"talking-turns","name":"Talking Turns","maintainer":"Apple / CMU","tagline":"A supervised turn-taking judge trained on human-human conversations to score audio foundation models.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["turn-event F1","backchannel rate","interruption rate"],"models_scored":3,"updated":"2025-04","url":"https://machinelearning.apple.com/research/talking-turns","paper_url":"https://arxiv.org/abs/2503.01174","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/talking-turns"}
+{"slug":"voicebench","name":"VoiceBench","maintainer":"NUS","tagline":"Evaluates LLM-based voice assistants across knowledge, reasoning, safety, and instruction-following — judged by a GPT-4-class model.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["overall","QA","reasoning (BBH)","IFEval","safety"],"models_scored":7,"updated":"2025-06","url":"https://github.com/MatthewCYM/VoiceBench","paper_url":"https://arxiv.org/abs/2410.17196","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voicebench"}
+{"slug":"vocalbench","name":"VocalBench","maintainer":"SJTU (OmniAgent)","tagline":"A four-dimensional stress test for speech-interaction models — semantic, acoustic, conversational, and robust.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding"],"metrics":["ACC","LLM-judge","UTMOS","WER","EER"],"models_scored":0,"updated":"2026-01","url":"https://github.com/SJTU-OmniAgent/VocalBench","paper_url":"https://arxiv.org/abs/2505.15727","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/vocalbench"}
+{"slug":"uro-bench","name":"URO-Bench","maintainer":"Ruiqi Yan et al.","tagline":"The first end-to-end S2S benchmark with multilingual, multi-turn, and paralinguistic coverage.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following","emotion"],"metrics":["U/R/O composite","paralinguistic score"],"models_scored":2,"updated":"2025-08","url":"https://github.com/Ruiqi-Yan/URO-Bench","paper_url":"https://arxiv.org/abs/2502.17810","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/uro-bench"}
+{"slug":"air-bench","name":"AIR-Bench","maintainer":"Alibaba DAMO (OFA-Sys)","tagline":"A two-tier stress test for audio-LLMs covering speech, natural sound, and music — MCQ plus chat.","full_duplex":false,"tier":"native","setting":"lab","focus":["audio-understanding","instruction-following"],"metrics":["MCQ acc","GPT-4 judge score"],"models_scored":3,"updated":"2024-12","url":"https://github.com/OFA-Sys/AIR-Bench","paper_url":"https://arxiv.org/abs/2402.07729","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/air-bench"}
+{"slug":"big-bench-audio","name":"Big Bench Audio","maintainer":"Artificial Analysis","tagline":"1 000 audio-question reasoning benchmark — the dataset behind the Artificial Analysis S2S speech-reasoning leaderboard.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["accuracy","TTFA (s)"],"models_scored":10,"updated":"2026-05","url":"https://huggingface.co/datasets/ArtificialAnalysis/big_bench_audio","paper_url":null,"leaderboard_url":"https://artificialanalysis.ai/speech-to-speech","page_url":"https://www.fullduplex.ai/benchmarks/big-bench-audio"}
+{"slug":"aa-speech-to-speech-index","name":"Artificial Analysis Speech to Speech Index","maintainer":"Artificial Analysis","tagline":"Composite score for native speech-to-speech models: speech reasoning, agentic task completion, human preference and task success, weighted equally.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","vertical-agent"],"metrics":["index score","Big Bench Audio accuracy","τ-Voice pass rate","Arena Score","task success rate"],"models_scored":0,"updated":"2026-08","url":"https://artificialanalysis.ai/methodology/speech-to-speech-benchmarking","paper_url":null,"leaderboard_url":"https://artificialanalysis.ai/speech-to-speech","page_url":"https://www.fullduplex.ai/benchmarks/aa-speech-to-speech-index"}
+{"slug":"mmar","name":"MMAR","maintainer":"SJTU / MBZUAI · NeurIPS 2025","tagline":"1 000 human-curated deep-reasoning audio QA items spanning speech, sound, music, and their mix — four hierarchical reasoning layers.","full_duplex":false,"tier":"component","setting":"lab","focus":["audio-understanding","speech-lm"],"metrics":["overall accuracy","per-layer accuracy","rubric score"],"models_scored":2,"updated":"2026-02","url":"https://github.com/ddlBoJack/MMAR","paper_url":"https://arxiv.org/abs/2505.13032","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mmar"}
+{"slug":"ifeval-audio","name":"IFEval-Audio","maintainer":"AudioLLMs (I2R Singapore)","tagline":"Format-constrained spoken instructions, verified programmatically — an audio port of text IFEval.","full_duplex":false,"tier":"native","setting":"lab","focus":["instruction-following","speech-lm"],"metrics":["IFR","SCR","OSR"],"models_scored":0,"updated":"2025-05","url":"https://github.com/AudioLLMs/AudioBench/tree/main/IFEval-Audio","paper_url":"https://arxiv.org/abs/2505.16774","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/ifeval-audio"}
+{"slug":"scale-voice-showdown","name":"Scale Voice Showdown","maintainer":"Scale AI","tagline":"First in-the-wild preference arena for voice AI, launched March 2026 with 11 frontier models, 60+ languages and real user speech. The public leaderboard is offline as of October 2026.","full_duplex":false,"tier":"native","setting":"arena","focus":["speech-lm","tts"],"metrics":["preference win-rate","per-language score"],"models_scored":1,"updated":"2026-03","url":"https://labs.scale.com/","paper_url":null,"leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/scale-voice-showdown"}
+{"slug":"aa-speech-arena","name":"Artificial Analysis Speech Arena","maintainer":"Artificial Analysis","tagline":"Blind user votes turn TTS models into a live Elo leaderboard — the de-facto preference number for TTS. Despite the 'Speech Arena' label, this is a TTS-only (text → speech) arena, not an STS ranking.","full_duplex":false,"tier":"adjacent","setting":"arena","focus":["tts"],"metrics":["Elo"],"models_scored":0,"updated":"2026-10","url":"https://artificialanalysis.ai/text-to-speech/arena","paper_url":null,"leaderboard_url":"https://artificialanalysis.ai/text-to-speech/leaderboard","page_url":"https://www.fullduplex.ai/benchmarks/aa-speech-arena"}
+{"slug":"aa-speech-agent-arena","name":"Artificial Analysis Speech Agent Arena","maintainer":"Artificial Analysis","tagline":"Human participants complete the same task with two hidden speech-to-speech systems over live calls, then vote for the conversation they preferred.","full_duplex":false,"tier":"native","setting":"arena","focus":["speech-lm","vertical-agent"],"metrics":["preference Elo","agentic Elo","non-agentic Elo","task success rate"],"models_scored":0,"updated":"2026-10","url":"https://artificialanalysis.ai/speech-to-speech/arena/overview","paper_url":null,"leaderboard_url":"https://artificialanalysis.ai/speech-to-speech/arena","page_url":"https://www.fullduplex.ai/benchmarks/aa-speech-agent-arena"}
+{"slug":"rw-voice-eq","name":"Real World Voice EQ Bench","maintainer":"Hume AI","tagline":"Human-rated leaderboards across TTS, speech-to-speech, speech understanding and ASR, built to test whether systems use what is in the audio and not only the words.","full_duplex":false,"tier":"component","setting":"arena","focus":["tts","speech-lm","asr","emotion"],"metrics":["rater score (1 to 5) per factor","WER (ASR board)"],"models_scored":0,"updated":"2026-07","url":"https://huggingface.co/spaces/HumeAI/rw-voice-eq","paper_url":"https://arxiv.org/abs/2607.14846","leaderboard_url":"https://huggingface.co/spaces/HumeAI/rw-voice-eq","page_url":"https://www.fullduplex.ai/benchmarks/rw-voice-eq"}
+{"slug":"mtalk-bench","name":"MTalk-Bench","maintainer":"FreedomIntelligence / CUHK-SZ","tagline":"First multi-turn S2S benchmark with both arena-style (pairwise) and rubric-based (absolute) protocols for dialogue evaluation.","full_duplex":false,"tier":"native","setting":"arena","focus":["speech-lm","instruction-following","emotion"],"metrics":["arena win-rate","rubric score","per-dimension score"],"models_scored":0,"updated":"2025-09","url":"https://github.com/FreedomIntelligence/MTalk-Bench","paper_url":"https://arxiv.org/abs/2508.18240","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mtalk-bench"}
+{"slug":"tts-arena","name":"TTS Arena","maintainer":"Hugging Face","tagline":"The original blind A/B voting space for TTS, feeding an Elo ladder on Hugging Face.","full_duplex":false,"tier":"adjacent","setting":"arena","focus":["tts"],"metrics":["Elo"],"models_scored":0,"updated":"2026-03","url":"https://huggingface.co/spaces/TTS-AGI/TTS-Arena","paper_url":null,"leaderboard_url":"https://huggingface.co/spaces/TTS-AGI/TTS-Arena","page_url":"https://www.fullduplex.ai/benchmarks/tts-arena"}
+{"slug":"tau3-bench","name":"τ-bench 2.0 (τ-Voice)","maintainer":"Sierra Research","tagline":"The first agent benchmark with a live, full-duplex voice track, now in six languages with version 2.0.","full_duplex":true,"tier":"native","setting":"vertical","focus":["vertical-agent","full-duplex","speech-lm"],"metrics":["pass^k","tool-use success","voice task success","per-language task completion"],"models_scored":1,"updated":"2026-10","url":"https://sierra.ai/blog/bench-advancing-agent-benchmarking-to-knowledge-and-voice","paper_url":"https://arxiv.org/abs/2603.13686","leaderboard_url":"https://www.taubench.com","page_url":"https://www.fullduplex.ai/benchmarks/tau3-bench"}
+{"slug":"voiceagenteval","name":"VoiceAgentEval","maintainer":"Meituan / Xbench / Agora","tagline":"A dual-dimensional benchmark for expert-level outbound-calling agents — interaction fluency vs task flow compliance.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm"],"metrics":["GIC","TFC","task success"],"models_scored":0,"updated":"2025-10","url":"https://www.shengwang.cn/news/blogdetail/VoiceAgentEval/","paper_url":"https://arxiv.org/abs/2510.21244","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceagenteval"}
+{"slug":"healthbench","name":"HealthBench","maintainer":"OpenAI","tagline":"5,000 multi-turn health conversations × 48k physician-written rubric criteria — text-only, listed as a transferable foundation rubric rather than a voice benchmark.","full_duplex":false,"tier":"adjacent","setting":"vertical","focus":["vertical-agent","instruction-following"],"metrics":["rubric score","per-theme breakdown"],"models_scored":0,"updated":"2025-05","url":"https://huggingface.co/datasets/openai/healthbench","paper_url":"https://arxiv.org/abs/2505.08775","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/healthbench"}
+{"slug":"seamless-expressive","name":"SeamlessExpressive (mExpresso)","maintainer":"Meta AI","tagline":"Expressive speech-to-speech translation judged on translation, prosody, and vocal-style preservation.","full_duplex":false,"tier":"component","setting":"lab","focus":["s2st"],"metrics":["ASR-BLEU","AutoPCP","Vocal Style Sim"],"models_scored":0,"updated":"2023-12","url":"https://github.com/facebookresearch/seamless_communication","paper_url":"https://arxiv.org/abs/2312.05187","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/seamless-expressive"}
+{"slug":"covost2","name":"CoVoST-2 BLEU","maintainer":"Meta AI","tagline":"21 language pairs of speech-to-text translation, long the S2T standard.","full_duplex":false,"tier":"component","setting":"lab","focus":["s2st","asr"],"metrics":["BLEU","chrF"],"models_scored":0,"updated":"2020-07","url":"https://github.com/facebookresearch/covost","paper_url":"https://arxiv.org/abs/2007.10310","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/covost2"}
+{"slug":"seed-tts-eval","name":"Seed-TTS Eval","maintainer":"ByteDance","tagline":"The de-facto evaluation set for zero-shot voice cloning — WER, speaker similarity, and CMOS in one recipe.","full_duplex":false,"tier":"component","setting":"lab","focus":["tts"],"metrics":["WER","SECS","CMOS"],"models_scored":0,"updated":"2024-11","url":"https://bytedancespeech.github.io/seedtts_tech_report/","paper_url":"https://arxiv.org/abs/2406.02430","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/seed-tts-eval"}
+{"slug":"nv-bench","name":"NV-Bench","maintainer":"NVSpeech authors","tagline":"Standardised test for generating laughs, sighs, fillers, and back-channels in TTS.","full_duplex":false,"tier":"component","setting":"lab","focus":["tts","emotion"],"metrics":["PCER","FAD","SIM","DNSMOS"],"models_scored":0,"updated":"2025-08","url":"https://nvspeech.github.io","paper_url":"https://arxiv.org/abs/2508.04195","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/nv-bench"}
+{"slug":"superb","name":"SUPERB","maintainer":"NTU et al. (s3prl)","tagline":"The classic: general-purpose speech representations scored on ten frozen-encoder downstream tasks.","full_duplex":false,"tier":"peripheral","setting":"lab","focus":["speech-lm","asr","audio-understanding"],"metrics":["composite score","per-task scores"],"models_scored":0,"updated":"2024-06","url":"https://github.com/s3prl/s3prl","paper_url":"https://arxiv.org/abs/2105.01051","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/superb"}
+{"slug":"librispeech","name":"LibriSpeech WER","maintainer":"OpenSLR","tagline":"The coordinate system of English ASR — test-clean and test-other WER.","full_duplex":false,"tier":"peripheral","setting":"lab","focus":["asr"],"metrics":["WER (test-clean)","WER (test-other)"],"models_scored":0,"updated":"2024-01","url":"https://www.openslr.org/12","paper_url":null,"leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/librispeech"}
+{"slug":"meld","name":"MELD / IEMOCAP","maintainer":"Academic","tagline":"The classic emotion-recognition pair, evaluated with full dialogue context.","full_duplex":false,"tier":"peripheral","setting":"lab","focus":["emotion","audio-understanding"],"metrics":["Weighted F1","Accuracy"],"models_scored":0,"updated":"2023-06","url":"https://affective-meld.github.io","paper_url":null,"leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/meld"}
