consome2 commited on
Commit
37daf11
·
verified ·
1 Parent(s): b6d3026

Weekly snapshot 2026-W47

Browse files
Files changed (4) hide show
  1. README.md +4 -4
  2. data/benchmarks.jsonl +48 -19
  3. data/datasets.jsonl +23 -4
  4. data/models.jsonl +0 -0
README.md CHANGED
@@ -25,11 +25,11 @@ The models, datasets and benchmarks for full-duplex and speech-to-speech voice A
25
 
26
  | Config | Rows | One row per |
27
  |---|---|---|
28
- | `models` | 70 | model, hosted API, voice-agent platform or framework |
29
- | `datasets` | 107 | speech dataset |
30
- | `benchmarks` | 64 | benchmark |
31
 
32
- Snapshot of 2026-11-09. The site is updated every week, so it can be newer than this file.
33
 
34
  ## Fields worth knowing
35
 
 
25
 
26
  | Config | Rows | One row per |
27
  |---|---|---|
28
+ | `models` | 204 | model, hosted API, voice-agent platform or framework |
29
+ | `datasets` | 126 | speech dataset |
30
+ | `benchmarks` | 93 | benchmark |
31
 
32
+ Snapshot of 2026-11-16. The site is updated every week, so it can be newer than this file.
33
 
