Fullduplex
Subscribe
Datasetsverified 2026-11. Download on Hugging Face
107 of 107Human speech and open licences first
Full-duplexLanguagesLicenceSourceSpeech
Common Voice (Scripted Speech 27.0)Mozilla
—42.6k h29.3k validated295PermissiveConsentedhuman2019-06
IndicVoices (conversational slice)AI4Bharat (IIT Madras)
—~3,500 hconv.hi bn ta +19PermissiveConsentedhuman2024-03
CoVoST-2Meta AI
—2,880 h21 + enPermissiveConsentedhuman2020-07
FLEURSGoogle Research
—~1k h102+PermissiveConsentedhuman2022-05
otoSpeech-full-duplex-280hotoearth
Full-duplex280 henPermissiveConsentedhuman2026-02
AISHELL-1Beijing Shell Shell Technology
—178 hzhPermissiveConsentedhuman2017-09
otoSpeech-full-duplex-processed-141hotoearth
Full-duplex141 henPermissiveConsentedhuman2026-02
AMI Meeting CorpusUniversity of Edinburgh / Idiap
—100 henPermissiveConsentedhuman2006-06
AISHELL-3Beijing Shell Shell Technology
—85 hzhPermissiveConsentedhuman2020-10
ICSI Meeting CorpusICSI / University of Edinburgh (AMI site)
—72 henPermissiveConsentedhuman2004-02
VCTK CorpusCSTR, University of Edinburgh
—44 henPermissiveConsentedhuman2019-11
otoSpeech-full-duplex-task-oriented-20hotoearth
Full-duplex20 henPermissiveConsentedhuman2026-08
otoSpeech-full-duplex-task-oriented-v2otoearth
Full-duplex17.4 henPermissiveConsentedhuman2026-09
PriMock57Babylon Health
—~9 henPermissiveConsentedhuman2022-05
Perit Duplex Support Conversations (English)Perit AI
Full-duplex5.1 henPermissiveConsentedhuman2026-10
Indic DiarBenchSarvam AI
—~108 hhi bn ta +19PermissiveMixedhuman2026-07
YODAS v3ESPnet / WavLab (CMU)
—1.1M h100+PermissivePublic mediahuman2026-09
YODAS / YODAS2WavLab (CMU)
—500k h100+PermissivePublic mediahuman2024-06
VoxPopuliMeta AI
—400k hen de fr +20PermissivePublic mediahuman2021-01
Libri-LightMeta AI (FAIR)
—60k henPermissivePublic mediahuman2019-12
Multilingual LibriSpeech (MLS)Meta AI
—50k hen de nl +5PermissivePublic mediahuman2020-12
VoxLingua107Tallinn University of Technology
—6,628 hmultiPermissivePublic mediahuman2020-11
LibriSpeechVassil Panayotov et al.
—960 henPermissivePublic mediahuman2015-04
LibriTTSGoogle / Heiga Zen et al.
—585 henPermissivePublic mediahuman2019-04
LibriTTS-RGoogle
—585 henPermissivePublic mediahuman2023-05
Hi-Fi TTSNVIDIA
—292 henPermissivePublic mediahuman2021-04
CSS10Kyubyong Park et al.
—140 hde el es +7PermissivePublic mediahuman2019-03
MUSANJohns Hopkins University
—109 hmultiPermissivePublic mediahuman2015-10
LJSpeechKeith Ito
—24 henPermissivePublic mediahuman2017-07
Real-TurnTurkAta Technology Platforms
Full-duplex1.94 hpublic subsettrPermissiveNot statedhuman2026-08
AppTek Call-Center DialoguesAppTek
Full-duplex128.6 henShare-alikeConsentedhuman2026-04
CREMA-DUniversity of Pennsylvania
—~5 henShare-alikeConsentedhuman2014-08
People's SpeechMLCommons
—30k+ henShare-alikePublic mediahuman2021-11
Earnings-22Rev.com
—119 henShare-alikePublic mediahuman2022-03
DailyTalkKAIST
—20 henShare-alikeConsentedhuman2022-07
GranaryNVIDIA
—643k h192k h with audioen de fr +22PermissivePublic mediahuman2025-05
ReazonSpeechReazon Holdings
—35k hjaShare-alikePublic mediahuman2023-03
VoxCeleb2VGG, University of Oxford
—2,442 hmultiShare-alikePublic mediahuman2018-06
VoxCeleb1VGG, University of Oxford
—352 henShare-alikePublic mediahuman2017-06
Seamless InteractionMeta FAIR
Full-duplex4,000+ henNon-commercialConsentedhuman2025-06
SmoothConv / DuplexConvASLP@NPU / QualiaLabs
Full-duplex2,100 h100 h + 2,000 hzhNon-commercialConsentedhuman2026-06
CANDORBetterUp Labs (Reece, Cooney et al.)