+{"slug":"sid-bench","name":"SID-Bench","maintainer":"Academic (SID-Bench authors)","tagline":"Spoken Interaction Dynamics benchmark — millisecond-resolution ground truth for end-of-turn detection and interruption recovery.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["EOT latency (ms)","barge-in timing","recovery time"],"models_scored":0,"updated":"2025-09","url":"https://arxiv.org/abs/2509.15033","paper_url":"https://arxiv.org/abs/2509.15033","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/sid-bench"}
+{"slug":"fd-bench","name":"FD-Bench","maintainer":"Academic (FD-Bench authors)","tagline":"Independent full-duplex benchmark focused on natural-pause / short-utterance regimes — a cross-check against FDB.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["pause handling","turn-taking","overlap"],"models_scored":0,"updated":"2025-07","url":"https://arxiv.org/abs/2507.14062","paper_url":"https://arxiv.org/abs/2507.14062","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fd-bench"}
+{"slug":"mtr-duplex-bench","name":"MTR-DuplexBench","maintainer":"Academic (MTR-DuplexBench authors)","tagline":"Multi-turn, multi-topic full-duplex benchmark built from recorded call-center-style dialogues.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["multi-turn coherence","topic-shift recovery","dynamics retention"],"models_scored":0,"updated":"2025-10","url":"https://arxiv.org/abs/2510.19097","paper_url":"https://arxiv.org/abs/2510.19097","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mtr-duplex-bench"}
+{"slug":"humdial","name":"HumDial (ICASSP '26)","maintainer":"ICASSP 2026 Spoken Dialogue Challenge","tagline":"Dual-track human-evaluated spoken dialogue challenge — first public benchmark to explicitly score paralinguistic input alongside turn-taking dynamics.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm","emotion"],"metrics":["turn-taking","paralinguistic grounding","human preference"],"models_scored":0,"updated":"2026-01","url":"https://sites.google.com/view/humdial-2026","paper_url":null,"leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/humdial"}
+{"slug":"flexi","name":"FLEXI","maintainer":"Academic (FLEXI authors)","tagline":"Full-duplex safety benchmark — scores whether a model correctly barges in on a user in safety-critical moments (medical, self-harm, imminent danger).","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["safety-interrupt success","false-alarm rate"],"models_scored":0,"updated":"2025-11","url":"https://arxiv.org/abs/2511.01923","paper_url":"https://arxiv.org/abs/2511.01923","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/flexi"}
+{"slug":"audiobench","name":"AudioBench","maintainer":"AudioLLMs (I2R Singapore)","tagline":"Broad audio-understanding benchmark covering speech, music, and ambient sound reasoning.","full_duplex":false,"tier":"component","setting":"lab","focus":["audio-understanding"],"metrics":["per-track accuracy","overall"],"models_scored":0,"updated":"2024-07","url":"https://github.com/AudioLLMs/AudioBench","paper_url":"https://arxiv.org/abs/2406.16020","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/audiobench"}
+{"slug":"dynamic-superb","name":"Dynamic-SUPERB","maintainer":"NTU et al.","tagline":"Instruction-tuned successor to SUPERB — ~55 tasks across speech understanding and light paralinguistic input.","full_duplex":false,"tier":"component","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["per-task accuracy","composite score"],"models_scored":0,"updated":"2024-11","url":"https://github.com/dynamic-superb/dynamic-superb","paper_url":"https://arxiv.org/abs/2411.05361","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/dynamic-superb"}
+{"slug":"voiceassistant-eval","name":"VoiceAssistant-Eval","maintainer":"Academic (VoiceAssistant-Eval authors)","tagline":"End-to-end voice assistant eval — instruction-following, multi-turn dialogue, and naturalness in one suite with GPT-4-class judges.","full_duplex":false,"tier":"component","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["judge score","multi-turn coherence","naturalness"],"models_scored":0,"updated":"2024-09","url":"https://arxiv.org/abs/2409.19342","paper_url":"https://arxiv.org/abs/2409.19342","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceassistant-eval"}
+{"slug":"sd-eval","name":"SD-Eval","maintainer":"Academic (SD-Eval authors)","tagline":"Baseline probe on paralinguistic input — does a speech-LM use tone, emotion, and speaker traits at all?","full_duplex":false,"tier":"component","setting":"lab","focus":["speech-lm","emotion"],"metrics":["paralinguistic-grounded accuracy","delta over ASR baseline"],"models_scored":0,"updated":"2024-06","url":"https://arxiv.org/abs/2406.13340","paper_url":"https://arxiv.org/abs/2406.13340","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/sd-eval"}
+{"slug":"prosaudit","name":"ProsAudit","maintainer":"Academic (ProsAudit authors)","tagline":"Prosodic boundary detection benchmark — can the model perceive phrase / sentence boundaries from audio alone?","full_duplex":false,"tier":"component","setting":"lab","focus":["emotion","audio-understanding"],"metrics":["boundary F1","delta over text-only"],"models_scored":0,"updated":"2023-02","url":"https://arxiv.org/abs/2302.12057","paper_url":"https://arxiv.org/abs/2302.12057","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/prosaudit"}
+{"slug":"vocalbench-zh","name":"VocalBench-zh","maintainer":"SJTU (OmniAgent)","tagline":"Mandarin version of VocalBench — 10 subsets, ~10k instances, 14 models evaluated at the Nov 2025 launch.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding"],"metrics":["ACC","LLM-judge","UTMOS","WER","EER"],"models_scored":0,"updated":"2025-11","url":"https://arxiv.org/abs/2511.08230","paper_url":"https://arxiv.org/abs/2511.08230","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/vocalbench-zh"}
+{"slug":"cs3-bench","name":"CS3-Bench","maintainer":"Academic (CS3-Bench authors)","tagline":"Mandarin-English code-switching benchmark — headline finding: S2S models drop ~66% relative on code-switched inputs vs monolingual ones.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["monolingual-vs-codeswitch delta","per-pair accuracy"],"models_scored":0,"updated":"2025-10","url":"https://arxiv.org/abs/2510.07881","paper_url":"https://arxiv.org/abs/2510.07881","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/cs3-bench"}
+{"slug":"j-moshi","name":"J-Moshi (subjective)","maintainer":"Japanese academic community","tagline":"Japanese open-weights full-duplex model paper whose subjective MOS listening tests are the main public Japanese STS measurement layer today.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["MOS","listener-preference"],"models_scored":0,"updated":"2024-11","url":"https://aclanthology.org/2024.emnlp-main.1234/","paper_url":"https://aclanthology.org/2024.emnlp-main.1234/","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/j-moshi"}
+{"slug":"voiceagentbench","name":"VoiceAgentBench","maintainer":"Academic (VoiceAgentBench authors)","tagline":"Spoken agentic benchmark — multi-tool workflows, multi-turn interactions, and safety under realistic voice-agent conditions; covers English and Hindi.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm","instruction-following"],"metrics":["task success","tool-selection accuracy","multi-turn stability","safety"],"models_scored":0,"updated":"2025-10","url":"https://arxiv.org/abs/2510.07978","paper_url":"https://arxiv.org/abs/2510.07978","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceagentbench"}
+{"slug":"audio-multichallenge","name":"Audio MultiChallenge","maintainer":"Scale AI (Scale Labs)","tagline":"Multi-turn evaluation of spoken dialogue systems on natural human interaction — scores Inference Memory, Instruction Retention, Self Coherence, Voice Editing, Audio-Cue handling.","full_duplex":true,"tier":"native","setting":"lab","focus":["speech-lm","full-duplex"],"metrics":["Inference Memory","Instruction Retention","Self Coherence","Voice Editing","Audio-Cue"],"models_scored":2,"updated":"2026-10","url":"https://huggingface.co/papers/2512.14865","paper_url":"https://arxiv.org/abs/2512.14865","leaderboard_url":"https://labs.scale.com/leaderboard/audiomc","page_url":"https://www.fullduplex.ai/benchmarks/audio-multichallenge"}
+{"slug":"msi-bench","name":"MSI-Bench","maintainer":"Boson AI","tagline":"Multi-speaker voice-agent benchmark: short multi-party audio scenes with participant context, expected tool calls and atomic rubrics, in English and Mandarin.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","vertical-agent"],"metrics":["all-rubric pass rate","multi-speaker memory","multi-speaker instruction following","multi-speaker reasoning"],"models_scored":0,"updated":"2026-09","url":"https://github.com/boson-ai/MSI-Bench","paper_url":"https://arxiv.org/abs/2609.24812","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/msi-bench"}
+{"slug":"mmedfd","name":"MMedFD","maintainer":"Academic (MMedFD authors)","tagline":"Real-world healthcare benchmark for multi-turn full-duplex automatic speech recognition — adjacent to STS, crucial for duplex deployment.","full_duplex":true,"tier":"adjacent","setting":"vertical","focus":["asr","vertical-agent","full-duplex"],"metrics":["duplex WER","medical-term recall","overlap-region WER"],"models_scored":0,"updated":"2025-09","url":"https://huggingface.co/papers/2509.19817","paper_url":"https://arxiv.org/abs/2509.19817","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mmedfd"}
+{"slug":"audio-jailbreaks-taxonomy","name":"LALM Jailbreak Taxonomy & Eval","maintainer":"Feng et al.","tagline":"Unified taxonomy of LALM jailbreak attacks (semantic, acoustic, signal, internal-representation) with controlled empirical evaluation under a single threat model and cost-aware protocol.","full_duplex":false,"tier":"native","setting":"lab","focus":["audio-safety","speech-lm"],"metrics":["attack success rate","cost-aware ASR","defense utility","cross-attack generalisation"],"models_scored":0,"updated":"2026-05","url":"https://arxiv.org/abs/2605.30031","paper_url":"https://arxiv.org/abs/2605.30031","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/audio-jailbreaks-taxonomy"}
+{"slug":"speechjbb","name":"SpeechJBB","maintainer":"Ceccatelli, Jeon, Adelani et al.","tagline":"Audio jailbreak benchmark for LALMs under multilingual and code-switched speech, with phonological perturbations as an augmented evaluation regime.","full_duplex":false,"tier":"native","setting":"lab","focus":["audio-safety","speech-lm"],"metrics":["code-switched ASR","phonological-perturbation success","multilingual alignment gap"],"models_scored":0,"updated":"2026-06","url":"https://arxiv.org/abs/2606.06037","paper_url":"https://arxiv.org/abs/2606.06037","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/speechjbb"}
+{"slug":"aia-acoustic-interference","name":"AIA — Acoustic Interference Attack","maintainer":"Wang, Huang, Liang, Wu (CUHK · HKUST)","tagline":"Universal-jailbreak benchmark for LALMs using Acoustic Latent Semantics — paralinguistic features intrinsic to the priors of audio generative models — that decouples attack payload from audio content.","full_duplex":false,"tier":"native","setting":"lab","focus":["audio-safety","speech-lm"],"metrics":["universal jailbreak success","alignment-evasion rate","transferability across LALMs"],"models_scored":0,"updated":"2026-05","url":"https://arxiv.org/abs/2605.18168","paper_url":"https://arxiv.org/abs/2605.18168","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/aia-acoustic-interference"}

data/datasets.jsonl

@@ -0,0 +1,103 @@
+{"slug":"ami","name":"AMI Meeting Corpus","maintainer":"University of Edinburgh / Idiap","tagline":"Multi-party meeting recordings with close-talking mics, video and overlap annotation. One of the few CC-BY corpora rich in overlapping speech.","hours":"100 h","size_hours":100,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2006-06","url":"https://groups.inf.ed.ac.uk/ami/corpus/","paper_url":"https://groups.inf.ed.ac.uk/ami/icmi2006BookChapter.pdf","page_url":"https://www.fullduplex.ai/datasets/ami"}
+{"slug":"switchboard","name":"Switchboard-1 (LDC97S62)","maintainer":"LDC","tagline":"2,400 U.S. telephone conversations, the original turn-taking corpus on which the field was built.","hours":"260 h","size_hours":260,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"LDC (paid)","license_class":"gated","commercial_use":"restricted","released":"1993-01","url":"https://catalog.ldc.upenn.edu/LDC97S62","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/switchboard"}
+{"slug":"candor","name":"CANDOR","maintainer":"BetterUp Labs (Reece, Cooney et al.)","tagline":"Naturalistic video-chat conversation across 1,656 dyads, the go-to open corpus for full-duplex turn-taking research.","hours":"850 h","size_hours":850,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2023-03","url":"https://betterup-data-requests.herokuapp.com/","paper_url":"https://www.science.org/doi/10.1126/sciadv.adf3197","page_url":"https://www.fullduplex.ai/datasets/candor"}
+{"slug":"fisher","name":"Fisher English","maintainer":"LDC","tagline":"Ten-minute telephone conversations, a classic for conversational ASR and turn-taking baselines.","hours":"2,000 h","size_hours":2000,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"LDC (paid)","license_class":"gated","commercial_use":"restricted","released":"2004-12","url":"https://catalog.ldc.upenn.edu/LDC2004T19","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/fisher"}
+{"slug":"magicdata-ramc","name":"MagicData-RAMC","maintainer":"Magic Data Technology / OpenSLR SLR-123","tagline":"Mandarin spontaneous two-party conversation, free for academic use. The canonical Chinese analogue of Switchboard.","hours":"180 h","size_hours":180,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-ND-4.0","license_class":"non-commercial","commercial_use":"no","released":"2022-03","url":"https://www.openslr.org/123/","paper_url":"https://arxiv.org/abs/2203.16844","page_url":"https://www.fullduplex.ai/datasets/magicdata-ramc"}