34
  ## Fields worth knowing
35
 
data/benchmarks.jsonl CHANGED
@@ -1,13 +1,13 @@
1
  {"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":3,"updated":"2026-11","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"}
2
- {"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":5,"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"}
3
- {"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":null,"page_url":"https://www.fullduplex.ai/benchmarks/fdb-v15"}
4
- {"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"}
5
- {"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":2,"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"}
6
  {"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/leaderboard.html","page_url":"https://www.fullduplex.ai/benchmarks/videofdb"}
7
  {"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":8,"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"}
8
  {"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":2,"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"}
9
  {"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"}
10
- {"slug":"m3-duplexbench","name":"M3-DuplexBench","maintainer":"NTT (Fukuda et al.)","tagline":"English and Japanese full-duplex benchmark that scores interaction timing and response content together, in casual chat and multi-turn QA.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["turn-taking rate and latency","pause handling","backchannel and barge-in response","response content"],"models_scored":0,"updated":"2026-10","url":"https://github.com/nttcslab-sp/m3-duplexbench","paper_url":"https://arxiv.org/abs/2607.29125","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/m3-duplexbench"}
11
  {"slug":"duplex-mpe","name":"Duplex-MPE","maintainer":"Ma et al.","tagline":"Multi-party full-duplex test: with three or four people talking, should the assistant answer, stay silent or stop?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["response initiation","answer accuracy","silence preservation","stopping when resolved"],"models_scored":0,"updated":"2026-11","url":"https://step-out.github.io/Duplex-MPE-Page/","paper_url":"https://arxiv.org/abs/2609.31948","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplex-mpe"}
12
  {"slug":"duplexledger","name":"DuplexLedger","maintainer":"Li et al.","tagline":"Does a full-duplex agent remember the right things? Write, update, ignore and retain decisions during live interaction.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["latest-value accuracy","final-state fact coverage","stale assertions","distractor adoption"],"models_scored":0,"updated":"2026-10","url":"https://github.com/lyx-researcher/DuplexLedger","paper_url":"https://arxiv.org/abs/2610.27581","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplexledger"}
13
  {"slug":"interactionbench","name":"InteractionBench","maintainer":"Avera Labs (He et al.)","tagline":"Timing, spoken-task accuracy, whisper and volume, and groundedness over 20+ turns, measured on the same recorded user audio for every system.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex","speech-lm"],"metrics":["conversational timing","spoken-task accuracy","whisper and volume control","groundedness"],"models_scored":0,"updated":"2026-11","url":"https://avera-labs.github.io/InteractionBench/","paper_url":"https://arxiv.org/abs/2611.02430","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/interactionbench"}
@@ -15,13 +15,13 @@
15
  {"slug":"mp-bench","name":"MP-Bench","maintainer":"Shih et al.","tagline":"The voice agent is one participant in a spoken group conversation: does it follow who said what, and does it take its turn?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["turn-taking accuracy","response appropriateness","comprehension accuracy","latency"],"models_scored":0,"updated":"2026-10","url":"https://github.com/atosystem/MP-Bench","paper_url":"https://arxiv.org/abs/2609.13076","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mp-bench"}
16
  {"slug":"dsb-ifeval","name":"DuplexSpeechBench-IFEval","maintainer":"Mathur and Manocha","tagline":"Does a full-duplex agent infer from its persona when to speak, backchannel, interrupt or yield, and not only from explicit rules?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","instruction-following"],"metrics":["Instruction Adherence Score","Persona Adherence Score"],"models_scored":0,"updated":"2026-09","url":"https://dsb-ifeval.github.io/","paper_url":"https://arxiv.org/abs/2609.03423","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/dsb-ifeval"}
17
  {"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"}
18
- {"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":"https://matthewcym.github.io/VoiceBench/","page_url":"https://www.fullduplex.ai/benchmarks/voicebench"}
19
- {"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"}
20
- {"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"}
21
- {"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"}
22
- {"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":7,"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"}
23
  {"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"}
24
- {"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"}
25
  {"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"}
26
  {"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"}
27
  {"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"}
@@ -29,36 +29,65 @@
29
  {"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"}
30
  {"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"}
31
  {"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-V2","paper_url":null,"leaderboard_url":"https://huggingface.co/spaces/TTS-AGI/TTS-Arena-V2","page_url":"https://www.fullduplex.ai/benchmarks/tts-arena"}
32
- {"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":3,"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"}
33
  {"slug":"eva-bench","name":"EVA-Bench","maintainer":"ServiceNow","tagline":"End-to-end voice-agent evaluation with bot-to-bot calls, separate accuracy and experience scores, and accent and noise perturbations.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm"],"metrics":["EVA-A (accuracy)","EVA-X (experience)","pass@k and pass^k","accent and noise robustness"],"models_scored":1,"updated":"2026-10","url":"https://servicenow.github.io/eva/","paper_url":"https://arxiv.org/abs/2605.13841","leaderboard_url":"https://servicenow.github.io/eva/","page_url":"https://www.fullduplex.ai/benchmarks/eva-bench"}
34
  {"slug":"vamos-bench","name":"VAmoS Bench","maintainer":"Veris AI","tagline":"Simulated phone calls graded on what the voice agent said and on what it actually changed in the backend.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent"],"metrics":["task completion","latency","barge-ins per call","cost per call"],"models_scored":0,"updated":"2026-10","url":"https://veris.ai/leaderboard/vamos","paper_url":"https://arxiv.org/abs/2609.38512","leaderboard_url":"https://veris.ai/leaderboard/vamos-pro","page_url":"https://www.fullduplex.ai/benchmarks/vamos-bench"}
35
  {"slug":"apex-voice","name":"APEX-Voice","maintainer":"Mathur and Manocha","tagline":"Can a full-duplex voice agent finish professional work, such as form completion or negotiation, while the user interrupts and corrects?","full_duplex":true,"tier":"native","setting":"vertical","focus":["vertical-agent","full-duplex"],"metrics":["Pass@1","Reliable@3","artifact field accuracy"],"models_scored":0,"updated":"2026-09","url":"https://apex-voice.github.io/","paper_url":"https://arxiv.org/abs/2609.34973","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/apex-voice"}
36
  {"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"}
37
  {"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"}
38
  {"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"}
39
- {"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"}
40
- {"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"}
41
  {"slug":"nv-bench","name":"NV-Bench","maintainer":"Ni et al.","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":"2026-03","url":"https://arxiv.org/abs/2603.15352","paper_url":"https://arxiv.org/abs/2603.15352","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/nv-bench"}
42
  {"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"}
43
- {"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"}
44
  {"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"}
45
  {"slug":"sid-bench","name":"SID-Bench","maintainer":"Xia et al.","tagline":"Semantic-aware interruption detection, built from real human dialogues and scored with a time penalty for both false alarms and late responses.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["Average Penalty Time (APT)","false alarms","response delay"],"models_scored":0,"updated":"2026-03","url":"https://github.com/xkx-hub/SID-bench","paper_url":"https://arxiv.org/abs/2603.24144","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/sid-bench"}
46
  {"slug":"fd-bench","name":"FD-Bench","maintainer":"Peng et al.","tagline":"A pipeline of LLM, TTS and ASR that simulates users and tests how full-duplex systems handle interruptions, delays and noise.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["interruption response","delay","robustness under noise"],"models_scored":0,"updated":"2025-07","url":"https://github.com/pengyizhou/FD-Bench","paper_url":"https://arxiv.org/abs/2507.19040","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fd-bench"}
47
  {"slug":"mtr-duplex-bench","name":"MTR-DuplexBench","maintainer":"Zhang et al.","tagline":"Multi-round benchmark that splits continuous full-duplex dialogue into turns and scores each round.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["conversational features","dialogue quality","instruction following","safety"],"models_scored":0,"updated":"2026-04","url":"https://github.com/ZhangHe0918/MTR-DuplexBench","paper_url":"https://arxiv.org/abs/2511.10262","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mtr-duplex-bench"}
48
- {"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://aslp-lab.github.io/HumDial-Challenge/","paper_url":"https://arxiv.org/abs/2601.05564","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/humdial"}
49
  {"slug":"flexi","name":"FLEXI","maintainer":"Ge et al.","tagline":"Full-duplex benchmark that includes the model interrupting the user in emergencies, across six interaction scenarios.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["latency","response quality","emergency interruption"],"models_scored":0,"updated":"2025-09","url":"https://github.com/ChristineCHEN274/FLEXI","paper_url":"https://arxiv.org/abs/2509.22243","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/flexi"}
50
  {"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"}
51
  {"slug":"dynamic-superb","name":"Dynamic-SUPERB","maintainer":"NTU et al.","tagline":"Instruction-tuned successor to SUPERB, with 180 tasks in Phase-2 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"}
52
  {"slug":"voiceassistant-eval","name":"VoiceAssistant-Eval","maintainer":"Wang et al.","tagline":"10,497 examples in 13 task categories that test voice assistants on listening, speaking and viewing.","full_duplex":false,"tier":"component","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["content quality","speech quality","consistency"],"models_scored":0,"updated":"2025-09","url":"https://mathllm.github.io/VoiceAssistantEval/","paper_url":"https://arxiv.org/abs/2509.22651","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceassistant-eval"}
53
- {"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"}
54
  {"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"}
55
  {"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://github.com/SJTU-OmniAgent/VocalBench-zh","paper_url":"https://arxiv.org/abs/2511.08230","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/vocalbench-zh"}
56
  {"slug":"cs3-bench","name":"CS3-Bench","maintainer":"Academic (CS3-Bench authors)","tagline":"Mandarin-English code-switching benchmark: across seven speech-to-speech models, knowledge-intensive question answering drops by up to 66% relative on code-switched input.","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://huggingface.co/datasets/VocalNet/CS3-Bench","paper_url":"https://arxiv.org/abs/2510.07881","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/cs3-bench"}
57
- {"slug":"j-moshi","name":"J-Moshi (subjective)","maintainer":"Ohashi et al.","tagline":"The first public Japanese full-duplex dialogue model, built on Moshi and rated by human listeners for naturalness and meaningfulness.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["naturalness","meaningfulness"],"models_scored":0,"updated":"2025-06","url":"https://nu-dialogue.github.io/j-moshi","paper_url":"https://arxiv.org/abs/2506.02979","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/j-moshi"}
58
  {"slug":"voiceagentbench","name":"VoiceAgentBench","maintainer":"Academic (VoiceAgentBench authors)","tagline":"Spoken agentic benchmark: 6,000+ synthetic spoken queries covering single-tool calls, multi-tool workflows, multi-turn dialogue and safety, in English and six Indic languages.","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":"2026-02","url":"https://github.com/ola-krutrim/VoiceAgentBench","paper_url":"https://arxiv.org/abs/2510.07978","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceagentbench"}
59
- {"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://labs.scale.com/papers/audiomc","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"}
60
  {"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"}
61
  {"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://github.com/Kinetics-JOJO/MMedFD","paper_url":"https://arxiv.org/abs/2509.19817","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mmedfd"}
62
  {"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"}
63
  {"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":"2026-08","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"}
64
  {"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://flaai.github.io/AIA_page/","paper_url":"https://arxiv.org/abs/2605.18168","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/aia-acoustic-interference"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  {"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":3,"updated":"2026-11","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"}
2
+ {"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":14,"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"}
3
+ {"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":6,"updated":"2025-08","url":"https://github.com/DanielLin94144/Full-Duplex-Bench","paper_url":"https://arxiv.org/abs/2507.23159","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fdb-v15"}
4
+ {"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":2,"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"}
5
+ {"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":6,"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"}
6
  {"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/leaderboard.html","page_url":"https://www.fullduplex.ai/benchmarks/videofdb"}
7
  {"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":8,"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"}
8
  {"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":2,"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"}
9
  {"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"}
10
+ {"slug":"m3-duplexbench","name":"M3-DuplexBench","maintainer":"NTT (Fukuda et al.)","tagline":"English and Japanese full-duplex benchmark that scores interaction timing and response content together, in casual chat and multi-turn QA.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["turn-taking rate and latency","pause handling","backchannel and barge-in response","response content"],"models_scored":2,"updated":"2026-10","url":"https://github.com/nttcslab-sp/m3-duplexbench","paper_url":"https://arxiv.org/abs/2607.29125","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/m3-duplexbench"}
11
  {"slug":"duplex-mpe","name":"Duplex-MPE","maintainer":"Ma et al.","tagline":"Multi-party full-duplex test: with three or four people talking, should the assistant answer, stay silent or stop?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["response initiation","answer accuracy","silence preservation","stopping when resolved"],"models_scored":0,"updated":"2026-11","url":"https://step-out.github.io/Duplex-MPE-Page/","paper_url":"https://arxiv.org/abs/2609.31948","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplex-mpe"}
12
  {"slug":"duplexledger","name":"DuplexLedger","maintainer":"Li et al.","tagline":"Does a full-duplex agent remember the right things? Write, update, ignore and retain decisions during live interaction.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["latest-value accuracy","final-state fact coverage","stale assertions","distractor adoption"],"models_scored":0,"updated":"2026-10","url":"https://github.com/lyx-researcher/DuplexLedger","paper_url":"https://arxiv.org/abs/2610.27581","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplexledger"}
13
  {"slug":"interactionbench","name":"InteractionBench","maintainer":"Avera Labs (He et al.)","tagline":"Timing, spoken-task accuracy, whisper and volume, and groundedness over 20+ turns, measured on the same recorded user audio for every system.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex","speech-lm"],"metrics":["conversational timing","spoken-task accuracy","whisper and volume control","groundedness"],"models_scored":0,"updated":"2026-11","url":"https://avera-labs.github.io/InteractionBench/","paper_url":"https://arxiv.org/abs/2611.02430","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/interactionbench"}
 