Full-duplex850 henNon-commercialConsentedhuman2023-03
MyST Children's Conversational SpeechBoulder Learning / LDC
—393 henNon-commercialConsentedhuman2021-06
MultiDialogIVLLab (Yonsei / Naver)
—340 henNon-commercialConsentedhuman2024-06
SpokenWOZAlibaba DAMO Academy
—249 henNon-commercialConsentedhuman2023-05
MagicData-RAMCMagic Data Technology / OpenSLR SLR-123
—180 hzhNon-commercialConsentedhuman2022-03
WHAM! / WHAMR!Whisper.ai / MERL
—~80 henNon-commercialConsentedhuman2019-07
CHiME-6CHiME Challenge
—40 henNon-commercialConsentedhuman2020-04
ExpressoMeta AI
—40 henNon-commercialConsentedhuman2023-08
Multi-stream Spontaneous Conversation (zh+en)Magic Data Technology
Full-duplex15 h10 zh + 5 enzh enNon-commercialConsentedhuman2024-11
Japanese Duplex Conversation DatasetMagicData
Full-duplex10.35 hjaNon-commercialConsentedhuman2025-04
RAVDESSRyerson University
—~3 henNon-commercialConsentedhuman2018-04
Emilia / Emilia-LargeAmphion (CUHK-Shenzhen)
—215.6k hEmilia-Largeen zh ja +3Non-commercialPublic mediahuman2024-07
J-CHATSarulab (UTokyo) / Keio
—76k hjaNon-commercialPublic mediahuman2024-07
WenetSpeechWeNet community / NPU-ASLP
—10k hlabelledzhNon-commercialPublic mediahuman2021-10
MuST-CFBK (Fondazione Bruno Kessler)
—~4k hen → 14Non-commercialPublic mediahuman2019-06
Multilingual TEDxJohns Hopkins / CMU
—760 hes fr pt +5Non-commercialPublic mediahuman2021-02
TED-LIUM 3LIUM, Université du Mans
—452 henNon-commercialPublic mediahuman2018-05
CN-CelebCSLT, Tsinghua University
—274 hzhNon-commercialPublic mediahuman2019-11
OleSpeech-IVOlewave / Stanford / Microsoft / PingAn / KU / CMU
—100 hopen subsetenNon-commercialPublic mediahuman2025-09
Fisher EnglishLDC
Full-duplex2,000 henGatedConsentedhuman2004-12
MLC-SLM (Interspeech 2025)Nexdata / Interspeech 2025 challenge
—1,604 hen fr de +8GatedConsentedhuman2025-09
Basis Conversations 1500Basis
Full-duplex1,502 hen es fr +19GatedConsentedhuman2026-09
Open Yap 1KThe Agentic Data Company
Full-duplex1,000 h8.9 h open sampleenGatedConsentedhuman2026-09
AISHELL-2Beijing Shell Shell Technology
—1,000 hzhGatedConsentedhuman2018-08
KsponSpeechETRI / Korean AI Hub
—969 hkoGatedConsentedhuman2020-03
Corpus of Spontaneous Japanese (CSJ)NINJAL / NICT / Tokyo Tech
—661 hjaGatedConsentedhuman2004-01
Switchboard-1 (LDC97S62)LDC
Full-duplex260 henGatedConsentedhuman1993-01
CALLHOME (JA / ZH / ES / DE / AR …)LDC
Full-duplex~229 h2026 Second Editionsja zh es +3GatedConsentedhuman1996-01
otoSpeech-full-duplex-turn-104hotoearth
Full-duplex105 henGatedConsentedhuman2026-06
Wall Street Journal (WSJ0 / WSJ1)LDC
—~80 henGatedConsentedhuman1993-01
DAI-ASR-I18NDavid AI
Full-duplex42.1 hen zh ja +18GatedConsentedhuman2026-10