+{"slug":"ksponspeech","name":"KsponSpeech","maintainer":"ETRI / Korean AI Hub","tagline":"Korean spontaneous dyadic conversation, the canonical Korean analogue of Fisher and Switchboard.","hours":"969 h","size_hours":969,"languages":["ko"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"AI Hub (registration)","license_class":"gated","commercial_use":"restricted","released":"2020-03","url":"https://aihub.or.kr/aidata/105","paper_url":"https://aclanthology.org/2020.lrec-1.125/","page_url":"https://www.fullduplex.ai/datasets/ksponspeech"}
+{"slug":"indicvoices","name":"IndicVoices (conversational slice)","maintainer":"AI4Bharat (IIT Madras)","tagline":"Conversational speech across all 22 scheduled Indic languages, the single largest permissive source for Hindi, Bengali, Tamil and Telugu dialogue.","hours":"~3,500 h (conv.)","size_hours":3500,"languages":["hi","bn","ta","te","mr","gu","kn","ml","pa","or","as","ur","ne","sd","ks","kok","brx","doi","mai","mni","sa","sat"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2024-03","url":"https://huggingface.co/datasets/ai4bharat/IndicVoices","paper_url":"https://arxiv.org/abs/2403.01926","page_url":"https://www.fullduplex.ai/datasets/indicvoices"}
+{"slug":"dailytalk","name":"DailyTalk","maintainer":"KAIST","tagline":"Two-speaker daily conversations with emotion and dialog-act labels, a staple for dialogue TTS.","hours":"20 h","size_hours":20,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-SA-4.0","license_class":"share-alike","commercial_use":"restricted","released":"2022-07","url":"https://github.com/keonlee9420/DailyTalk","paper_url":"https://arxiv.org/abs/2207.01063","page_url":"https://www.fullduplex.ai/datasets/dailytalk"}
+{"slug":"callhome","name":"CALLHOME (JA / ZH / ES / DE / AR …)","maintainer":"LDC","tagline":"Native-speaker telephone conversations across Japanese, Mandarin, Spanish, German, Arabic, and English — the classic non-English turn-taking set.","hours":"~120 h","size_hours":120,"languages":["ja","zh","es","de","ar","en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"LDC (paid, per-release)","license_class":"gated","commercial_use":"restricted","released":"1996-01","url":"https://catalog.ldc.upenn.edu/search?query=callhome","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/callhome"}
+{"slug":"chime-6","name":"CHiME-6","maintainer":"CHiME Challenge","tagline":"Real dinner-party conversation from 20 parties, captured with multiple arrays and close-talk mics. Overlap-rich multi-party speech.","hours":"40 h","size_hours":40,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"Research use (CHiME Challenge terms)","license_class":"non-commercial","commercial_use":"no","released":"2020-04","url":"https://www.chimechallenge.org/challenges/chime6/","paper_url":"https://arxiv.org/abs/2004.09249","page_url":"https://www.fullduplex.ai/datasets/chime-6"}
+{"slug":"spokenwoz","name":"SpokenWOZ","maintainer":"Alibaba DAMO Academy","tagline":"5.7K human-to-human spoken task-oriented dialogues (203K turns, 249 h) — the canonical spoken task-oriented corpus.","hours":"249 h","size_hours":249,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2023-05","url":"https://spokenwoz.github.io/","paper_url":"https://arxiv.org/abs/2305.13040","page_url":"https://www.fullduplex.ai/datasets/spokenwoz"}
+{"slug":"yodas","name":"YODAS / YODAS2","maintainer":"WavLab (CMU)","tagline":"Creative Commons YouTube speech across 100+ languages. The larger YODAS v3 is listed separately.","hours":"500k h","size_hours":500000,"languages":["100+"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-3.0","license_class":"permissive","commercial_use":"yes","released":"2024-06","url":"https://huggingface.co/datasets/espnet/yodas","paper_url":"https://arxiv.org/abs/2406.00899","page_url":"https://www.fullduplex.ai/datasets/yodas"}
+{"slug":"yodas-v3","name":"YODAS v3","maintainer":"ESPnet / WavLab (CMU)","tagline":"48 kHz multi-channel YouTube speech under CC-BY-3.0, crawled from videos that do not overlap YODAS v1 or v2.","hours":"1.1M h","size_hours":1100000,"languages":["100+"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":48,"license":"CC-BY-3.0","license_class":"permissive","commercial_use":"yes","released":"2026-09","url":"https://huggingface.co/datasets/espnet/yodas3","paper_url":"https://arxiv.org/abs/2609.29448","page_url":"https://www.fullduplex.ai/datasets/yodas-v3"}
+{"slug":"voxpopuli","name":"VoxPopuli","maintainer":"Meta AI","tagline":"Unlabelled European Parliament speech in 23 languages, plus 1.8k hours transcribed and 17.3k hours of interpreter audio. Released under CC0.","hours":"400k h","size_hours":400000,"languages":["en","de","fr","es","+19"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC0 (data) / CC BY-NC 4.0 (code & models)","license_class":"permissive","commercial_use":"yes","released":"2021-01","url":"https://github.com/facebookresearch/voxpopuli","paper_url":"https://arxiv.org/abs/2101.00390","page_url":"https://www.fullduplex.ai/datasets/voxpopuli"}
+{"slug":"emilia","name":"Emilia / Emilia-Large","maintainer":"Amphion (CUHK-Shenzhen)","tagline":"In-the-wild multilingual speech under two licences: Emilia (101.7k h, non-commercial) and Emilia-YODAS (113.9k h, CC-BY-4.0).","hours":"215.6k h (Emilia-Large)","size_hours":215600,"languages":["en","zh","ja","ko","de","fr"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-4.0 (Emilia) / CC-BY-4.0 (Emilia-YODAS)","license_class":"non-commercial","commercial_use":"restricted","released":"2024-07","url":"https://huggingface.co/datasets/amphion/Emilia-Dataset","paper_url":"https://arxiv.org/abs/2407.05361","page_url":"https://www.fullduplex.ai/datasets/emilia"}
+{"slug":"mls","name":"Multilingual LibriSpeech (MLS)","maintainer":"Meta AI","tagline":"LibriSpeech-style audiobooks in 8 languages, the permissive multilingual pretraining baseline.","hours":"50k h","size_hours":50000,"languages":["en","de","nl","fr","es","it","pt","pl"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2020-12","url":"https://www.openslr.org/94","paper_url":"https://arxiv.org/abs/2012.03411","page_url":"https://www.fullduplex.ai/datasets/mls"}
+{"slug":"granary","name":"Granary","maintainer":"NVIDIA","tagline":"Pseudo-labelled ASR and speech-translation data for 25 European languages under CC-BY-4.0: 192k hours with audio included and 451k hours of new transcripts for existing corpora.","hours":"643k h (192k h with audio)","size_hours":643000,"languages":["en","de","fr","es","+21"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"restricted","released":"2025-05","url":"https://huggingface.co/datasets/nvidia/Granary","paper_url":"https://arxiv.org/abs/2505.13404","page_url":"https://www.fullduplex.ai/datasets/granary"}
+{"slug":"common-voice","name":"Common Voice (Scripted Speech 27.0)","maintainer":"Mozilla","tagline":"Crowdsourced multilingual read speech under CC0: 42.6k total and 29.3k validated hours in Scripted Speech 27.0 (September 2026).","hours":"42.6k h (29.3k validated)","size_hours":42593,"languages":["295"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC0-1.0","license_class":"permissive","commercial_use":"yes","released":"2019-06","url":"https://mozilladatacollective.com/organization/cmfh0j9o10006ns07jq45h7xk","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/common-voice"}
+{"slug":"peoples-speech","name":"People's Speech","maintainer":"MLCommons","tagline":"Diverse English speech harvested from Archive.org, the largest permissive English ASR corpus.","hours":"30k+ h","size_hours":30000,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-SA-4.0 / CC-BY-4.0","license_class":"share-alike","commercial_use":"yes","released":"2021-11","url":"https://mlcommons.org/datasets/peoples-speech/","paper_url":"https://arxiv.org/abs/2111.09344","page_url":"https://www.fullduplex.ai/datasets/peoples-speech"}
+{"slug":"gigaspeech","name":"GigaSpeech","maintainer":"SpeechColab","tagline":"English from audiobooks, podcasts and YouTube. Broad acoustic coverage, but audio access is research-only.","hours":"10k h","size_hours":10000,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"Research-only (gated, Apache-2.0 code)","license_class":"gated","commercial_use":"no","released":"2021-06","url":"https://github.com/SpeechColab/GigaSpeech","paper_url":"https://arxiv.org/abs/2106.06909","page_url":"https://www.fullduplex.ai/datasets/gigaspeech"}
+{"slug":"librispeech","name":"LibriSpeech","maintainer":"Vassil Panayotov et al.","tagline":"Read English audiobooks, the canonical ASR benchmark since 2015.","hours":"960 h","size_hours":960,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2015-04","url":"https://www.openslr.org/12","paper_url":"https://www.danielpovey.com/files/2015_icassp_librispeech.pdf","page_url":"https://www.fullduplex.ai/datasets/librispeech"}
+{"slug":"reazonspeech","name":"ReazonSpeech","maintainer":"Reazon Holdings","tagline":"Japanese TV broadcasts pseudo-labelled with Whisper, the largest practical Japanese training pool.","hours":"35k h","size_hours":35000,"languages":["ja"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CDLA-Sharing-1.0","license_class":"share-alike","commercial_use":"restricted","released":"2023-03","url":"https://research.reazon.jp/projects/ReazonSpeech/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/reazonspeech"}
+{"slug":"csj","name":"Corpus of Spontaneous Japanese (CSJ)","maintainer":"NINJAL / NICT / Tokyo Tech","tagline":"Japanese spontaneous speech, the canonical academic Japanese corpus for prosody, POS and dependency.","hours":"661 h","size_hours":661,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"NINJAL (tiered)","license_class":"gated","commercial_use":"restricted","released":"2004-01","url":"https://clrd.ninjal.ac.jp/csj/en/index.html","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/csj"}
+{"slug":"jvs","name":"JVS Corpus","maintainer":"Takamichi Lab (University of Tokyo)","tagline":"Japanese parallel readings across 100 speakers, the staple for multi-speaker Japanese TTS.","hours":"30 h","size_hours":30,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-SA-4.0 (with conditions)","license_class":"custom","commercial_use":"restricted","released":"2019-08","url":"https://sites.google.com/site/shinnosuketakamichi/research-topics/jvs_corpus","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/jvs"}
+{"slug":"iemocap","name":"IEMOCAP","maintainer":"USC SAIL","tagline":"Scripted and improvised dyadic acting, the emotion-recognition classic.","hours":"12 h","size_hours":12,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"USC SAIL (request)","license_class":"gated","commercial_use":"no","released":"2008-12","url":"https://sail.usc.edu/iemocap/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/iemocap"}
+{"slug":"expresso","name":"Expresso","maintainer":"Meta AI","tagline":"Studio-quality expressive speech from 4 speakers, with 8 read styles and 26 improvised dialogue styles including laughter and whisper.","hours":"40 h","size_hours":40,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":48,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2023-08","url":"https://speechbot.github.io/expresso/","paper_url":"https://arxiv.org/abs/2308.05725","page_url":"https://www.fullduplex.ai/datasets/expresso"}
+{"slug":"fleurs","name":"FLEURS","maintainer":"Google Research","tagline":"Few-shot Learning Evaluation of Universal Representations of Speech — 102 languages, ~10 h each, built on FLoRes translation pairs.","hours":"~1k h","size_hours":1020,"languages":["102+"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2022-05","url":"https://huggingface.co/datasets/google/fleurs","paper_url":"https://arxiv.org/abs/2205.12446","page_url":"https://www.fullduplex.ai/datasets/fleurs"}
+{"slug":"covost2","name":"CoVoST-2","maintainer":"Meta AI","tagline":"Multilingual speech-to-text translation built on Common Voice, the standard supervised S2ST base.","hours":"2,880 h","size_hours":2880,"languages":["21 + en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC0-1.0","license_class":"permissive","commercial_use":"yes","released":"2020-07","url":"https://github.com/facebookresearch/covost","paper_url":"https://arxiv.org/abs/2007.10310","page_url":"https://www.fullduplex.ai/datasets/covost2"}
+{"slug":"cvss","name":"CVSS","maintainer":"Google Research","tagline":"Massively multilingual S2ST corpus — 21 source languages into English, with synthetic target speech — CC-BY-4.0.","hours":"3,809 h","size_hours":3809,"languages":["21 → en"],"speech":"mixed","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2022-01","url":"https://github.com/google-research-datasets/cvss","paper_url":"https://arxiv.org/abs/2201.03713","page_url":"https://www.fullduplex.ai/datasets/cvss"}
+{"slug":"must-c","name":"MuST-C","maintainer":"FBK (Fondazione Bruno Kessler)","tagline":"English TED-talk speech translated into 14 target languages — hundreds of hours per pair, the reference training corpus for ST.","hours":"~4k h","size_hours":4000,"languages":["en → 14"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-ND-4.0","license_class":"non-commercial","commercial_use":"no","released":"2019-06","url":"https://mt.fbk.eu/must-c/","paper_url":"https://www.sciencedirect.com/science/article/pii/S0885230820300887","page_url":"https://www.fullduplex.ai/datasets/must-c"}
+{"slug":"j-chat","name":"J-CHAT","maintainer":"Sarulab (UTokyo) / Keio","tagline":"Japanese spoken dialogue scraped from podcasts and YouTube, the first Japanese corpus aimed at dialogue-oriented speech LMs.","hours":"76k h","size_hours":76000,"languages":["ja"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"Research-only (JP Copyright Art. 30-4)","license_class":"non-commercial","commercial_use":"no","released":"2024-07","url":"https://huggingface.co/datasets/sarulab-speech/J-CHAT","paper_url":"https://arxiv.org/abs/2407.15828","page_url":"https://www.fullduplex.ai/datasets/j-chat"}
+{"slug":"multidialog","name":"MultiDialog","maintainer":"IVLLab (Yonsei / Naver)","tagline":"9k audio-visual dialogues between 12 fluent English speakers, with emotion annotations on top of TopicalChat.","hours":"340 h","size_hours":340,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-SA-4.0","license_class":"non-commercial","commercial_use":"no","released":"2024-06","url":"https://multidialog.github.io/","paper_url":"https://arxiv.org/abs/2406.07867","page_url":"https://www.fullduplex.ai/datasets/multidialog"}