15
  {"slug":"mp-bench","name":"MP-Bench","maintainer":"Shih et al.","tagline":"The voice agent is one participant in a spoken group conversation: does it follow who said what, and does it take its turn?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["turn-taking accuracy","response appropriateness","comprehension accuracy","latency"],"models_scored":0,"updated":"2026-10","url":"https://github.com/atosystem/MP-Bench","paper_url":"https://arxiv.org/abs/2609.13076","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mp-bench"}
16
  {"slug":"dsb-ifeval","name":"DuplexSpeechBench-IFEval","maintainer":"Mathur and Manocha","tagline":"Does a full-duplex agent infer from its persona when to speak, backchannel, interrupt or yield, and not only from explicit rules?","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","instruction-following"],"metrics":["Instruction Adherence Score","Persona Adherence Score"],"models_scored":0,"updated":"2026-09","url":"https://dsb-ifeval.github.io/","paper_url":"https://arxiv.org/abs/2609.03423","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/dsb-ifeval"}
17
  {"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"}
18
+ {"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":24,"updated":"2025-06","url":"https://github.com/MatthewCYM/VoiceBench","paper_url":"https://arxiv.org/abs/2410.17196","leaderboard_url":"https://matthewcym.github.io/VoiceBench/","page_url":"https://www.fullduplex.ai/benchmarks/voicebench"}
19
+ {"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":1,"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"}
20
+ {"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":5,"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"}
21
+ {"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":4,"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"}
22
+ {"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":12,"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"}
23
  {"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"}
24
+ {"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":5,"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"}
25
  {"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"}
26
  {"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"}
27
  {"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"}
 