IEMOCAPUSC SAIL
—12 henGatedConsentedhuman2008-12
TIMITDARPA / LDC
—5.4 henGatedConsentedhuman1993-10
GigaSpeechSpeechColab
—10k henGatedPublic mediahuman2021-06
SPGISpeechKensho (S&P Global)
—5,000 henGatedPublic mediahuman2021-04
MSP-PodcastUT Dallas (MSP Lab)
—409 hv2.0enGatedPublic mediahuman2017-01
otoSpeech 1,125hotoearth
Full-duplex1,125 henCustomConsentedhuman—
JVS CorpusTakamichi Lab (University of Tokyo)
—30 hjaCustomConsentedhuman2019-08
ESD (Emotional Speech Dataset)NUS / HLT Lab
—29 hen zhCustomConsentedhuman2021-05
L2-ARCTICTexas A&M / Iowa State
—27 henCustomConsentedhuman2018-09
Dialogs (Russian)Shigabeev, Latyshev
—20.6 hruCustomConsentedhuman2026-07
JSUTShinnosuke Takamichi (U. Tokyo)
—10 hjaCustomConsentedhuman2017-11
otoSpeech Videootoearth
Full-duplex—enCustomConsentedhuman—
otoSpeech Face-to-Faceotoearth
Full-duplex—enCustomConsentedhuman—
otoSpeech 3-Speakerotoearth
Full-duplex—enCustomConsentedhuman—
otoSpeech Koreanotoearth
Full-duplex—koCustomConsentedhuman—
otoSpeech Japaneseotoearth
Full-duplex—jaCustomConsentedhuman—
otoSpeech Role-Playotoearth
Full-duplex—jaCustomConsentedhuman—
otoSpeech Scriptedotoearth
Full-duplex—jaCustomConsentedhuman—
otoSpeech Egocentricotoearth
Full-duplex—enCustomConsentedhuman—
DuplexChatNakata, Saito, Saruwatari (U. Tokyo / AIST)
Full-duplex415k hen jaCustomPublic mediahuman2026-07
MM-F2FXiamen University
—210 henCustomPublic mediahuman2025-07
MELDDeCLaRe Lab / SUTD
—~12 henCustomPublic mediahuman2018-10
CVSSGoogle Research
—3,809 h21 → enPermissiveConsentedmixed2022-01
NaijaS2STMaltais et al.
—~200 hig ha yo +2PermissiveConsentedmixed2026-04
CVSS-XGris et al.
—~16,070 hsynthetic target speechen +28Non-commercialConsentedmixed2026-09
SD-EvalAmphion (CUHK-Shenzhen)
—8.76 henNon-commercialMixedmixed2024-06
InteractSpeechInteractSpeech authors (EMNLP 2025)
—150 haudio not releasedenCustomConsentedmixed2025-11
VoxDialogueVoxDialogue authors (ICLR 2025)
——enPermissivePublic mediasynthetic2025-05
Behavior-SDSeoul National University (NAACL 2025)
—2,164 henPermissiveNot statedsynthetic2025-04
Korean full-duplex synthetic (preview)Wi-Fi (Hugging Face)
—preview only2,000.5 h claimedkoPermissiveNot statedsynthetic2026-08
VartalaapKapture CX
Full-duplex5,446 hhi enNon-commercialConsentedsynthetic2026-08
InstructS2S-200KICTNLP (LLaMA-Omni team)
——enNon-commercialPublic mediasynthetic2024-09
MultiTalkPT / MultiTalkFTCUHK MMLab
Full-duplex57.6k hen zhNon-commercialNot statedsynthetic2026-09
MoVE (paired EN/ZH emotional speech)Chen et al.
—1,056 hEnglish sideen zhNon-commercialNot statedsynthetic2026-04
DEMANDInria / University of Edinburgh
—~5 hmultiShare-alike——2013-06