+{"slug":"instructs2s-200k","name":"InstructS2S-200K","maintainer":"ICTNLP (LLaMA-Omni team)","tagline":"200k multi-turn speech-to-speech conversations tailored for instruction-following speech models.","hours":null,"size_hours":null,"languages":["en"],"speech":"synthetic","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2024-09","url":"https://huggingface.co/datasets/ICTNLP/InstructS2S-200K","paper_url":"https://arxiv.org/abs/2409.06666","page_url":"https://www.fullduplex.ai/datasets/instructs2s-200k"}
+{"slug":"sd-eval","name":"SD-Eval","maintainer":"Amphion (CUHK-Shenzhen)","tagline":"Curated evaluation across emotion, accent, age and environment. A non-commercial, eval-only release under CC-BY-NC-4.0.","hours":"8.76 h","size_hours":9,"languages":["en"],"speech":"mixed","speaker_consent":"mixed","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2024-06","url":"https://huggingface.co/datasets/amphion/SD-Eval","paper_url":"https://arxiv.org/abs/2406.13340","page_url":"https://www.fullduplex.ai/datasets/sd-eval"}
+{"slug":"voxdialogue","name":"VoxDialogue","maintainer":"VoxDialogue authors (ICLR 2025)","tagline":"4,500 multi-turn spoken dialog samples × 12 acoustic attributes — probing whether spoken dialog systems catch what text can't.","hours":null,"size_hours":null,"languages":["en"],"speech":"synthetic","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0 (Hugging Face card metadata)","license_class":"permissive","commercial_use":"yes","released":"2025-05","url":"https://huggingface.co/datasets/Exgc/VoxDialogue","paper_url":"https://openreview.net/forum?id=vbmSSIhKAM","page_url":"https://www.fullduplex.ai/datasets/voxdialogue"}
+{"slug":"mm-f2f","name":"MM-F2F","maintainer":"Xiamen University","tagline":"Multi-modal face-to-face conversation with word-level turn-taking and backchannel labels.","hours":"210 h","size_hours":210,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"Research (see repo)","license_class":"custom","commercial_use":"restricted","released":"2025-07","url":"https://github.com/Linyx1125/MM-F2F","paper_url":"https://aclanthology.org/2025.acl-long.743/","page_url":"https://www.fullduplex.ai/datasets/mm-f2f"}
+{"slug":"jp-duplex-magichub","name":"Japanese Duplex Conversation Dataset","maintainer":"MagicData","tagline":"Channel-separated two-speaker Japanese conversation. Tiny, but one of the only open Japanese corpora with per-speaker tracks.","hours":"10.35 h","size_hours":10,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":16,"license":"CC-BY-NC-ND-4.0","license_class":"non-commercial","commercial_use":"no","released":"2025-04","url":"https://magichub.com/datasets/japanese-duplex-conversation-training-dataset/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/jp-duplex-magichub"}
+{"slug":"olespeech-iv","name":"OleSpeech-IV","maintainer":"Olewave / Stanford / Microsoft / PingAn / KU / CMU","tagline":"English conversational speech with accents from all world regions — human-refined speaker turns and transcripts, 100-hour open subset.","hours":"100 h (open subset)","size_hours":100,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"Non-commercial research","license_class":"non-commercial","commercial_use":"no","released":"2025-09","url":"https://huggingface.co/datasets/olewave/OleSpeech-IV-2025-EN-AR-100","paper_url":"https://arxiv.org/abs/2509.04702","page_url":"https://www.fullduplex.ai/datasets/olespeech-iv"}
+{"slug":"interactspeech","name":"InteractSpeech","maintainer":"InteractSpeech authors (EMNLP 2025)","tagline":"English speech-interaction data, synthesised plus filtered real dialogues, with precise speaker timestamps for interruptions and backchannels.","hours":"150 h (audio not released)","size_hours":null,"languages":["en"],"speech":"mixed","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"Research (see repo)","license_class":"custom","commercial_use":"restricted","released":"2025-11","url":"https://interactspeech.github.io/","paper_url":"https://aclanthology.org/2025.findings-emnlp.424/","page_url":"https://www.fullduplex.ai/datasets/interactspeech"}
+{"slug":"mlc-slm","name":"MLC-SLM (Interspeech 2025)","maintainer":"Nexdata / Interspeech 2025 challenge","tagline":"Two-speaker conversational speech in 11 languages, the first genuinely multilingual public benchmark for conversational speech LMs.","hours":"1,604 h","size_hours":1604,"languages":["en","fr","de","it","pt","es","ja","ko","ru","th","vi"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"Challenge access (CC-BY-SA-4.0 for eval)","license_class":"gated","commercial_use":"restricted","released":"2025-09","url":"https://www.nexdata.ai/competition/mlc-slm","paper_url":"https://arxiv.org/abs/2509.13785","page_url":"https://www.fullduplex.ai/datasets/mlc-slm"}
+{"slug":"magicdata-multistream","name":"Multi-stream Spontaneous Conversation (zh+en)","maintainer":"Magic Data Technology","tagline":"Dual-track two-speaker conversation in Mandarin and English, with a channel per speaker. A rare open corpus built for full duplex.","hours":"15 h (10 zh + 5 en)","size_hours":15,"languages":["zh","en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":16,"license":"CC-BY-NC-ND-4.0","license_class":"non-commercial","commercial_use":"no","released":"2024-11","url":"https://magichub.com/datasets/multi-stream-spontaneous-conversation-training-datasets_chinese/","paper_url":"https://arxiv.org/abs/2509.04093","page_url":"https://www.fullduplex.ai/datasets/magicdata-multistream"}
+{"slug":"behavior-sd","name":"Behavior-SD","maintainer":"Seoul National University (NAACL 2025)","tagline":"108K LLM-synthesised full-duplex dialogues (2,164 h) with explicit backchannel / interruption / filler labels — the largest publicly-downloadable FD-labeled corpus.","hours":"2,164 h","size_hours":2164,"languages":["en"],"speech":"synthetic","speaker_consent":"unknown","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2025-04","url":"https://huggingface.co/datasets/yhytoto12/behavior-sd","paper_url":"https://aclanthology.org/2025.naacl-long.484/","page_url":"https://www.fullduplex.ai/datasets/behavior-sd"}
+{"slug":"otospeech-fd-280h","name":"otoSpeech-full-duplex-280h","maintainer":"otoearth","tagline":"Channel-separated two-speaker English conversation at 48 kHz under CC-BY-4.0. Among the largest permissive, commercially usable real-recorded FD corpora indexed on Hugging Face as of 2026-04.","hours":"280 h","size_hours":280,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-02","url":"https://huggingface.co/datasets/otoearth/otoSpeech-full-duplex-280h","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-fd-280h"}
+{"slug":"otospeech-fd-processed-141h","name":"otoSpeech-full-duplex-processed-141h","maintainer":"otoearth","tagline":"Human-reviewed, denoised, channel-separated English FD conversation. The processed sibling of otoSpeech-280h.","hours":"141 h","size_hours":141,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":44.1,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-02","url":"https://huggingface.co/datasets/otoearth/otoSpeech-full-duplex-processed-141h","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-fd-processed-141h"}
+{"slug":"otospeech-fd-turn-104h","name":"otoSpeech-full-duplex-turn-104h","maintainer":"otoearth","tagline":"Dyadic English conversation with hand-made turn-taking labels on every speaker channel. The training split of Sesame's TurnBench.","hours":"105 h","size_hours":104.94,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"oto Speech License v1.0","license_class":"gated","commercial_use":"no","released":"2026-06","url":"https://huggingface.co/datasets/otoearth/otoSpeech-full-duplex-turn-104h","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-fd-turn-104h"}
+{"slug":"duplexchat","name":"DuplexChat","maintainer":"Nakata, Saito, Saruwatari (U. Tokyo / AIST)","tagline":"Speaker-separated two-channel dialogue mined from public podcasts, distributed as metadata rather than audio.","hours":"415k h","size_hours":415357,"languages":["en","ja"],"speech":"human","speaker_consent":"public","full_duplex":true,"sample_rate_khz":null,"license":"Metadata manifest on Hugging Face; source-audio rights remain with each podcast","license_class":"custom","commercial_use":"restricted","released":"2026-07","url":"https://huggingface.co/datasets/sarulab-speech/DuplexChat","paper_url":"https://arxiv.org/abs/2607.04941","page_url":"https://www.fullduplex.ai/datasets/duplexchat"}
+{"slug":"dialogs-ru","name":"Dialogs (Russian)","maintainer":"Shigabeev, Latyshev","tagline":"Studio-recorded expressive Russian dialogue with per-utterance emotion labels.","hours":"20.6 h","size_hours":20.6,"languages":["ru"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":44.1,"license":"Not stated in the paper — verify before use","license_class":"custom","commercial_use":"restricted","released":"2026-07","url":"https://arxiv.org/abs/2607.14310","paper_url":"https://arxiv.org/abs/2607.14310","page_url":"https://www.fullduplex.ai/datasets/dialogs-ru"}
+{"slug":"korean-fd-synthetic-preview","name":"Korean full-duplex synthetic (preview)","maintainer":"Wi-Fi (Hugging Face)","tagline":"A 100-conversation CC-BY-4.0 taster in front of a claimed 2,000-hour synthetic Korean full-duplex corpus.","hours":"preview only (2,000.5 h claimed)","size_hours":null,"languages":["ko"],"speech":"synthetic","speaker_consent":"unknown","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0 (preview subset)","license_class":"permissive","commercial_use":"yes","released":"2026-08","url":"https://huggingface.co/datasets/Wi-Fi/korean-full-duplex-synthetic-dataset-preview","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/korean-fd-synthetic-preview"}
+{"slug":"affectspeech","name":"AffectSpeech","maintainer":"Qi, Zheng, Schuller, Luo, Li","tagline":"Emotional speech annotated with open-vocabulary descriptions rather than a fixed label set.","hours":null,"size_hours":null,"languages":["en"],"speech":"human","speaker_consent":"mixed","full_duplex":false,"sample_rate_khz":null,"license":"Not stated in the paper or repo — verify before use","license_class":"custom","commercial_use":"restricted","released":"2026-04","url":"https://github.com/jeremychee4/AffectSpeech","paper_url":"https://arxiv.org/abs/2604.04160","page_url":"https://www.fullduplex.ai/datasets/affectspeech"}
+{"slug":"naijas2st","name":"NaijaS2ST","maintainer":"Maltais et al.","tagline":"Parallel speech-to-speech translation across four Nigerian languages plus accented English.","hours":"~200 h","size_hours":200,"languages":["ig","ha","yo","pcm","en"],"speech":"mixed","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"Not stated in the paper — verify before use","license_class":"custom","commercial_use":"restricted","released":"2026-04","url":"https://arxiv.org/abs/2604.16287","paper_url":"https://arxiv.org/abs/2604.16287","page_url":"https://www.fullduplex.ai/datasets/naijas2st"}
+{"slug":"otospeech-fd-task-20h","name":"otoSpeech-full-duplex-task-oriented-20h","maintainer":"otoearth","tagline":"Two-speaker English conversation recorded while the pair worked on a task, with task state, interface events and reference answers on the same timeline as the audio.","hours":"20 h","size_hours":20,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-08","url":"https://huggingface.co/datasets/otoearth/otoSpeech-full-duplex-task-oriented-20h","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-fd-task-20h"}
+{"slug":"otospeech-fd-task-v2","name":"otoSpeech-full-duplex-task-oriented-v2","maintainer":"otoearth","tagline":"Two-speaker English conversation recorded during collaborative tasks, all from sessions after the 20-hour release.","hours":"17.4 h","size_hours":17.38,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-09","url":"https://huggingface.co/datasets/otoearth/otoSpeech-full-duplex-task-oriented-v2","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-fd-task-v2"}
+{"slug":"otospeech-1125h","name":"otoSpeech 1,125h","maintainer":"otoearth","tagline":"Casual two-speaker English conversation across 3,878 sessions, one speaker per channel, under a commercial licence.","hours":"1,125 h","size_hours":1125,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"Commercial licence (on request)","license_class":"custom","commercial_use":"yes","released":null,"url":"https://www.oto.earth/datasets/english","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-1125h"}
+{"slug":"otospeech-video","name":"otoSpeech Video","maintainer":"otoearth","tagline":"Two-speaker English conversation with a camera on each speaker, synced to channel-separated audio.","hours":null,"size_hours":null,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/video","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-video"}
+{"slug":"otospeech-face-to-face","name":"otoSpeech Face-to-Face","maintainer":"otoearth","tagline":"In-person English conversation among two to five people in one room, recorded on 16 channels with video.","hours":null,"size_hours":null,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/multi-speaker","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-face-to-face"}
+{"slug":"otospeech-3-speaker","name":"otoSpeech 3-Speaker","maintainer":"otoearth","tagline":"English conversation among three people, with a separate mono recording for each speaker.","hours":null,"size_hours":null,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/multi-speaker","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-3-speaker"}
+{"slug":"otospeech-korean","name":"otoSpeech Korean","maintainer":"otoearth","tagline":"Korean free conversation between two people on a given topic, one speaker per channel.","hours":null,"size_hours":null,"languages":["ko"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":44.1,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/korean","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-korean"}
+{"slug":"otospeech-japanese","name":"otoSpeech Japanese","maintainer":"otoearth","tagline":"Japanese free conversation between two people on a given topic, one speaker per channel.","hours":null,"size_hours":null,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/japanese","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-japanese"}