29
  {"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"}
30
  {"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"}
31
  {"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-V2","paper_url":null,"leaderboard_url":"https://huggingface.co/spaces/TTS-AGI/TTS-Arena-V2","page_url":"https://www.fullduplex.ai/benchmarks/tts-arena"}
32
+ {"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":5,"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"}
33
  {"slug":"eva-bench","name":"EVA-Bench","maintainer":"ServiceNow","tagline":"End-to-end voice-agent evaluation with bot-to-bot calls, separate accuracy and experience scores, and accent and noise perturbations.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm"],"metrics":["EVA-A (accuracy)","EVA-X (experience)","pass@k and pass^k","accent and noise robustness"],"models_scored":1,"updated":"2026-10","url":"https://servicenow.github.io/eva/","paper_url":"https://arxiv.org/abs/2605.13841","leaderboard_url":"https://servicenow.github.io/eva/","page_url":"https://www.fullduplex.ai/benchmarks/eva-bench"}
34
  {"slug":"vamos-bench","name":"VAmoS Bench","maintainer":"Veris AI","tagline":"Simulated phone calls graded on what the voice agent said and on what it actually changed in the backend.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent"],"metrics":["task completion","latency","barge-ins per call","cost per call"],"models_scored":0,"updated":"2026-10","url":"https://veris.ai/leaderboard/vamos","paper_url":"https://arxiv.org/abs/2609.38512","leaderboard_url":"https://veris.ai/leaderboard/vamos-pro","page_url":"https://www.fullduplex.ai/benchmarks/vamos-bench"}
35
  {"slug":"apex-voice","name":"APEX-Voice","maintainer":"Mathur and Manocha","tagline":"Can a full-duplex voice agent finish professional work, such as form completion or negotiation, while the user interrupts and corrects?","full_duplex":true,"tier":"native","setting":"vertical","focus":["vertical-agent","full-duplex"],"metrics":["Pass@1","Reliable@3","artifact field accuracy"],"models_scored":0,"updated":"2026-09","url":"https://apex-voice.github.io/","paper_url":"https://arxiv.org/abs/2609.34973","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/apex-voice"}
36
  {"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"}
37
  {"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"}
38
  {"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"}
39
+ {"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":5,"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"}
40
+ {"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":3,"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"}
41
  {"slug":"nv-bench","name":"NV-Bench","maintainer":"Ni et al.","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":"2026-03","url":"https://arxiv.org/abs/2603.15352","paper_url":"https://arxiv.org/abs/2603.15352","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/nv-bench"}
42
  {"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"}
43
+ {"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":13,"updated":"2024-01","url":"https://www.openslr.org/12","paper_url":null,"leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/librispeech"}
44
  {"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"}
45
  {"slug":"sid-bench","name":"SID-Bench","maintainer":"Xia et al.","tagline":"Semantic-aware interruption detection, built from real human dialogues and scored with a time penalty for both false alarms and late responses.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["Average Penalty Time (APT)","false alarms","response delay"],"models_scored":0,"updated":"2026-03","url":"https://github.com/xkx-hub/SID-bench","paper_url":"https://arxiv.org/abs/2603.24144","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/sid-bench"}
46
  {"slug":"fd-bench","name":"FD-Bench","maintainer":"Peng et al.","tagline":"A pipeline of LLM, TTS and ASR that simulates users and tests how full-duplex systems handle interruptions, delays and noise.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["interruption response","delay","robustness under noise"],"models_scored":0,"updated":"2025-07","url":"https://github.com/pengyizhou/FD-Bench","paper_url":"https://arxiv.org/abs/2507.19040","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/fd-bench"}
47
  {"slug":"mtr-duplex-bench","name":"MTR-DuplexBench","maintainer":"Zhang et al.","tagline":"Multi-round benchmark that splits continuous full-duplex dialogue into turns and scores each round.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["conversational features","dialogue quality","instruction following","safety"],"models_scored":0,"updated":"2026-04","url":"https://github.com/ZhangHe0918/MTR-DuplexBench","paper_url":"https://arxiv.org/abs/2511.10262","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mtr-duplex-bench"}
48
+ {"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":1,"updated":"2026-01","url":"https://aslp-lab.github.io/HumDial-Challenge/","paper_url":"https://arxiv.org/abs/2601.05564","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/humdial"}
49
  {"slug":"flexi","name":"FLEXI","maintainer":"Ge et al.","tagline":"Full-duplex benchmark that includes the model interrupting the user in emergencies, across six interaction scenarios.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["latency","response quality","emergency interruption"],"models_scored":0,"updated":"2025-09","url":"https://github.com/ChristineCHEN274/FLEXI","paper_url":"https://arxiv.org/abs/2509.22243","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/flexi"}
50
  {"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"}
51
  {"slug":"dynamic-superb","name":"Dynamic-SUPERB","maintainer":"NTU et al.","tagline":"Instruction-tuned successor to SUPERB, with 180 tasks in Phase-2 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"}
52
  {"slug":"voiceassistant-eval","name":"VoiceAssistant-Eval","maintainer":"Wang et al.","tagline":"10,497 examples in 13 task categories that test voice assistants on listening, speaking and viewing.","full_duplex":false,"tier":"component","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["content quality","speech quality","consistency"],"models_scored":0,"updated":"2025-09","url":"https://mathllm.github.io/VoiceAssistantEval/","paper_url":"https://arxiv.org/abs/2509.22651","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceassistant-eval"}
53
+ {"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":1,"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"}
54
  {"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"}
55
  {"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://github.com/SJTU-OmniAgent/VocalBench-zh","paper_url":"https://arxiv.org/abs/2511.08230","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/vocalbench-zh"}
56
  {"slug":"cs3-bench","name":"CS3-Bench","maintainer":"Academic (CS3-Bench authors)","tagline":"Mandarin-English code-switching benchmark: across seven speech-to-speech models, knowledge-intensive question answering drops by up to 66% relative on code-switched input.","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://huggingface.co/datasets/VocalNet/CS3-Bench","paper_url":"https://arxiv.org/abs/2510.07881","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/cs3-bench"}
57
+ {"slug":"j-moshi","name":"J-Moshi (subjective)","maintainer":"Ohashi et al.","tagline":"The first public Japanese full-duplex dialogue model, built on Moshi and rated by human listeners for naturalness and meaningfulness.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","speech-lm"],"metrics":["naturalness","meaningfulness"],"models_scored":1,"updated":"2025-06","url":"https://nu-dialogue.github.io/j-moshi","paper_url":"https://arxiv.org/abs/2506.02979","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/j-moshi"}
58
  {"slug":"voiceagentbench","name":"VoiceAgentBench","maintainer":"Academic (VoiceAgentBench authors)","tagline":"Spoken agentic benchmark: 6,000+ synthetic spoken queries covering single-tool calls, multi-tool workflows, multi-turn dialogue and safety, in English and six Indic languages.","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":"2026-02","url":"https://github.com/ola-krutrim/VoiceAgentBench","paper_url":"https://arxiv.org/abs/2510.07978","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voiceagentbench"}
59
+ {"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":4,"updated":"2026-10","url":"https://labs.scale.com/papers/audiomc","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"}
60
  {"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"}
61
  {"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://github.com/Kinetics-JOJO/MMedFD","paper_url":"https://arxiv.org/abs/2509.19817","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/mmedfd"}
62
  {"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"}
63
  {"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":"2026-08","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"}
64
  {"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://flaai.github.io/AIA_page/","paper_url":"https://arxiv.org/abs/2605.18168","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/aia-acoustic-interference"}
65
+ {"slug":"omni-duplexeval","name":"Omni-DuplexEval","maintainer":"OpenBMB (Tsinghua University / Shanghai Qi Zhi Institute / ModelBest)","tagline":"Streaming video benchmark that scores what a duplex omni-modal model says and when it says it, in live description and proactive reminder tasks.","full_duplex":true,"tier":"adjacent","setting":"lab","focus":["full-duplex","audio-understanding"],"metrics":["overall score (0 to 100)","Content Consistency","Temporal Sensitivity","Proactive Reminder score"],"models_scored":0,"updated":"2026-07","url":"https://github.com/OpenBMB/Omni-DuplexEval","paper_url":"https://arxiv.org/abs/2605.17360","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/omni-duplexeval"}
66
+ {"slug":"omniinteract","name":"OmniInteract","maintainer":"CUHK MMLab / SJTU / NTU / McMaster / CityUHK / JUFE","tagline":"Streaming audio-visual benchmark where spoken queries sit in the audio track and the model must decide whether, when and what to answer.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex","audio-understanding"],"metrics":["IA-QTF1 (Interaction-Aware Quality-Timeliness F1)","Interruption Diagnostic Suite (NOR, PAQ, CSM)","Nested Chain Completion Score"],"models_scored":0,"updated":"2026-05","url":"https://github.com/Lucky-Lance/OmniInteract","paper_url":"https://arxiv.org/abs/2605.26485","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/omniinteract"}
67
+ {"slug":"provoice-bench","name":"ProVoice-Bench","maintainer":"Shanghai Jiao Tong University (Xu et al.)","tagline":"Tests whether a voice agent stays silent or steps in unprompted, across four proactive tasks built from synthesized conversations and device context.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding"],"metrics":["trigger accuracy","recall","false-positive rate","response accuracy (R_acc)"],"models_scored":0,"updated":"2026-06","url":"https://huggingface.co/datasets/overji/ProVoice-Bench","paper_url":"https://arxiv.org/abs/2604.15037","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/provoice-bench"}
68
+ {"slug":"eot-bench","name":"eot-bench","maintainer":"LiveKit","tagline":"End-of-turn detection scored at real pauses in human-to-agent turns across 14 languages, as a trade-off between false cutoffs and response delay.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["false cutoffs at 300 ms and 600 ms latency budgets","latency at 5% and 10% false-cutoff budgets","latency and false-cutoff Pareto frontier"],"models_scored":0,"updated":"2026-11","url":"https://github.com/livekit/eot-bench","paper_url":null,"leaderboard_url":"https://livekit.com/benchmarks/eot-bench","page_url":"https://www.fullduplex.ai/benchmarks/eot-bench"}
69
+ {"slug":"audio2tool","name":"Audio2Tool","maintainer":"Rivian and Volkswagen Group Technologies","tagline":"Spoken tool-calling benchmark for smart car, smart home and wearable assistants, in eight tiers from direct commands to overlapping speakers.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","vertical-agent"],"metrics":["tool accuracy","exact match","parameter F1"],"models_scored":0,"updated":"2026-09","url":"https://audio2tool.github.io/","paper_url":"https://arxiv.org/abs/2604.22821","leaderboard_url":"https://huggingface.co/spaces/RVtech/Audio2ToolLeaderboard","page_url":"https://www.fullduplex.ai/benchmarks/audio2tool"}
70
+ {"slug":"c3-benchmark","name":"C3 Benchmark","maintainer":"Ma et al.","tagline":"English and Chinese test of whether spoken dialogue models handle ambiguity, omission, coreference and multi-turn context in speech.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["accuracy (LLM-judged)","per-phenomenon accuracy"],"models_scored":0,"updated":"2025-10","url":"https://step-out.github.io/C3-web/","paper_url":"https://arxiv.org/abs/2507.22968","leaderboard_url":"https://step-out.github.io/C3-web/","page_url":"https://www.fullduplex.ai/benchmarks/c3-benchmark"}
71
+ {"slug":"cekura-s2s-agent-bench","name":"Cekura Speech-to-Speech Agent Benchmark","maintainer":"Cekura","tagline":"Vendor-run leaderboard that swaps realtime speech models into one voice agent and scores 82 clinic and Medicare calls, three runs each.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm"],"metrics":["pass³ (all three runs pass)","success rate","data accuracy","median response time"],"models_scored":0,"updated":"2026-11","url":"https://benchmarks.cekura.ai/speech-to-speech","paper_url":null,"leaderboard_url":"https://benchmarks.cekura.ai/speech-to-speech","page_url":"https://www.fullduplex.ai/benchmarks/cekura-s2s-agent-bench"}
72
+ {"slug":"cekura-turn-bench","name":"Cekura Turn-Taking Benchmark","maintainer":"Cekura","tagline":"Scripted calls that test whether realtime speech models wait through pauses, reply at turn ends, stop on cut-ins and keep talking through backchannels.","full_duplex":true,"tier":"native","setting":"live","focus":["full-duplex"],"metrics":["pause handling","turn taking","barge-in yield","backchannel continue rate","end-of-turn latency"],"models_scored":0,"updated":"2026-11","url":"https://benchmarks.cekura.ai/speech-to-speech/turn-bench","paper_url":null,"leaderboard_url":"https://benchmarks.cekura.ai/speech-to-speech/turn-bench","page_url":"https://www.fullduplex.ai/benchmarks/cekura-turn-bench"}
73
+ {"slug":"codett","name":"CoDeTT","maintainer":"Bairong (BRVoice team)","tagline":"Turn-taking as a four-way decision with 14 labelled intents, so a model that picks the right action for the wrong reason is caught.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex"],"metrics":["4-action accuracy (ACC)","Semantic Misalignment Rate (SMR)"],"models_scored":0,"updated":"2026-04","url":"https://huggingface.co/datasets/YingaoWang-casia/CoDeTT","paper_url":"https://arxiv.org/abs/2603.25434","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/codett"}
74
+ {"slug":"contextdialog","name":"ContextDialog","maintainer":"Seoul National University (Kim et al.)","tagline":"Spoken questions about something said earlier in a conversation, testing whether voice interaction models recall and use past turns.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["GPT Score (five-point)","word error rate"],"models_scored":0,"updated":"2025-05","url":"https://huggingface.co/datasets/ContextDialog/ContextDialog","paper_url":"https://arxiv.org/abs/2502.19759","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/contextdialog"}
75
+ {"slug":"coval-benchmarks","name":"Coval Benchmarks","maintainer":"Coval","tagline":"Continuously run vendor boards for speech-to-text, text-to-speech and speech-to-speech models, with an open-source runner for the first two and a dated methodology.","full_duplex":false,"tier":"component","setting":"live","focus":["asr","tts","speech-lm"],"metrics":["WER","TTFS","TTFA","voice-to-voice latency","instruction adherence"],"models_scored":0,"updated":"2026-11","url":"https://benchmarks.coval.ai/overview","paper_url":null,"leaderboard_url":"https://benchmarks.coval.ai/overview","page_url":"https://www.fullduplex.ai/benchmarks/coval-benchmarks"}
76
+ {"slug":"duplexworld","name":"DuplexWorld","maintainer":"Centific / University of Maryland","tagline":"Speech-to-speech voice agents on simulated telephone tasks in six worlds, scored on task completion, conversational dynamics and speech naturalness.","full_duplex":true,"tier":"native","setting":"vertical","focus":["vertical-agent","full-duplex"],"metrics":["Pass@1","goal-state check","turn-taking","selectivity","DNSMOS"],"models_scored":0,"updated":"2026-08","url":"https://huggingface.co/datasets/CentificAIResearch/Duplex-World","paper_url":"https://arxiv.org/abs/2608.10716","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/duplexworld"}
77
+ {"slug":"game-time","name":"Game-Time Benchmark","maintainer":"MIT / NTU / Academia Sinica","tagline":"Tests whether full-duplex spoken language models can follow spoken instructions about timing, tempo and simultaneous speech.","full_duplex":true,"tier":"native","setting":"lab","focus":["full-duplex","instruction-following"],"metrics":["instruction following (1 to 5)","response appropriateness (1 to 5)","turn-taking (1 to 5)"],"models_scored":0,"updated":"2026-06","url":"https://github.com/ga642381/Game-Time-Benchmark","paper_url":"https://arxiv.org/abs/2509.26388","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/game-time"}
78
+ {"slug":"ihbench","name":"IHBench","maintainer":"Boson AI","tagline":"Scores what a voice agent says after the user interrupts it in the middle of a structured workflow.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","vertical-agent"],"metrics":["task fulfillment win rate","recovery quality pass rate"],"models_scored":0,"updated":"2026-06","url":"https://huggingface.co/datasets/bosonai/IHBench","paper_url":"https://arxiv.org/abs/2606.19595","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/ihbench"}
79
+ {"slug":"kvoicebench","name":"KVoiceBench","maintainer":"KRAFTON","tagline":"Korean spoken question answering benchmark for speech language models, transferred from VoiceBench with human-agent translation and normalization rulebooks.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["accuracy","judge score (open-ended)","prompt and instruction accuracy (KIFEval)","refusal rate (KAdvBench)"],"models_scored":0,"updated":"2026-08","url":"https://huggingface.co/datasets/KRAFTON/KVoiceBench","paper_url":"https://arxiv.org/abs/2605.27984","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/kvoicebench"}
80
+ {"slug":"mivas-bench","name":"MIVAS Bench","maintainer":"Bluejay","tagline":"Speech-to-speech models placed inside multi-agent call environments for healthcare, legal services and customer support, scored by deterministic verifiers.","full_duplex":false,"tier":"native","setting":"vertical","focus":["vertical-agent","speech-lm"],"metrics":["Pass^5","Pass@1","tool adherence","handoff adherence","final database state"],"models_scored":0,"updated":"2026-11","url":"https://github.com/bluejay-ai-dev/mivas-bench","paper_url":null,"leaderboard_url":"https://research.getbluejay.ai/benchmarks/mivas/leaderboard","page_url":"https://www.fullduplex.ai/benchmarks/mivas-bench"}
81
+ {"slug":"openaudiobench","name":"OpenAudioBench","maintainer":"Baichuan Inc.","tagline":"Spoken question answering set from Baichuan covering logical reasoning, general knowledge and open-ended questions, judged by GPT-4o.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["accuracy (Llama Questions, Web Questions, TriviaQA)","score (Reasoning QA, AlpacaEval)"],"models_scored":0,"updated":"2025-01","url":"https://huggingface.co/datasets/baichuan-inc/OpenAudioBench","paper_url":"https://arxiv.org/abs/2501.15368","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/openaudiobench"}
82
+ {"slug":"rolebreak","name":"RoleBreak","maintainer":"University of Hong Kong / Kami AI","tagline":"Tests whether speech-to-speech models keep a persona, stay safe and express emotion over role-play conversations of about 22 turns.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","emotion"],"metrics":["persona pass rate","interaction pass rate","safety pass rate","first-failure turn (P-FFT, S-FFT)","naturalness","emotion"],"models_scored":0,"updated":"2026-09","url":"https://huggingface.co/datasets/Greenbean/RoleBreak","paper_url":"https://arxiv.org/abs/2609.16614","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/rolebreak"}
83
+ {"slug":"s2s-arena","name":"S2S-Arena","maintainer":"FreedomIntelligence / CUHK-SZ / Shenzhen University of Advanced Technology","tagline":"Arena-style pairwise comparison of speech-to-speech models on spoken instructions that depend on paralinguistic cues, judged directly on audio.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following","emotion"],"metrics":["Elo","win rate"],"models_scored":0,"updated":"2026-06","url":"https://github.com/FreedomIntelligence/S2S-Arena","paper_url":"https://arxiv.org/abs/2503.05085","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/s2s-arena"}
84
+ {"slug":"speechinstructbench","name":"SpeechInstructBench","maintainer":"CUHK / Alibaba Group","tagline":"Spoken instruction-following benchmark in English and Chinese with closed-ended, open-ended and adjustment tasks under accent, noise and disfluency variants.","full_duplex":false,"tier":"native","setting":"lab","focus":["instruction-following","speech-lm"],"metrics":["prompt-level accuracy","instruction-level accuracy","IAR","ECR"],"models_scored":0,"updated":"2026-01","url":"https://github.com/dingdongwang/SpeechInstructBench","paper_url":"https://arxiv.org/abs/2503.02769","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/speechinstructbench"}
85
+ {"slug":"steerbench","name":"SteerBench","maintainer":"Scale AI / University of Maryland","tagline":"Tests whether a speech-to-speech assistant changes its tone, persona, style or accent, and speed or length when the user asks for it out loud.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["audio APR","sample APR","rubric pass rate"],"models_scored":0,"updated":"2026-11","url":"https://huggingface.co/datasets/utkarsh4430/SteerBench","paper_url":"https://arxiv.org/abs/2609.12623","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/steerbench"}
86
+ {"slug":"televal","name":"TELEVAL","maintainer":"China Telecom AI (Tele-AI)","tagline":"Chinese benchmark that checks whether spoken language models answer correctly and react appropriately to cues in the user's voice, without being told what to listen for.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding","emotion"],"metrics":["average score","Reliable Content Fulfillment","Interactional Appropriateness","speech-text consistency","DNSMOS"],"models_scored":0,"updated":"2026-08","url":"https://github.com/Tele-AI/TELEVAL","paper_url":"https://arxiv.org/abs/2507.18061","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/televal"}
87
+ {"slug":"tpi-bench","name":"TPI-Bench","maintainer":"Seoul National University / Yonsei University / University of Seoul","tagline":"Checks whether a voice assistant notices that a second person cut in on the user, and whether it then uses or ignores what that person said.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding"],"metrics":["Response Strategy Following (RSF)","Overall Helpfulness (OH)","BLEU","ROUGE-L"],"models_scored":0,"updated":"2026-04","url":"https://tpi-va.github.io/","paper_url":"https://arxiv.org/abs/2604.17358","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/tpi-bench"}
88
+ {"slug":"vera","name":"VERA","maintainer":"Duke University / Adobe","tagline":"Measures how much reasoning accuracy a model loses when the same questions are asked and answered by voice under real-time limits.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["accuracy (macro-averaged over tracks)","time to first response (TTFR)","WER"],"models_scored":0,"updated":"2025-09","url":"https://github.com/linyueqian/VERA","paper_url":"https://arxiv.org/abs/2509.26542","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/vera"}
89
+ {"slug":"voicebench-ja","name":"VoiceBench-ja","maintainer":"SB Intuitions","tagline":"Japanese test set for the gap between spoken and typed input, with the same prompts as audio and as text in four subsets.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","instruction-following"],"metrics":["LLM-judge score (1 to 5)","format following","exact match"],"models_scored":0,"updated":"2026-03","url":"https://huggingface.co/datasets/sbintuitions/voicebench-ja","paper_url":"https://arxiv.org/abs/2603.12565","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voicebench-ja"}
90
+ {"slug":"voxeval","name":"VoxEval","maintainer":"CUHK / Lightspeed Studios / NUS","tagline":"MMLU in spoken form: multiple-choice knowledge questions asked and answered in speech, repeated across speakers, speaking styles and audio quality.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["accuracy (ASR + string matching)"],"models_scored":0,"updated":"2025-07","url":"https://github.com/dreamtheater123/VoxEval","paper_url":"https://arxiv.org/abs/2501.04962","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/voxeval"}
91
+ {"slug":"wavbench","name":"WavBench","maintainer":"Xiamen University / Zhejiang University / CUHK-Shenzhen","tagline":"Scores spoken dialogue models on hard reasoning, on whether answers sound like natural talk, and on hearing and producing paralinguistic cues.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","emotion","audio-understanding"],"metrics":["colloquial score (Pro, Basic)","explicit understanding accuracy","explicit generation accuracy","implicit dialogue score"],"models_scored":0,"updated":"2026-02","url":"https://naruto-2024.github.io/wavbench.github.io/","paper_url":"https://arxiv.org/abs/2602.12135","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/wavbench"}
92
+ {"slug":"wearvox","name":"WearVox","maintainer":"Meta","tagline":"Voice assistant benchmark recorded on AI glasses, with 3,842 multi-channel egocentric clips that include real indoor and outdoor noise and bystander speech.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm","audio-understanding"],"metrics":["turn-based micro-average accuracy","tool-call AST accuracy","side-talk rejection accuracy","speech translation judge score"],"models_scored":0,"updated":"2026-04","url":"https://github.com/facebookresearch/WearVox","paper_url":"https://arxiv.org/abs/2601.02391","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/wearvox"}
93
+ {"slug":"wildspeech-bench","name":"WildSpeech-Bench","maintainer":"Tencent (WeChat AI)","tagline":"Speech-to-speech test built from real user chat queries, with varied speakers, added noise, and a human-recorded set of pauses, stress, tone, stuttering and near-homophones.","full_duplex":false,"tier":"native","setting":"lab","focus":["speech-lm"],"metrics":["average score (1 to 10)","per-category score","UTMOS"],"models_scored":0,"updated":"2025-09","url":"https://github.com/Tencent/WildSpeech-Bench","paper_url":"https://arxiv.org/abs/2506.21875","leaderboard_url":null,"page_url":"https://www.fullduplex.ai/benchmarks/wildspeech-bench"}
data/datasets.jsonl CHANGED
@@ -1,7 +1,7 @@
1
  {"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://doi.org/10.1007/11677482_3","page_url":"https://www.fullduplex.ai/datasets/ami"}
2
  {"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"}
3
- {"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"}
4
- {"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"}
5
  {"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"}
6
  {"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"}
7
  {"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"}
@@ -9,6 +9,13 @@
9
  {"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":"~229 h (2026 Second Editions)","size_hours":229,"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"}
10
  {"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"}
11
  {"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"}
 