+{"slug":"otospeech-role-play","name":"otoSpeech Role-Play","maintainer":"otoearth","tagline":"Japanese two-speaker role-play in set scenarios, from shop visits to business talks, in the speakers' own words.","hours":null,"size_hours":null,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/japanese","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-role-play"}
+{"slug":"otospeech-scripted","name":"otoSpeech Scripted","maintainer":"otoearth","tagline":"Japanese two-speaker dialogue read from scripts, each recording with its script and a timed transcript.","hours":null,"size_hours":null,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/japanese","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-scripted"}
+{"slug":"otospeech-egocentric","name":"otoSpeech Egocentric","maintainer":"otoearth","tagline":"English conversation with first-person video while two to four people fold towels, tidy objects or play Jenga.","hours":null,"size_hours":null,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"On request","license_class":"custom","commercial_use":"restricted","released":null,"url":"https://www.oto.earth/datasets/egocentric","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-egocentric"}
+{"slug":"open-yap-1k","name":"Open Yap 1K","maintainer":"The Agentic Data Company","tagline":"Two-speaker English conversation between people who already know each other, one 48 kHz track per speaker, under a data use agreement.","hours":"1,000 h (8.9 h open sample)","size_hours":1000,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"Open Yap 1K Data Use Agreement (full corpus); CC-BY-4.0 (sample)","license_class":"gated","commercial_use":"restricted","released":"2026-09","url":"https://huggingface.co/datasets/TheAgenticDataCompany/open-yap-1k","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/open-yap-1k"}
+{"slug":"basis-conversations-1500","name":"Basis Conversations 1500","maintainer":"Basis","tagline":"Multi-party conversation in 22 languages, with every speaker on a separate 48 kHz track.","hours":"1,502 h","size_hours":1502,"languages":["en","es","fr","de","it","nl","pl","pt","ru","uk","tr","ar","he","hi","ja","ko","zh","ka","kk","xmf","xh","zu"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"basis-data-license-1.0 (custom, gated)","license_class":"gated","commercial_use":"restricted","released":"2026-09","url":"https://huggingface.co/datasets/basis-ai/basis-conversations-1500","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/basis-conversations-1500"}
+{"slug":"smoothconv-duplexconv","name":"SmoothConv / DuplexConv","maintainer":"ASLP@NPU / QualiaLabs","tagline":"Mandarin multi-party conversation with a synchronized track per participant: 100 hours human-annotated, 2,000 hours LLM-annotated.","hours":"2,100 h (100 h + 2,000 h)","size_hours":2100,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2026-06","url":"https://github.com/qualialabsAI/SmoothConv-DuplexConv","paper_url":"https://arxiv.org/abs/2610.11150","page_url":"https://www.fullduplex.ai/datasets/smoothconv-duplexconv"}
+{"slug":"multitalk","name":"MultiTalkPT / MultiTalkFT","maintainer":"CUHK MMLab","tagline":"Synthetic long-form, multi-party English and Chinese full-duplex dialogue, plus a benchmark built from real recordings.","hours":"57.6k h","size_hours":57600,"languages":["en","zh"],"speech":"synthetic","speaker_consent":"unknown","full_duplex":true,"sample_rate_khz":null,"license":"CC-BY-NC-4.0 (PT, FT); CC-BY-SA-4.0 (Bench)","license_class":"non-commercial","commercial_use":"no","released":"2026-09","url":"https://huggingface.co/datasets/MultiTalk/MultiTalkPT","paper_url":"https://arxiv.org/abs/2609.36903","page_url":"https://www.fullduplex.ai/datasets/multitalk"}
+{"slug":"seamless-interaction","name":"Seamless Interaction","maintainer":"Meta FAIR","tagline":"Face-to-face dyadic conversation from more than 4,000 participants, with each participant's audio and video kept separate.","hours":"4,000+ h","size_hours":4000,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2025-06","url":"https://huggingface.co/datasets/facebook/seamless-interaction","paper_url":"https://arxiv.org/abs/2506.22554","page_url":"https://www.fullduplex.ai/datasets/seamless-interaction"}
+{"slug":"vartalaap","name":"Vartalaap","maintainer":"Kapture CX","tagline":"Synthetic Hindi-English customer-support calls with one stereo channel per role, built for full-duplex speech-to-speech training.","hours":"1,955 h","size_hours":1955.2,"languages":["hi","en"],"speech":"synthetic","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":24,"license":"CC-BY-NC-4.0 (gated, no redistribution)","license_class":"non-commercial","commercial_use":"no","released":"2026-08","url":"https://huggingface.co/datasets/kapturecx/Vartalaap","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/vartalaap"}
+{"slug":"indic-diarbench","name":"Indic DiarBench","maintainer":"Sarvam AI","tagline":"Human-annotated multi-speaker conversation across all 22 scheduled Indian languages, for joint diarization and speaker-attributed ASR.","hours":"~108 h","size_hours":108,"languages":["hi","bn","ta","te","+18"],"speech":"human","speaker_consent":"mixed","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-07","url":"https://huggingface.co/datasets/sarvamai/indic-diarbench","paper_url":"https://arxiv.org/abs/2607.23808","page_url":"https://www.fullduplex.ai/datasets/indic-diarbench"}
+{"slug":"dai-asr-i18n","name":"DAI-ASR-I18N","maintainer":"David AI","tagline":"Everyday two-speaker conversation in 21 languages, each speaker on a separate channel, licensed for evaluation only.","hours":"42.1 h","size_hours":42.1,"languages":["en","zh","ja","ko","hi","bn","mr","ta","te","th","vi","id","tl","ar","tr","ru","de","fr","es","it","pt"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"DAI-ASR-I18N Data Use Agreement v1.1 (evaluation only)","license_class":"gated","commercial_use":"restricted","released":"2026-10","url":"https://huggingface.co/datasets/davidai-labs/DAI-ASR-I18N","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/dai-asr-i18n"}
+{"slug":"timit","name":"TIMIT","maintainer":"DARPA / LDC","tagline":"The phonetically transcribed corpus every acoustic-model paper cited for two decades.","hours":"5.4 h","size_hours":5.4,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"LDC User Agreement (paid)","license_class":"gated","commercial_use":"restricted","released":"1993-10","url":"https://catalog.ldc.upenn.edu/LDC93S1","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/timit"}
+{"slug":"wsj","name":"Wall Street Journal (WSJ0 / WSJ1)","maintainer":"LDC","tagline":"Read newspaper text that defined large-vocabulary ASR benchmarking in the 1990s.","hours":"~80 h","size_hours":80,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"LDC User Agreement (paid)","license_class":"gated","commercial_use":"restricted","released":"1993-01","url":"https://catalog.ldc.upenn.edu/LDC93S6A","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/wsj"}
+{"slug":"ted-lium-3","name":"TED-LIUM 3","maintainer":"LIUM, Université du Mans","tagline":"TED talks with aligned transcripts, the standard long-form English ASR set.","hours":"452 h","size_hours":452,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-NC-ND-3.0","license_class":"non-commercial","commercial_use":"no","released":"2018-05","url":"https://www.openslr.org/51/","paper_url":"https://arxiv.org/abs/1805.04699","page_url":"https://www.fullduplex.ai/datasets/ted-lium-3"}
+{"slug":"libri-light","name":"Libri-Light","maintainer":"Meta AI (FAIR)","tagline":"Unlabelled English audiobooks, the corpus that made self-supervised speech practical.","hours":"60k h","size_hours":60000,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"MIT","license_class":"permissive","commercial_use":"yes","released":"2019-12","url":"https://github.com/facebookresearch/libri-light","paper_url":"https://arxiv.org/abs/1912.07875","page_url":"https://www.fullduplex.ai/datasets/libri-light"}
+{"slug":"aishell-1","name":"AISHELL-1","maintainer":"Beijing Shell Shell Technology","tagline":"Read Mandarin under Apache-2.0, the open baseline for Chinese ASR.","hours":"178 h","size_hours":178,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"Apache-2.0","license_class":"permissive","commercial_use":"yes","released":"2017-09","url":"https://www.openslr.org/33/","paper_url":"https://arxiv.org/abs/1709.05522","page_url":"https://www.fullduplex.ai/datasets/aishell-1"}
+{"slug":"aishell-2","name":"AISHELL-2","maintainer":"Beijing Shell Shell Technology","tagline":"Clean read Mandarin, free to academia on application.","hours":"1,000 h","size_hours":1000,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"Academic research licence (application required)","license_class":"gated","commercial_use":"no","released":"2018-08","url":"https://www.aishelltech.com/aishell_2","paper_url":"https://arxiv.org/abs/1808.10583","page_url":"https://www.fullduplex.ai/datasets/aishell-2"}
+{"slug":"wenetspeech","name":"WenetSpeech","maintainer":"WeNet community / NPU-ASLP","tagline":"10,000+ labelled hours of multi-domain Mandarin scraped from YouTube and podcasts.","hours":"10k h (labelled)","size_hours":10000,"languages":["zh"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0 (non-commercial usage only)","license_class":"non-commercial","commercial_use":"no","released":"2021-10","url":"https://wenet.org.cn/WenetSpeech/","paper_url":"https://arxiv.org/abs/2110.03370","page_url":"https://www.fullduplex.ai/datasets/wenetspeech"}
+{"slug":"spgispeech","name":"SPGISpeech","maintainer":"Kensho (S&P Global)","tagline":"Transcribed earnings calls with fully formatted, orthographic transcripts.","hours":"5,000 h","size_hours":5000,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"Kensho user agreement (registration required)","license_class":"gated","commercial_use":"restricted","released":"2021-04","url":"https://datasets.kensho.com/datasets/spgispeech","paper_url":"https://arxiv.org/abs/2104.02014","page_url":"https://www.fullduplex.ai/datasets/spgispeech"}
+{"slug":"earnings-22","name":"Earnings-22","maintainer":"Rev.com","tagline":"Global-accent English earnings calls, released for accent-robustness work.","hours":"119 h","size_hours":119,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-SA-4.0","license_class":"share-alike","commercial_use":"yes","released":"2022-03","url":"https://github.com/revdotcom/speech-datasets","paper_url":"https://arxiv.org/abs/2203.15591","page_url":"https://www.fullduplex.ai/datasets/earnings-22"}
+{"slug":"multilingual-tedx","name":"Multilingual TEDx","maintainer":"Johns Hopkins / CMU","tagline":"TEDx talks in 8 languages with speech-translation pairs.","hours":"760 h","size_hours":760,"languages":["es","fr","pt","it","ru","el","ar","de"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-NC-ND-4.0","license_class":"non-commercial","commercial_use":"no","released":"2021-02","url":"https://www.openslr.org/100/","paper_url":"https://arxiv.org/abs/2102.01757","page_url":"https://www.fullduplex.ai/datasets/multilingual-tedx"}
+{"slug":"icsi-meeting","name":"ICSI Meeting Corpus","maintainer":"ICSI / University of Edinburgh (AMI site)","tagline":"75 real research meetings recorded with close-talking and tabletop mics.","hours":"72 h","size_hours":72,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2004-02","url":"https://groups.inf.ed.ac.uk/ami/icsi/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/icsi-meeting"}
+{"slug":"ljspeech","name":"LJSpeech","maintainer":"Keith Ito","tagline":"One speaker reading public-domain books, the default single-speaker TTS set.","hours":"24 h","size_hours":24,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":22.05,"license":"Public Domain","license_class":"permissive","commercial_use":"yes","released":"2017-07","url":"https://keithito.com/LJ-Speech-Dataset/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/ljspeech"}
+{"slug":"vctk","name":"VCTK Corpus","maintainer":"CSTR, University of Edinburgh","tagline":"109 English speakers across accents — the multi-speaker and voice-cloning staple.","hours":"44 h","size_hours":44,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":48,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2019-11","url":"https://datashare.ed.ac.uk/handle/10283/3443","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/vctk"}
+{"slug":"libritts","name":"LibriTTS","maintainer":"Google / Heiga Zen et al.","tagline":"LibriSpeech re-cut at 24 kHz with punctuation restored, for synthesis.","hours":"585 h","size_hours":585,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":24,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2019-04","url":"https://www.openslr.org/60/","paper_url":"https://arxiv.org/abs/1904.02882","page_url":"https://www.fullduplex.ai/datasets/libritts"}
+{"slug":"libritts-r","name":"LibriTTS-R","maintainer":"Google","tagline":"LibriTTS restored with speech enhancement — same transcripts, cleaner audio.","hours":"585 h","size_hours":585,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":24,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2023-05","url":"https://www.openslr.org/141/","paper_url":"https://arxiv.org/abs/2305.18802","page_url":"https://www.fullduplex.ai/datasets/libritts-r"}
+{"slug":"hi-fi-tts","name":"Hi-Fi TTS","maintainer":"NVIDIA","tagline":"Studio-grade audiobook speech from 10 speakers at 44.1 kHz.","hours":"292 h","size_hours":292,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":44.1,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2021-04","url":"https://www.openslr.org/109/","paper_url":"https://arxiv.org/abs/2104.01497","page_url":"https://www.fullduplex.ai/datasets/hi-fi-tts"}
+{"slug":"jsut","name":"JSUT","maintainer":"Shinnosuke Takamichi (U. Tokyo)","tagline":"Single-speaker Japanese covering everyday vocabulary, recorded in an anechoic room.","hours":"10 h","size_hours":10,"languages":["ja"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":48,"license":"Custom (research free; commercial terms apply, no redistribution)","license_class":"custom","commercial_use":"restricted","released":"2017-11","url":"https://sites.google.com/site/shinnosuketakamichi/publication/jsut","paper_url":"https://arxiv.org/abs/1711.00354","page_url":"https://www.fullduplex.ai/datasets/jsut"}