 
 
 
 
 
 
12
  {"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"}
13
  {"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"}
14
  {"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"}
@@ -44,7 +51,7 @@
44
  {"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"}
45
  {"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"}
46
  {"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"}
47
- {"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"}
48
  {"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":"MIT (manifest only); audio rights stay with each podcast's rights holders","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"}
49
  {"slug":"apptek-callcenter-dialogues","name":"AppTek Call-Center Dialogues","maintainer":"AppTek","tagline":"Role-played agent and customer calls in 14 English accent groups, each speaker in a separate file, released for evaluation.","hours":"128.6 h","size_hours":128.6,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":16,"license":"CC-BY-SA-4.0","license_class":"share-alike","commercial_use":"yes","released":"2026-04","url":"https://huggingface.co/datasets/apptek-com/apptek_callcenter_dialogues","paper_url":"https://arxiv.org/abs/2604.27543","page_url":"https://www.fullduplex.ai/datasets/apptek-callcenter-dialogues"}
50
  {"slug":"perit-duplex-support-en","name":"Perit Duplex Support Conversations (English)","maintainer":"Perit AI","tagline":"Performed English customer-support calls with one speaker per channel at 48 kHz, under CC-BY-4.0.","hours":"5.1 h","size_hours":5.1,"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":"2026-10","url":"https://huggingface.co/datasets/perit-ai/perit-duplex-support-en","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/perit-duplex-support-en"}
@@ -54,7 +61,7 @@
54
  {"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":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-04","url":"https://huggingface.co/datasets/McGill-NLP/NaijaS2ST","paper_url":"https://arxiv.org/abs/2604.16287","page_url":"https://www.fullduplex.ai/datasets/naijas2st"}
55
  {"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"}
56
  {"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"}
57
- {"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"}
58
  {"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"}
59
  {"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"}
60
  {"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"}
@@ -64,6 +71,8 @@
64
  {"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"}
65
  {"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"}
66
  {"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":"yes","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"}
 
 
67
  {"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"}
68
  {"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"}
69
  {"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"}
@@ -71,6 +80,16 @@
71
  {"slug":"vartalaap","name":"Vartalaap","maintainer":"Kapture CX","tagline":"Synthetic Hindi-English two-party calls, customer support plus open-topic chat, with one stereo channel per role, built for full-duplex speech-to-speech training.","hours":"5,446 h","size_hours":5446.1,"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"}
72
  {"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"}
73
  {"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"}
 
 
 
 
 
 
 
 
 
 
74
  {"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"}
75
  {"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"}
76
  {"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"}
 
1
  {"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://doi.org/10.1007/11677482_3","page_url":"https://www.fullduplex.ai/datasets/ami"}
2
  {"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"}
3
+ {"slug":"candor","name":"CANDOR","maintainer":"BetterUp Labs (Reece, Cooney et al.)","tagline":"Unscripted video-call conversations between 1,656 pairs of strangers, with one participant per audio channel.","hours":"~860 h","size_hours":860,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"CANDOR Corpus License 2.1 (commercial use allowed, no redistribution)","license_class":"custom","commercial_use":"yes","released":"2022-02","url":"https://huggingface.co/datasets/betterup/candor-processed","paper_url":"https://www.science.org/doi/10.1126/sciadv.adf3197","page_url":"https://www.fullduplex.ai/datasets/candor"}
4
+ {"slug":"fisher","name":"Fisher English","maintainer":"LDC","tagline":"Ten-minute telephone conversations, a classic for conversational ASR and turn-taking baselines.","hours":"1,960 h","size_hours":1960,"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"}
5
  {"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"}
6
  {"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"}
7
  {"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"}
 
9
  {"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":"~229 h (2026 Second Editions)","size_hours":229,"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"}
10
  {"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"}
11
  {"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"}
12
+ {"slug":"callfriend","name":"CALLFRIEND","maintainer":"LDC","tagline":"Unscripted telephone calls between native speakers in 13 languages, with one end of the call on each channel.","hours":"~313 h (10 sets with stated hours)","size_hours":313,"languages":["en","fr","ar","fa","de","hi","ja","ko","zh","es","ta","vi","ru"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"LDC (paid, per-release)","license_class":"gated","commercial_use":"restricted","released":null,"url":"https://catalog.ldc.upenn.edu/LDC2023S08","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/callfriend"}
13
+ {"slug":"hkust-mandarin-telephone","name":"HKUST Mandarin Telephone Speech, Part 1 (LDC2005S15)","maintainer":"LDC","tagline":"About 149 hours of Mandarin telephone conversation, mostly between strangers in mainland China, with each call side recorded separately.","hours":"149 h","size_hours":149,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"LDC (paid)","license_class":"gated","commercial_use":"restricted","released":"2005-07","url":"https://catalog.ldc.upenn.edu/LDC2005S15","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/hkust-mandarin-telephone"}
14
+ {"slug":"greybeard","name":"Greybeard (LDC2013S05)","maintainer":"LDC","tagline":"About 590 hours of English telephone conversation among people who had taken part in earlier LDC collections, two channels per call.","hours":"590 h","size_hours":590,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"LDC (paid)","license_class":"gated","commercial_use":"restricted","released":"2013-06","url":"https://catalog.ldc.upenn.edu/LDC2013S05","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/greybeard"}
15
+ {"slug":"hcrc-map-task","name":"HCRC Map Task Corpus","maintainer":"University of Edinburgh / University of Glasgow (HCRC)","tagline":"128 task-oriented dialogues in which one speaker guides the other along a route on a map, recorded with two channels per conversation.","hours":"18 h","size_hours":18,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":20,"license":"CC-BY-4.0 (downloads from the corpus site)","license_class":"permissive","commercial_use":"yes","released":null,"url":"https://groups.inf.ed.ac.uk/maptask/","paper_url":"https://aclanthology.org/H93-1005/","page_url":"https://www.fullduplex.ai/datasets/hcrc-map-task"}
16
+ {"slug":"alimeeting","name":"AliMeeting","maintainer":"Alibaba Group / OpenSLR SLR-119","tagline":"Mandarin meetings of two to four people, recorded with an 8-channel microphone array and a headset microphone on each participant.","hours":"118.75 h","size_hours":118.75,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-SA-4.0","license_class":"share-alike","commercial_use":"yes","released":null,"url":"https://www.openslr.org/119/","paper_url":"https://arxiv.org/abs/2110.07393","page_url":"https://www.fullduplex.ai/datasets/alimeeting"}
17
+ {"slug":"dipco","name":"DiPCo (Dinner Party Corpus)","maintainer":"Amazon","tagline":"Ten dinner-party conversations among four people, each wearing a close-talk microphone, with five far-field arrays around the room.","hours":"5.3 h","size_hours":5.3,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":16,"license":"CDLA-Permissive-1.0","license_class":"permissive","commercial_use":"yes","released":"2019-09","url":"https://zenodo.org/records/8122551","paper_url":"https://arxiv.org/abs/1909.13447","page_url":"https://www.fullduplex.ai/datasets/dipco"}
18
+ {"slug":"harper-valley","name":"Gridspace-Stanford Harper Valley","maintainer":"Gridspace / Stanford CS224S","tagline":"Simulated calls to a bank's contact centre, with the agent and the caller in separate single-channel files.","hours":"23.7 h","size_hours":23.7,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":8,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2020-10","url":"https://github.com/cricketclub/gridspace-stanford-harper-valley","paper_url":"https://arxiv.org/abs/2010.13929","page_url":"https://www.fullduplex.ai/datasets/harper-valley"}
19
  {"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"}
20
  {"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"}
21
  {"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"}
 
51
  {"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"}
52
  {"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"}
53
  {"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"}
54
+ {"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":"104.9 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"}
55
  {"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":"MIT (manifest only); audio rights stay with each podcast's rights holders","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"}
56
  {"slug":"apptek-callcenter-dialogues","name":"AppTek Call-Center Dialogues","maintainer":"AppTek","tagline":"Role-played agent and customer calls in 14 English accent groups, each speaker in a separate file, released for evaluation.","hours":"128.6 h","size_hours":128.6,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":16,"license":"CC-BY-SA-4.0","license_class":"share-alike","commercial_use":"yes","released":"2026-04","url":"https://huggingface.co/datasets/apptek-com/apptek_callcenter_dialogues","paper_url":"https://arxiv.org/abs/2604.27543","page_url":"https://www.fullduplex.ai/datasets/apptek-callcenter-dialogues"}
57
  {"slug":"perit-duplex-support-en","name":"Perit Duplex Support Conversations (English)","maintainer":"Perit AI","tagline":"Performed English customer-support calls with one speaker per channel at 48 kHz, under CC-BY-4.0.","hours":"5.1 h","size_hours":5.1,"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":"2026-10","url":"https://huggingface.co/datasets/perit-ai/perit-duplex-support-en","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/perit-duplex-support-en"}
 
61
  {"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":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-04","url":"https://huggingface.co/datasets/McGill-NLP/NaijaS2ST","paper_url":"https://arxiv.org/abs/2604.16287","page_url":"https://www.fullduplex.ai/datasets/naijas2st"}
62
  {"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"}
63
  {"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"}
64
+ {"slug":"otospeech-1125h","name":"otoSpeech 1,125h","maintainer":"otoearth","tagline":"Casual two-speaker English conversation: 3,878 sessions and 565 speakers, one speaker per channel. Sold online in packs from 100 to 1,125 hours.","hours":"1,125 h","size_hours":1125,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"Commercial data licence, bought online","license_class":"custom","commercial_use":"yes","released":"2026-11","url":"https://www.oto.earth/store/otospeech-1125h","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/otospeech-1125h"}
65
  {"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"}
66
  {"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"}
67
  {"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"}
 
71
  {"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"}
72
  {"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"}
73
  {"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":"yes","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"}
74
+ {"slug":"usergyai-global-conversational-speech","name":"Global Conversational Speech","maintainer":"UsergyAI","tagline":"Two-person conversations in 18 locales across healthcare, meetings and call centres, one speaker per stereo channel. Listed at $65 per hour.","hours":"305 h","size_hours":305,"languages":["zh","en","es","pt","fr","de","it","ja","ko","yue","hi"],"speech":"human","speaker_consent":"unknown","full_duplex":true,"sample_rate_khz":null,"license":"Commercial licence, by request","license_class":"custom","commercial_use":"yes","released":"2026-04","url":"https://huggingface.co/datasets/UsergyAI/Global-Conversational-Speech","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/usergyai-global-conversational-speech"}
75
+ {"slug":"speechdata-english-us","name":"English (US) Conversational Speech","maintainer":"SpeechData.ai","tagline":"Spontaneous two-person US English conversation, one channel per speaker, with human-verified transcripts. Listed at $90 per hour.","hours":"1,500 h","size_hours":1500,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":null,"license":"Commercial (training and evaluation), by written quote","license_class":"custom","commercial_use":"yes","released":null,"url":"https://www.speechdata.ai/datasets/english-usa","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/speechdata-english-us"}
76
  {"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"}
77
  {"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"}
78
  {"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"}
 