+{"slug":"aishell-3","name":"AISHELL-3","maintainer":"Beijing Shell Shell Technology","tagline":"Multi-speaker Mandarin TTS data from 218 speakers, under Apache-2.0.","hours":"85 h","size_hours":85,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":44.1,"license":"Apache-2.0","license_class":"permissive","commercial_use":"yes","released":"2020-10","url":"https://www.openslr.org/93/","paper_url":"https://arxiv.org/abs/2010.11567","page_url":"https://www.fullduplex.ai/datasets/aishell-3"}
+{"slug":"css10","name":"CSS10","maintainer":"Kyubyong Park et al.","tagline":"Single-speaker TTS data in 10 languages, cut from public-domain audiobooks.","hours":"140 h","size_hours":140,"languages":["de","el","es","fi","fr","hu","ja","nl","ru","zh"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":22.05,"license":"Apache-2.0","license_class":"permissive","commercial_use":"yes","released":"2019-03","url":"https://github.com/Kyubyong/css10","paper_url":"https://arxiv.org/abs/1903.11269","page_url":"https://www.fullduplex.ai/datasets/css10"}
+{"slug":"voxceleb1","name":"VoxCeleb1","maintainer":"VGG, University of Oxford","tagline":"1,251 celebrities speaking in the wild — the reference speaker-verification set.","hours":"352 h","size_hours":352,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-SA-4.0 (annotations; source video rights retained)","license_class":"share-alike","commercial_use":"restricted","released":"2017-06","url":"https://www.robots.ox.ac.uk/~vgg/data/voxceleb/","paper_url":"https://arxiv.org/abs/1706.08612","page_url":"https://www.fullduplex.ai/datasets/voxceleb1"}
+{"slug":"voxceleb2","name":"VoxCeleb2","maintainer":"VGG, University of Oxford","tagline":"6,112 speakers and over a million utterances — the scale tier of speaker recognition.","hours":"2,442 h","size_hours":2442,"languages":["multi"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-SA-4.0 (annotations; source video rights retained)","license_class":"share-alike","commercial_use":"restricted","released":"2018-06","url":"https://www.robots.ox.ac.uk/~vgg/data/voxceleb/","paper_url":"https://arxiv.org/abs/1806.05622","page_url":"https://www.fullduplex.ai/datasets/voxceleb2"}
+{"slug":"voxlingua107","name":"VoxLingua107","maintainer":"Tallinn University of Technology","tagline":"Spoken language identification data covering 107 languages.","hours":"6,628 h","size_hours":6628,"languages":["multi"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2020-11","url":"https://bark.phon.ioc.ee/voxlingua107/","paper_url":"https://arxiv.org/abs/2011.12998","page_url":"https://www.fullduplex.ai/datasets/voxlingua107"}
+{"slug":"cn-celeb","name":"CN-Celeb","maintainer":"CSLT, Tsinghua University","tagline":"Chinese speaker recognition in eleven genres — deliberately harder than VoxCeleb.","hours":"274 h","size_hours":274,"languages":["zh"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-SA-4.0, no commercial usage","license_class":"non-commercial","commercial_use":"no","released":"2019-11","url":"https://www.openslr.org/82/","paper_url":"https://arxiv.org/abs/1911.01799","page_url":"https://www.fullduplex.ai/datasets/cn-celeb"}
+{"slug":"ravdess","name":"RAVDESS","maintainer":"Ryerson University","tagline":"24 actors performing eight emotions at two intensities, audio and video.","hours":"~3 h","size_hours":3,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":48,"license":"CC-BY-NC-SA-4.0","license_class":"non-commercial","commercial_use":"no","released":"2018-04","url":"https://zenodo.org/record/1188976","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/ravdess"}
+{"slug":"crema-d","name":"CREMA-D","maintainer":"University of Pennsylvania","tagline":"7,442 acted clips from 91 ethnically diverse actors, rated by crowdworkers.","hours":"~5 h","size_hours":5,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"Open Database License (ODbL)","license_class":"share-alike","commercial_use":"yes","released":"2014-08","url":"https://github.com/CheyneyComputerScience/CREMA-D","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/crema-d"}
+{"slug":"msp-podcast","name":"MSP-Podcast","maintainer":"UT Dallas (MSP Lab)","tagline":"The largest naturalistic emotional speech corpus — podcast audio, continuously growing.","hours":"~230 h","size_hours":230,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"Academic licence (application required)","license_class":"gated","commercial_use":"restricted","released":"2017-01","url":"https://ecs.utdallas.edu/research/researchlabs/msp-lab/MSP-Podcast.html","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/msp-podcast"}
+{"slug":"meld","name":"MELD","maintainer":"DeCLaRe Lab / SUTD","tagline":"Emotion recognition in multi-party conversation, cut from a sitcom script.","hours":"~12 h","size_hours":12,"languages":["en"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"Not stated (repository code is GPL-3.0)","license_class":"custom","commercial_use":"restricted","released":"2018-10","url":"https://affective-meld.github.io/","paper_url":"https://arxiv.org/abs/1810.02508","page_url":"https://www.fullduplex.ai/datasets/meld"}
+{"slug":"esd","name":"ESD (Emotional Speech Dataset)","maintainer":"NUS / HLT Lab","tagline":"Parallel emotional speech in English and Mandarin from 20 speakers.","hours":"29 h","size_hours":29,"languages":["en","zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"Research use (see repository terms)","license_class":"custom","commercial_use":"restricted","released":"2021-05","url":"https://hltsingapore.github.io/ESD/","paper_url":"https://arxiv.org/abs/2105.14762","page_url":"https://www.fullduplex.ai/datasets/esd"}
+{"slug":"musan","name":"MUSAN","maintainer":"Johns Hopkins University","tagline":"Music, speech and noise — the corpus behind almost every augmentation recipe.","hours":"109 h","size_hours":109,"languages":["multi"],"speech":"human","speaker_consent":"public","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2015-10","url":"https://www.openslr.org/17/","paper_url":"https://arxiv.org/abs/1510.08484","page_url":"https://www.fullduplex.ai/datasets/musan"}
+{"slug":"wham","name":"WHAM! / WHAMR!","maintainer":"Whisper.ai / MERL","tagline":"Real ambient noise recorded in cafés and bars, paired with speech separation mixtures.","hours":"~80 h","size_hours":80,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-NC-4.0","license_class":"non-commercial","commercial_use":"no","released":"2019-07","url":"http://wham.whisper.ai/","paper_url":"https://arxiv.org/abs/1907.01160","page_url":"https://www.fullduplex.ai/datasets/wham"}
+{"slug":"demand","name":"DEMAND","maintainer":"Inria / University of Edinburgh","tagline":"Multi-channel recordings of 18 real acoustic environments.","hours":"~5 h","size_hours":5,"languages":["multi"],"speech":null,"speaker_consent":null,"full_duplex":false,"sample_rate_khz":48,"license":"CC-BY-SA-3.0","license_class":"share-alike","commercial_use":"yes","released":"2013-06","url":"https://zenodo.org/record/1227121","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/demand"}
+{"slug":"myst-children","name":"MyST Children's Conversational Speech","maintainer":"Boulder Learning / LDC","tagline":"Children talking to a virtual science tutor, the largest child-speech corpus.","hours":"393 h","size_hours":393,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-NC-SA-4.0 (non-commercial); commercial licence available","license_class":"non-commercial","commercial_use":"restricted","released":"2021-06","url":"https://catalog.ldc.upenn.edu/LDC2021S05","paper_url":"https://arxiv.org/abs/2309.13347","page_url":"https://www.fullduplex.ai/datasets/myst-children"}
+{"slug":"l2-arctic","name":"L2-ARCTIC","maintainer":"Texas A&M / Iowa State","tagline":"Non-native English from 24 speakers across six first languages, with error annotation.","hours":"27 h","size_hours":27,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":44.1,"license":"Research use (registration required)","license_class":"custom","commercial_use":"restricted","released":"2018-09","url":"https://psi.engr.tamu.edu/l2-arctic-corpus/","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/l2-arctic"}
+{"slug":"primock57","name":"PriMock57","maintainer":"Babylon Health","tagline":"57 mock primary-care consultations released for clinical speech research.","hours":"~9 h","size_hours":9,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2022-05","url":"https://github.com/babylonhealth/primock57","paper_url":"https://aclanthology.org/2022.acl-short.65/","page_url":"https://www.fullduplex.ai/datasets/primock57"}

data/models.jsonl

@@ -0,0 +1,57 @@
+{"slug":"tml-interaction-small","name":"TML-Interaction-Small","vendor":"Thinking Machines Lab","tagline":"276B-A12B mixture-of-experts interaction model with VAD-free, time-aligned 200 ms micro-turns across audio, video, and text — Thinking Machines Lab's first public voice release.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-05","first_response_ms":400,"open_weights":false,"license":null,"commercial_use":"unclear","languages":["en"],"status":"preview","benchmarks":["fdb-v15","audio-multichallenge"],"url":"https://thinkingmachines.ai/blog/interaction-models/","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/tml-interaction-small"}
+{"slug":"moshi","name":"Moshi","vendor":"Kyutai","tagline":"A 7B speech-text foundation model that generates user and assistant audio streams in parallel.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2024-09","first_response_ms":160,"open_weights":true,"license":"Apache-2.0 (code) / CC-BY-4.0 (weights)","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":["fdb-v1","big-bench-audio","talking-turns"],"url":"https://kyutai.org/2024/09/18/moshi-release.html","paper_url":"https://arxiv.org/abs/2410.00037","repo_url":"https://github.com/kyutai-labs/moshi","page_url":"https://www.fullduplex.ai/models/moshi"}
+{"slug":"glm-4-voice","name":"GLM-4-Voice","vendor":"Zhipu AI (Z.ai)","tagline":"An end-to-end Chinese/English speech LM that exposes tone, dialect, and emotion as instructions.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2024-10","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["zh","en"],"status":"current","benchmarks":[],"url":"https://github.com/zai-org/GLM-4-Voice","paper_url":"https://arxiv.org/abs/2412.02612","repo_url":"https://github.com/zai-org/GLM-4-Voice","page_url":"https://www.fullduplex.ai/models/glm-4-voice"}
+{"slug":"sesame-csm","name":"Sesame CSM","vendor":"Sesame AI","tagline":"The conversational speech model everyone called \"uncanny-valley-free.\"","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-03","first_response_ms":200,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://www.sesame.com/research/crossing_the_uncanny_valley_of_voice","paper_url":null,"repo_url":"https://github.com/SesameAILabs/csm","page_url":"https://www.fullduplex.ai/models/sesame-csm"}
+{"slug":"kimi-audio","name":"Kimi-Audio","vendor":"Moonshot AI","tagline":"A 7B audio foundation model covering ASR, AQA, audio QA, and end-to-end voice chat in one net.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-04","first_response_ms":null,"open_weights":true,"license":"Modified MIT","commercial_use":"yes","languages":["zh","en"],"status":"current","benchmarks":["voicebench","mmar","air-bench"],"url":"https://github.com/MoonshotAI/Kimi-Audio","paper_url":"https://arxiv.org/abs/2504.18425","repo_url":"https://github.com/MoonshotAI/Kimi-Audio","page_url":"https://www.fullduplex.ai/models/kimi-audio"}
+{"slug":"personaplex","name":"PersonaPlex","vendor":"NVIDIA (ADLR)","tagline":"7B full-duplex speech LM with hybrid voice + text prompts — any role, any voice, on top of Moshi.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-01","first_response_ms":170,"open_weights":true,"license":"MIT (code) / NVIDIA Open Model License (weights)","commercial_use":"research-only","languages":["en"],"status":"current","benchmarks":[],"url":"https://research.nvidia.com/labs/adlr/personaplex/","paper_url":"https://arxiv.org/abs/2602.06053","repo_url":"https://github.com/NVIDIA/PersonaPlex","page_url":"https://www.fullduplex.ai/models/personaplex"}
+{"slug":"minicpm-o-4-5","name":"MiniCPM-o 4.5","vendor":"OpenBMB / ModelBest","tagline":"9B on-device omni-modal LLM with full-duplex speech + vision streaming, targeting Gemini 2.5 Flash quality on a laptop.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-02","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh","+30"],"status":"current","benchmarks":["videofdb"],"url":"https://github.com/OpenBMB/MiniCPM-o","paper_url":"https://arxiv.org/abs/2604.27393","repo_url":"https://github.com/OpenBMB/MiniCPM-o","page_url":"https://www.fullduplex.ai/models/minicpm-o-4-5"}
+{"slug":"salmonn-omni","name":"SALMONN-omni","vendor":"ByteDance · Tsinghua · Cambridge","tagline":"First standalone codec-free full-duplex speech LLM — no audio tokens in the vocabulary, RL-trained for barge-in.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-11","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://github.com/bytedance/SALMONN","paper_url":"https://arxiv.org/abs/2505.17060","repo_url":"https://github.com/bytedance/SALMONN","page_url":"https://www.fullduplex.ai/models/salmonn-omni"}
+{"slug":"step-audio-2-mini","name":"Step-Audio 2 mini","vendor":"StepFun","tagline":"End-to-end multimodal audio LLM for speech conversation with expressive TTS and tool / search integration.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-08","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["zh","en"],"status":"current","benchmarks":["big-bench-audio","voicebench","uro-bench"],"url":"https://github.com/stepfun-ai/Step-Audio2","paper_url":"https://arxiv.org/abs/2507.16632","repo_url":"https://github.com/stepfun-ai/Step-Audio2","page_url":"https://www.fullduplex.ai/models/step-audio-2-mini"}
+{"slug":"covo-audio-chat","name":"Covo-Audio-Chat / Covo-Audio-Chat-FD","vendor":"Tencent","tagline":"7B end-to-end audio LLM with a native full-duplex variant — tri-modal speech-text interleaving and dual-stream listen-and-speak.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-03","first_response_ms":null,"open_weights":true,"license":"Tencent research license (non-commercial)","commercial_use":"research-only","languages":["zh","en"],"status":"preview","benchmarks":["uro-bench","voicebench"],"url":"https://github.com/Tencent/Covo-Audio","paper_url":"https://arxiv.org/abs/2602.09823","repo_url":"https://github.com/Tencent/Covo-Audio","page_url":"https://www.fullduplex.ai/models/covo-audio-chat"}
+{"slug":"llama-omni2","name":"LLaMA-Omni 2","vendor":"ICT / CAS (ictnlp)","tagline":"Qwen2.5-based real-time spoken chatbot series (0.5B–14B) with autoregressive streaming speech synthesis.","category":"speech-lm-fd","duplex_type":"half-duplex","full_duplex":false,"released":"2025-04","first_response_ms":583,"open_weights":true,"license":"Apache-2.0 (code) / Non-commercial (weights)","commercial_use":"non-commercial","languages":["en"],"status":"current","benchmarks":["voicebench"],"url":"https://github.com/ictnlp/LLaMA-Omni2","paper_url":"https://arxiv.org/abs/2505.02625","repo_url":"https://github.com/ictnlp/LLaMA-Omni2","page_url":"https://www.fullduplex.ai/models/llama-omni2"}
+{"slug":"realtime-venus","name":"Realtime-Venus","vendor":"Ant Group (inclusionAI)","tagline":"Two open 9B full-duplex models, one audio-visual and one speech-only, that keep talking while a background harness runs reasoning and tools.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-09","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh"],"status":"current","benchmarks":["fdb-v15"],"url":"https://realtime-venus.github.io/","paper_url":"https://arxiv.org/abs/2609.13814","repo_url":"https://github.com/inclusionAI/Realtime-Venus","page_url":"https://www.fullduplex.ai/models/realtime-venus"}
+{"slug":"duplexomni","name":"DuplexOmni","vendor":"Xi'an Jiaotong University / Peking University / Meituan","tagline":"Open full-duplex omni model that listens and watches in 480 ms slices, answers in speech in real time, and hands hard questions to a pluggable thinking layer.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-07","first_response_ms":506,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["zh","en"],"status":"preview","benchmarks":["big-bench-audio"],"url":"https://huggingface.co/MuyeHuang/DuplexOmni","paper_url":"https://arxiv.org/abs/2606.09186","repo_url":"https://github.com/MuyeHuang/DuplexOmni","page_url":"https://www.fullduplex.ai/models/duplexomni"}
+{"slug":"nemotronlabs-voicechat","name":"NVIDIA NemotronLabs VoiceChat 11B","vendor":"NVIDIA","tagline":"Open 11B full-duplex speech model on a Nemotron Nano v2 backbone, with a separate output channel for tool calls so it keeps talking while tools run.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-08","first_response_ms":448,"open_weights":true,"license":"OpenMDW-1.1","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":["fdb-v1","voicebench"],"url":"https://huggingface.co/nvidia/NVIDIA-NemotronLabs-VoiceChat-11B","paper_url":"https://arxiv.org/abs/2609.21967","repo_url":"https://github.com/NVIDIA-NeMo/Speech/tree/nemotron-labs-voicechat","page_url":"https://www.fullduplex.ai/models/nemotronlabs-voicechat"}
+{"slug":"raon-speechchat","name":"Raon-SpeechChat-9B","vendor":"KRAFTON","tagline":"Open 9B English full-duplex speech model with a dedicated backchannel token, speak-first or listen-first modes, and voice conditioning.","category":"speech-lm-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-04","first_response_ms":null,"open_weights":true,"license":"CC-BY-NC-4.0","commercial_use":"non-commercial","languages":["en"],"status":"current","benchmarks":[],"url":"https://huggingface.co/KRAFTON/Raon-SpeechChat-9B","paper_url":"https://arxiv.org/abs/2605.23912","repo_url":"https://github.com/krafton-ai/Raon-Speech","page_url":"https://www.fullduplex.ai/models/raon-speechchat"}
+{"slug":"openai-realtime","name":"OpenAI Realtime (gpt-realtime-2.1)","vendor":"OpenAI","tagline":"Realtime API reasoning voice model: gpt-realtime-2.1 updates gpt-realtime-2 with better alphanumeric recognition and steadier handling of silence, noise, and interruptions, plus a cheaper mini sibling.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-07","first_response_ms":310,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","es","fr","+50"],"status":"current","benchmarks":["big-bench-audio","fdb-v1","audio-multichallenge","scale-voice-showdown"],"url":"https://developers.openai.com/api/docs/models/gpt-realtime-2.1","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/openai-realtime"}
+{"slug":"gemini-live","name":"Gemini 2.5 Live","vendor":"Google DeepMind","tagline":"Legacy native-audio Live API model, still available; Google now recommends Gemini 3.8 Live as the replacement.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-12","first_response_ms":300,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","+21"],"status":"legacy","benchmarks":["big-bench-audio","talking-turns","videofdb"],"url":"https://ai.google.dev/gemini-api/docs/live","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gemini-live"}
+{"slug":"gemini-3-live","name":"Gemini 3.1 Flash Live Preview","vendor":"Google DeepMind","tagline":"Preview Live API snapshot on 3.1 Flash with proactive audio, native thinking, and tighter tool-call latency; Google now recommends Gemini 3.8 Live instead.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-03","first_response_ms":260,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","+21"],"status":"legacy","benchmarks":["big-bench-audio","talking-turns","videofdb"],"url":"https://ai.google.dev/gemini-api/docs/live","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gemini-3-live"}
+{"slug":"nova-2-sonic","name":"Amazon Nova 2 Sonic","vendor":"Amazon / AWS","tagline":"Bedrock-hosted GA speech-to-speech model — polyglot voices, 1M-token context, async tool calls, telephony-ready.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2025-12","first_response_ms":400,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","es","fr","de","it","pt","hi","ja","ko","+"],"status":"current","benchmarks":["big-bench-audio"],"url":"https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-amazon-nova-2-sonic.html","paper_url":"https://assets.amazon.science/c5/3d/84514a224666b5be6de4b43ef4aa/nova-2-0-technical-report2.pdf","repo_url":null,"page_url":"https://www.fullduplex.ai/models/nova-2-sonic"}
+{"slug":"nova-2-5-sonic","name":"Amazon Nova 2.5 Sonic","vendor":"Amazon / AWS","tagline":"Bedrock speech-to-speech model that succeeds Nova 2 Sonic at the same price, with claimed gains in reasoning, instruction following, and tool-calling accuracy.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-10","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-nova-2.5-sonic/","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/nova-2-5-sonic"}
+{"slug":"grok-voice-agent","name":"Grok Voice 3.0 (xAI Speech to Speech API)","vendor":"xAI","tagline":"xAI's speech-to-speech API, now on Grok Voice 3.0: 20+ languages, $0.08 per minute, wire-compatible with the OpenAI Realtime API.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-10","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","es","fr","de","zh","ja","pt","+13"],"status":"current","benchmarks":["big-bench-audio"],"url":"https://docs.x.ai/developers/model-capabilities/audio/voice-agent","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/grok-voice-agent"}
+{"slug":"gpt-live-1","name":"GPT-Live 1","vendor":"OpenAI","tagline":"OpenAI's full-duplex voice model: it listens and talks at the same time and hands reasoning and tool calls to a backend model or agent.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-07","first_response_ms":798,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":["fdb-v1","fdb-v15","fdb-v3","tau3-bench"],"url":"https://developers.openai.com/api/docs/models/gpt-live-1","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gpt-live-1"}
+{"slug":"gemini-3-8-live","name":"Gemini 3.8 Live","vendor":"Google DeepMind","tagline":"Google's GA Live API voice model: native audio in and out, camera and screen frames as input, and tool calls that run in the background while it keeps talking.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-09","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","+94"],"status":"current","benchmarks":[],"url":"https://ai.google.dev/gemini-api/docs/models/gemini-3.8-live","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gemini-3-8-live"}
+{"slug":"seedrealtime","name":"SeedRealtime","vendor":"ByteDance Seed","tagline":"ByteDance Seed's audio-visual full-duplex model: it watches a live camera feed, decides when to speak without an external VAD, and runs video calls in the Doubao app.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-08","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"unclear","languages":[],"status":"current","benchmarks":[],"url":"https://seed.bytedance.com/en/blog/seedrealtime-audio-visual-full-duplex-llm-released-toward-omni-modal-natural-interaction","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/seedrealtime"}
+{"slug":"eleven-realtime","name":"Eleven Realtime (v1 Mini / v1 Max)","vendor":"ElevenLabs","tagline":"ElevenLabs' speech-to-speech models, each a complete voice agent: Mini for low-cost conversation, Max for reasoning and multi-step tool use, behind an OpenAI-Realtime-compatible API.","category":"assistant-fd","duplex_type":"cascade","full_duplex":false,"released":"2026-10","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["90+"],"status":"current","benchmarks":["big-bench-audio"],"url":"https://elevenlabs.io/blog/realtime","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/eleven-realtime"}
+{"slug":"stepaudio-3-realtime","name":"StepAudio 3 Realtime","vendor":"StepFun","tagline":"StepFun's full-duplex realtime voice model on its open platform: the user and the model can speak at once, and it thinks while it speaks.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-09","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"unclear","languages":[],"status":"preview","benchmarks":[],"url":"https://platform.stepfun.ai/docs/en/guides/models/stepaudio-3-realtime","paper_url":"https://arxiv.org/abs/2609.14005","repo_url":null,"page_url":"https://www.fullduplex.ai/models/stepaudio-3-realtime"}
+{"slug":"qwen-audio-realtime","name":"Qwen-Audio Realtime (3.0 / 3.1 Plus)","vendor":"Alibaba Cloud","tagline":"Alibaba Cloud Model Studio's duplex speech models for voice conversation, served over WebSocket, AOQ, or WebRTC with system voices and voice cloning.","category":"assistant-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-08","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["zh","en","ja","ko","de","es","fr","id","it","pt","ru"],"status":"current","benchmarks":[],"url":"https://www.alibabacloud.com/help/en/model-studio/qwen-audio-realtime-user-guides","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/qwen-audio-realtime"}
+{"slug":"elevenlabs-agents","name":"ElevenLabs Agents (ElevenAgents)","vendor":"ElevenLabs","tagline":"The TTS incumbent's turnkey agent platform — turn-taking, RAG, and telephony pre-wired.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2025-05","first_response_ms":450,"open_weights":false,"license":null,"commercial_use":"yes","languages":["90+"],"status":"current","benchmarks":[],"url":"https://elevenlabs.io/conversational-ai","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/elevenlabs-agents"}
+{"slug":"vapi","name":"Vapi","vendor":"Vapi","tagline":"Developer-first orchestration layer for inbound/outbound voice agents with BYO models.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2023-07","first_response_ms":500,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","es","+100"],"status":"current","benchmarks":[],"url":"https://vapi.ai","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/vapi"}
+{"slug":"retell-ai","name":"Retell AI","vendor":"Retell","tagline":"Low-latency voice agent API built for call-centre workloads and SIP carriers.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2024-02","first_response_ms":500,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","es","+15"],"status":"current","benchmarks":[],"url":"https://retellai.com","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/retell-ai"}
+{"slug":"deepgram-voice-agent","name":"Deepgram Voice Agent","vendor":"Deepgram","tagline":"STT+TTS incumbent's agent API using Nova-3 / Flux for transcription and Aura-2 for voice.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2024-12","first_response_ms":450,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","+30"],"status":"current","benchmarks":[],"url":"https://deepgram.com/product/voice-agent-api","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/deepgram-voice-agent"}
+{"slug":"bland","name":"Bland","vendor":"Bland AI","tagline":"Turnkey voice-agent platform for inbound and outbound telephony with web agents and an API.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2023-11","first_response_ms":400,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","es","+30"],"status":"current","benchmarks":[],"url":"https://www.bland.ai","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/bland"}
+{"slug":"polyai-dialog-rsn-1","name":"PolyAI Dialog-RSN-1","vendor":"PolyAI","tagline":"Audio-in, text-out dialog model that folds turn-taking, speech recognition, reasoning, and tool calls into one model and leaves speech output to a separate TTS.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2026-07","first_response_ms":300,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://poly.ai/blog/PolyAI-dialog-rsn-1","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/polyai-dialog-rsn-1"}