80
  {"slug":"vartalaap","name":"Vartalaap","maintainer":"Kapture CX","tagline":"Synthetic Hindi-English two-party calls, customer support plus open-topic chat, with one stereo channel per role, built for full-duplex speech-to-speech training.","hours":"5,446 h","size_hours":5446.1,"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"}
81
  {"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"}
82
  {"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"}
83
+ {"slug":"sssd","name":"SSSD (Scalable Spontaneous Speech Dataset)","maintainer":"Carnegie Mellon University (WAVLab)","tagline":"727 hours of unscripted English conversation between randomly paired crowd workers, each speaker on a separate channel at 48 kHz.","hours":"727 h","size_hours":727,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"SSSD Data Use and License Agreement (Carnegie Mellon, no fee)","license_class":"gated","commercial_use":"restricted","released":"2025-06","url":"https://wavlab-speech.github.io/SSSD/","paper_url":"https://www.isca-archive.org/interspeech_2025/sheikh25_interspeech.html","page_url":"https://www.fullduplex.ai/datasets/sssd"}
84
+ {"slug":"casper","name":"CASPER","maintainer":"Johns Hopkins University (The SSSD Project)","tagline":"Spontaneous two-person English conversations recorded on each participant's own device, one audio file per participant.","hours":"102 h","size_hours":102,"languages":["en"],"speech":"human","speaker_consent":"consented","full_duplex":true,"sample_rate_khz":48,"license":"CC-BY-NC-SA-4.0","license_class":"non-commercial","commercial_use":"no","released":"2025-06","url":"https://huggingface.co/datasets/CASPER-SSSD/CASPER","paper_url":"https://arxiv.org/abs/2506.00267","page_url":"https://www.fullduplex.ai/datasets/casper"}
85
+ {"slug":"aishell-5","name":"AISHELL-5","maintainer":"Beijing AISHELL Technology / OpenSLR SLR-159","tagline":"Mandarin conversation among two to four people inside a car, with a microphone on each door and a headset microphone on each speaker.","hours":"105 h","size_hours":104.96,"languages":["zh"],"speech":"human","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-SA-4.0 (OpenSLR) / CC-BY-NC-SA-4.0 (AISHELL's page)","license_class":"non-commercial","commercial_use":"restricted","released":"2025-05","url":"https://www.openslr.org/159/","paper_url":"https://arxiv.org/abs/2505.23036","page_url":"https://www.fullduplex.ai/datasets/aishell-5"}
86
+ {"slug":"easy-turn-trainset","name":"Easy Turn Trainset","maintainer":"ASLP Lab, Northwestern Polytechnical University / Huawei","tagline":"About 1,145 hours of English and Chinese speech in which each recording is labelled with one of four turn-taking states.","hours":"1,145 h","size_hours":1145,"languages":["en","zh"],"speech":"mixed","speaker_consent":"mixed","full_duplex":false,"sample_rate_khz":null,"license":"Apache-2.0 (card); real subsets derive from MagicData-RAMC (CC-BY-NC-ND-4.0)","license_class":"custom","commercial_use":"restricted","released":"2025-09","url":"https://huggingface.co/datasets/ASLP-lab/Easy-Turn-Trainset","paper_url":"https://arxiv.org/abs/2509.23938","page_url":"https://www.fullduplex.ai/datasets/easy-turn-trainset"}
87
+ {"slug":"smart-turn-data","name":"Smart Turn v3.2 training data","maintainer":"Pipecat","tagline":"270,946 audio clips labelled as a finished or an unfinished turn, the open training set of the Smart Turn v3.2 detector.","hours":null,"size_hours":null,"languages":["ar","bn","zh","da","nl","de","en","fi","fr","hi","id","it","ja","ko","mr","no","pl","pt","ru","es","tr","uk","vi"],"speech":"mixed","speaker_consent":"unknown","full_duplex":false,"sample_rate_khz":null,"license":"CC-BY-4.0","license_class":"permissive","commercial_use":"yes","released":"2026-01","url":"https://huggingface.co/datasets/pipecat-ai/smart-turn-data-v3.2-train","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/smart-turn-data"}
88
+ {"slug":"semantic-vad-eot","name":"Semantic-VAD EOT","maintainer":"Scicom (MSC) Bhd","tagline":"End-of-turn training data in which each row is one speaker turn, with its pauses marked as holds and its last silence as the end of the turn.","hours":"4,064 h (conversational configs)","size_hours":4064,"languages":["en","fr","de","it","ja","ko","zh","pl","pt","ru","es","th","tr","ms","ta"],"speech":"mixed","speaker_consent":"mixed","full_duplex":false,"sample_rate_khz":16,"license":"CC-BY-4.0 (card); conversational configs keep their sources' licences","license_class":"custom","commercial_use":"restricted","released":"2026-07","url":"https://huggingface.co/datasets/Scicom-intl/semantic-vad-eot","paper_url":null,"page_url":"https://www.fullduplex.ai/datasets/semantic-vad-eot"}
89
+ {"slug":"duplexgen","name":"DuplexGen","maintainer":"DuplexGen authors (Kim et al.)","tagline":"Synthetic human-AI dialogues with turn-taking decided at word-level slots and calibrated on human preference annotations, as text and as rendered audio.","hours":"1,015 h","size_hours":1015,"languages":["en"],"speech":"synthetic","speaker_consent":"public","full_duplex":false,"sample_rate_khz":24,"license":"Apache-2.0, MIT or CC-BY-4.0, by scenario","license_class":"permissive","commercial_use":"yes","released":"2026-07","url":"https://huggingface.co/datasets/DuplexGen/duplexgen-spoken","paper_url":"https://arxiv.org/abs/2607.26178","page_url":"https://www.fullduplex.ai/datasets/duplexgen"}
90
+ {"slug":"flexislm-data","name":"FlexiSLM-Data","maintainer":"Amphion team","tagline":"Single-turn English speech-to-speech pairs, a synthetic spoken prompt and a synthetic spoken reply, built to train the FlexiSLM spoken language model.","hours":"26.7k h","size_hours":26735,"languages":["en"],"speech":"synthetic","speaker_consent":"public","full_duplex":false,"sample_rate_khz":null,"license":"Not stated on the card; prompts inherit their source datasets' terms","license_class":"custom","commercial_use":"restricted","released":"2026-08","url":"https://huggingface.co/datasets/FlexiSLM/FlexiSLM-Data-4M-s2s","paper_url":"https://arxiv.org/abs/2606.31247","page_url":"https://www.fullduplex.ai/datasets/flexislm-data"}
91
+ {"slug":"ultravoice","name":"UltraVoice","maintainer":"SJTU X-LANCE / BIGAI (Tu et al.)","tagline":"Synthetic speech dialogues in which a spoken instruction asks for a speaking style and the spoken reply follows it, across emotion, speed, volume, accent and language.","hours":"833 h","size_hours":832.92,"languages":["en","zh","ja","ko"],"speech":"synthetic","speaker_consent":"consented","full_duplex":false,"sample_rate_khz":null,"license":"MIT (card and licence file); the paper says research use only","license_class":"custom","commercial_use":"restricted","released":"2025-10","url":"https://huggingface.co/datasets/tutu0604/UltraVoice","paper_url":"https://arxiv.org/abs/2510.22588","page_url":"https://www.fullduplex.ai/datasets/ultravoice"}
92
+ {"slug":"voiceassistant-400k","name":"VoiceAssistant-400K","maintainer":"gpt-omni (Mini-Omni authors)","tagline":"Spoken questions paired with assistant-style answers as text and audio tokens, made to fine-tune models that answer in speech.","hours":null,"size_hours":null,"languages":["en"],"speech":"synthetic","speaker_consent":"unknown","full_duplex":false,"sample_rate_khz":null,"license":"Apache-2.0","license_class":"permissive","commercial_use":"yes","released":"2024-09","url":"https://huggingface.co/datasets/gpt-omni/VoiceAssistant-400K","paper_url":"https://arxiv.org/abs/2408.16725","page_url":"https://www.fullduplex.ai/datasets/voiceassistant-400k"}
93
  {"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"}
94
  {"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"}
95
  {"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"}
data/models.jsonl CHANGED
The diff for this file is too large to render. See raw diff