+{"slug":"assemblyai-voice-agent-api","name":"AssemblyAI Voice Agent API","vendor":"AssemblyAI","tagline":"AssemblyAI's own speech understanding, LLM reasoning, and voice generation behind a single WebSocket, billed at $4.50 per hour.","category":"voice-agent-platform","duplex_type":"cascade","full_duplex":false,"released":"2026-04","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://www.assemblyai.com/blog/introducing-our-voice-agent-api","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/assemblyai-voice-agent-api"}
+{"slug":"livekit-agents","name":"LiveKit Agents","vendor":"LiveKit","tagline":"Open agent framework on top of the LiveKit WebRTC mesh — 50+ provider plugins, first-class FD.","category":"agent-framework","duplex_type":"cascade","full_duplex":false,"released":"2024-09","first_response_ms":null,"open_weights":false,"license":"Apache-2.0","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://github.com/livekit/agents","paper_url":null,"repo_url":"https://github.com/livekit/agents","page_url":"https://www.fullduplex.ai/models/livekit-agents"}
+{"slug":"pipecat","name":"Pipecat","vendor":"Daily","tagline":"Python framework for voice agents with swappable ASR / LLM / TTS and sane defaults for turn-taking.","category":"agent-framework","duplex_type":"cascade","full_duplex":false,"released":"2023-12","first_response_ms":null,"open_weights":false,"license":"BSD-2-Clause","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://pipecat.ai","paper_url":null,"repo_url":"https://github.com/pipecat-ai/pipecat","page_url":"https://www.fullduplex.ai/models/pipecat"}
+{"slug":"kyutai-unmute","name":"Kyutai Unmute","vendor":"Kyutai","tagline":"Open reference stack that gives any text LLM ears and a voice via delayed streams modelling.","category":"agent-framework","duplex_type":"cascade","full_duplex":false,"released":"2025-07","first_response_ms":null,"open_weights":true,"license":"MIT (code) / CC-BY-4.0 (weights)","commercial_use":"yes","languages":["en","fr"],"status":"current","benchmarks":[],"url":"https://github.com/kyutai-labs/unmute","paper_url":null,"repo_url":"https://github.com/kyutai-labs/unmute","page_url":"https://www.fullduplex.ai/models/kyutai-unmute"}
+{"slug":"hf-speech-to-speech","name":"HF speech-to-speech","vendor":"Hugging Face","tagline":"Modular reference pipeline that cascades VAD → ASR → LLM → TTS for local voice agents.","category":"agent-framework","duplex_type":"cascade","full_duplex":false,"released":"2024-08","first_response_ms":null,"open_weights":true,"license":"MIT","commercial_use":"yes","languages":["en","fr","es","zh","ja","ko"],"status":"current","benchmarks":[],"url":"https://github.com/huggingface/speech-to-speech","paper_url":null,"repo_url":"https://github.com/huggingface/speech-to-speech","page_url":"https://www.fullduplex.ai/models/hf-speech-to-speech"}
+{"slug":"qwen-audio-agent","name":"Qwen Audio Agent","vendor":"QwenAudio","tagline":"Apache-2.0 realtime voice runtime that keeps a conversation going while a backend agent runs tools and longer tasks.","category":"agent-framework","duplex_type":"cascade","full_duplex":false,"released":"2026-07","first_response_ms":null,"open_weights":false,"license":"Apache-2.0","commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://github.com/QwenAudio/qwen-audio-agent","paper_url":"https://arxiv.org/abs/2609.25195","repo_url":"https://github.com/QwenAudio/qwen-audio-agent","page_url":"https://www.fullduplex.ai/models/qwen-audio-agent"}
+{"slug":"qwen2-5-omni","name":"Qwen2.5-Omni","vendor":"Alibaba Cloud","tagline":"End-to-end multimodal Qwen (text / image / audio / video in, text + speech out) with streaming response.","category":"audio-llm","duplex_type":"half-duplex","full_duplex":false,"released":"2025-03","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh","ja","ko","+"],"status":"legacy","benchmarks":["air-bench","voicebench"],"url":"https://github.com/QwenLM/Qwen2.5-Omni","paper_url":"https://arxiv.org/abs/2503.20215","repo_url":"https://github.com/QwenLM/Qwen2.5-Omni","page_url":"https://www.fullduplex.ai/models/qwen2-5-omni"}
+{"slug":"qwen3-omni","name":"Qwen3-Omni","vendor":"Alibaba Cloud","tagline":"Natively end-to-end omni-LLM (text / image / audio / video in, text + speech out).","category":"audio-llm","duplex_type":"full-duplex","full_duplex":true,"released":"2025-09","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh","ja","+17"],"status":"current","benchmarks":["big-bench-audio","voicebench","mmar","air-bench"],"url":"https://github.com/QwenLM/Qwen3-Omni","paper_url":"https://arxiv.org/abs/2509.17765","repo_url":"https://github.com/QwenLM/Qwen3-Omni","page_url":"https://www.fullduplex.ai/models/qwen3-omni"}
+{"slug":"ultravox","name":"Ultravox","vendor":"Ultravox (Inworld)","tagline":"Lightweight audio adapter that bolts onto Llama / Qwen to give an LLM ears without retraining.","category":"audio-llm","duplex_type":"half-duplex","full_duplex":false,"released":"2024-08","first_response_ms":null,"open_weights":true,"license":"MIT","commercial_use":"yes","languages":["en"],"status":"current","benchmarks":[],"url":"https://www.ultravox.ai","paper_url":null,"repo_url":"https://github.com/fixie-ai/ultravox","page_url":"https://www.fullduplex.ai/models/ultravox"}
+{"slug":"cartesia-sonic","name":"Cartesia Sonic-3.6","vendor":"Cartesia","tagline":"State-space TTS with ~90 ms time-to-first-audio and realistic non-verbal sounds.","category":"tts","duplex_type":"other","full_duplex":false,"released":"2026-08","first_response_ms":90,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","+41"],"status":"current","benchmarks":[],"url":"https://cartesia.ai/blog/sonic-3.6/","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/cartesia-sonic"}
+{"slug":"cosyvoice-2","name":"CosyVoice 2","vendor":"Alibaba Tongyi","tagline":"Open expressive TTS with strong zero-shot voice cloning across Japanese, Chinese, and English.","category":"tts","duplex_type":"other","full_duplex":false,"released":"2024-12","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["zh","en","ja","ko"],"status":"current","benchmarks":[],"url":"https://funaudiollm.github.io/","paper_url":"https://arxiv.org/abs/2407.05407","repo_url":"https://github.com/FunAudioLLM/CosyVoice","page_url":"https://www.fullduplex.ai/models/cosyvoice-2"}
+{"slug":"qwen3-tts","name":"Qwen3-TTS","vendor":"Alibaba Cloud","tagline":"Open streaming TTS successor to CosyVoice with voice cloning, voice design, and 10-language coverage.","category":"tts","duplex_type":"other","full_duplex":false,"released":"2026-01","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh","ja","ko","es","fr","de","ar","ru","pt"],"status":"current","benchmarks":[],"url":"https://github.com/QwenLM/Qwen3-TTS","paper_url":null,"repo_url":"https://github.com/QwenLM/Qwen3-TTS","page_url":"https://www.fullduplex.ai/models/qwen3-tts"}
+{"slug":"moss-tts","name":"MOSS-TTS","vendor":"OpenMOSS (Fudan)","tagline":"Real-time streaming TTS family with voice cloning, multi-speaker dialogue, and sound generation.","category":"tts","duplex_type":"other","full_duplex":false,"released":"2026-01","first_response_ms":null,"open_weights":true,"license":"Apache-2.0","commercial_use":"yes","languages":["en","zh"],"status":"current","benchmarks":[],"url":"https://github.com/OpenMOSS/MOSS-TTS","paper_url":null,"repo_url":"https://github.com/OpenMOSS/MOSS-TTS","page_url":"https://www.fullduplex.ai/models/moss-tts"}
+{"slug":"gemini-3-8-flash-tts","name":"Gemini 3.8 Flash TTS / Flash-Lite TTS","vendor":"Google DeepMind","tagline":"Google's GA text-to-speech pair in the Gemini API: Flash TTS for studio-grade expressive voice, Flash-Lite TTS for high-volume and real-time voice-agent cascades.","category":"tts","duplex_type":"other","full_duplex":false,"released":"2026-09","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://ai.google.dev/gemini-api/docs/changelog","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gemini-3-8-flash-tts"}
+{"slug":"seamless-m4t-v2","name":"SeamlessM4T v2","vendor":"Meta AI","tagline":"Direct speech-to-speech translation across 100+ languages, with weights in the open.","category":"s2st","duplex_type":"other","full_duplex":false,"released":"2023-11","first_response_ms":null,"open_weights":true,"license":"CC-BY-NC-4.0","commercial_use":"non-commercial","languages":["100+"],"status":"current","benchmarks":[],"url":"https://ai.meta.com/research/seamless-communication/","paper_url":"https://arxiv.org/abs/2312.05187","repo_url":"https://github.com/facebookresearch/seamless_communication","page_url":"https://www.fullduplex.ai/models/seamless-m4t-v2"}
+{"slug":"hibiki","name":"Hibiki","vendor":"Kyutai","tagline":"On-device high-fidelity simultaneous speech-to-speech translation — Moshi-style dual-stream for FR → EN, with voice transfer.","category":"s2st","duplex_type":"full-duplex","full_duplex":true,"released":"2025-02","first_response_ms":null,"open_weights":true,"license":"CC-BY-4.0 (weights) / Apache-2.0 & MIT (code)","commercial_use":"yes","languages":["fr","en"],"status":"current","benchmarks":[],"url":"https://kyutai.org/2025/02/10/hibiki.html","paper_url":"https://arxiv.org/abs/2502.03382","repo_url":"https://github.com/kyutai-labs/hibiki","page_url":"https://www.fullduplex.ai/models/hibiki"}
+{"slug":"gpt-realtime-translate","name":"GPT-Realtime-Translate","vendor":"OpenAI","tagline":"Closed-API live speech-to-speech translation alongside the Realtime API GA — 70+ input languages into 13 output languages at $0.034 / minute.","category":"s2st","duplex_type":"full-duplex","full_duplex":true,"released":"2026-05","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","ja","zh","es","fr","+65"],"status":"current","benchmarks":[],"url":"https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api/","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gpt-realtime-translate"}
+{"slug":"tavus-griffin-lite","name":"Griffin-Lite","vendor":"Tavus","tagline":"Research preview of Tavus Griffin: one real-time model that watches and hears the user, decides every sub-second whether to speak, react, or wait, and renders its own voice and 720p face.","category":"video-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-10","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"unclear","languages":[],"status":"preview","benchmarks":["videofdb"],"url":"https://www.tavus.io/griffin","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/tavus-griffin-lite"}
+{"slug":"gemini-3-8-live-avatar","name":"Gemini 3.8 Live with Live Avatar","vendor":"Google DeepMind","tagline":"Gemini 3.8 Live sessions that return a 24 FPS avatar video stream alongside the model's own speech, generally available in Gemini Enterprise.","category":"video-fd","duplex_type":"full-duplex","full_duplex":true,"released":"2026-09","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":["en","+"],"status":"current","benchmarks":[],"url":"https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-8-live-with-live-avatar/","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/gemini-3-8-live-avatar"}
+{"slug":"tavus-cvi","name":"Tavus CVI (Phoenix-4.5 / Sparrow-2 / Raven-1)","vendor":"Tavus","tagline":"Tavus's production video-agent stack: camera perception, its own turn-taking model, STT, LLM, TTS, and a real-time face that keeps reacting while the user talks.","category":"video-fd","duplex_type":"cascade","full_duplex":false,"released":"2026-09","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":["turnbench"],"url":"https://docs.tavus.io/sections/conversational-video-interface/overview-cvi","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/tavus-cvi"}
+{"slug":"runway-characters","name":"Runway Characters","vendor":"Runway","tagline":"Real-time video agent API on Runway's GWM-1 world model: one image becomes a conversational character that keeps expressing while it speaks and listens.","category":"video-fd","duplex_type":"cascade","full_duplex":false,"released":"2026-03","first_response_ms":1750,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://runway.com/news/introducing-runway-characters","paper_url":null,"repo_url":null,"page_url":"https://www.fullduplex.ai/models/runway-characters"}
+{"slug":"heygen-liveavatar","name":"HeyGen LiveAvatar (FULL mode)","vendor":"HeyGen","tagline":"HeyGen's real-time avatar service; in FULL mode HeyGen runs speech detection, the LLM, TTS, and turn management, and switches the avatar into a listening state.","category":"video-fd","duplex_type":"cascade","full_duplex":false,"released":"2025-11","first_response_ms":null,"open_weights":false,"license":null,"commercial_use":"yes","languages":[],"status":"current","benchmarks":[],"url":"https://www.liveavatar.com/","paper_url":null,"repo_url":"https://github.com/heygen-com/liveavatar-web-sdk","page_url":"https://www.fullduplex.ai/models/heygen-liveavatar"}
+{"slug":"avtr-1","name":"AVTR-1","vendor":"Avaturn","tagline":"Open stack for real-time interactive avatars: a 153M-parameter motion model conditioned on both speakers' audio, so the face listens back while an external voice agent runs the turns.","category":"video-fd","duplex_type":"cascade","full_duplex":false,"released":"2026-05","first_response_ms":null,"open_weights":true,"license":"AVTR-1 Community License (weights) / PolyForm Noncommercial 1.0.0 (renderer, streamer)","commercial_use":"unclear","languages":[],"status":"current","benchmarks":[],"url":"https://avtr-1.avaturn.live/","paper_url":"https://arxiv.org/abs/2609.22913","repo_url":"https://github.com/avaturn-live/avtr-1","page_url":"https://www.fullduplex.ai/models/avtr-1"}
+{"slug":"avatar-forcing","name":"Avatar Forcing","vendor":"KAIST / NTU Singapore / DeepAuto.ai","tagline":"Research head avatar that takes the user's audio and head motion as live inputs and reacts to speech, nods, and laughter while it listens.","category":"video-fd","duplex_type":"cascade","full_duplex":false,"released":"2026-01","first_response_ms":null,"open_weights":true,"license":"CC BY-NC 4.0","commercial_use":"non-commercial","languages":[],"status":"current","benchmarks":[],"url":"https://taekyungki.github.io/AvatarForcing/","paper_url":"https://arxiv.org/abs/2601.00664","repo_url":"https://github.com/TaekyungKi/AvatarForcing","page_url":"https://www.fullduplex.ai/models/avatar-forcing"}