Skip to content

Datasets

Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.

Data type

For which models

Find a dataset for your goal

Answered by Gemini 4 · the chat is kept only in this tab

DatasetMLCommons/unsupervised_peoples_speechTop pick

Unsupervised People's Speech

Over one million hours of English audio from Archive.org for ASR and audio classification.

  • Automatic speech recognition
  • Audio classification
  • English
21.3K 83
DatasetHumanBehaviorAtlas/human_behavior_atlasTop pick

Multimodal dataset for behavior and emotion analysis

Combined corpus of 16 datasets with video, audio, and text for recognizing emotions, humor, and human state.

  • Video classification
  • Audio classification
  • Text classification
  • English
  • Chinese
20K 4100K–1M rows
Datasetawsaf49/epic_kitchens_100

EPIC-KITCHENS kitchen action video dataset mirror

A re-upload of the well-known first-person video benchmark with audio for quick downloading.

  • Voice activity detection
  • Video classification
19.8K 28up to 1K rows
Datasetsimon3000/genshin-voice

Genshin Impact voice lines

654k wav files of Genshin Impact character voice lines in Chinese, English, Japanese, and Korean.

  • Audio classification
  • Automatic speech recognition
  • Text → speech
  • Chinese
  • English
  • Japanese
  • +1
18.6K 307100K–1M rows
DatasetDBD-research-group/BirdSet

Large bird sound audio dataset for classification

Over 7,200 hours of recordings of nearly 10,000 bird species for training and evaluating audio classification models.

  • Audio classification
18K 33
Datasetbasis-ai/basis-conversations-1500Top pick

Multilingual multi-speaker conversations (48 kHz)

Multilingual dataset of live dialogues of up to 4 people with separate audio channels for ASR.

  • Automatic speech recognition
  • Audio classification
  • Arabic
  • German
  • English
  • +9
14.3K 6310K–100K rows🔒 on request
Datasetgoogle/svq

Short Spoken Questions in 17 Languages

Audio questions across 26 locales to evaluate ASR, audio search, and audio embeddings.

  • Question answering
  • Automatic speech recognition
  • Text retrieval
  • Arabic
  • Bangla
  • English
  • +9
14.1K 681M–10M rows
Datasetpsdn-ai/CROWS-multilingualTop pick

Multilingual crowdsourced speech corpus in 16 languages

A large audio corpus with text-reading recordings and annotations for training ASR and speech classification systems.

  • Automatic speech recognition
  • Audio classification
  • Arabic
  • German
  • English
  • +9
14K 5100K–1M rows
Datasetsimon3000/starrail-voice

Honkai: Star Rail Voice Lines

403,437 wav voice lines of Honkai: Star Rail characters in Chinese, English, Japanese, and Korean.

  • Audio classification
  • Automatic speech recognition
  • Text → speech
  • Chinese
  • English
  • Japanese
  • +1
13.1K 70100K–1M rows
DatasetCnam-LMSSC/vibravox

French speech recorded with bone conduction sensors

Audio dataset of French speech captured by body conduction sensors for recognition, synthesis, and audio enhancement tasks.

  • Audio → audio
  • Automatic speech recognition
  • Audio classification
  • French
12.4K 3110K–100K rows
Datasetgoogle/speech_commands

Basic Google voice commands dataset

Tens of thousands of one-second audio recordings of basic English words and background noise for command recognition training.

  • Audio classification
  • English
11K 66100K–1M rows
DatasetProgramComputer/voxceleb

VoxCeleb and VoxCeleb2 audio and video

Audio-video dataset for speech recognition and classification; combined multi-volume ZIP archives.

  • Automatic speech recognition
  • Audio classification
  • Image classification
10.6K 130100K–1M rows🔒 on request
DatasetCanCLID/zoengjyutgaai

Cantonese audiobooks and transcriptions

Recordings by famous storyteller Zhang Yuekai with verified texts for Cantonese TTS and ASR.

  • Automatic speech recognition
  • Text → speech
  • Text generation
  • Cantonese
10.5K 35100K–1M rows
DatasetMohammed01/ArFakeTop pick

Benchmark for Arabic speech deepfake detection

First multi-dialect dataset of synthetic and real Arabic speech for detecting audio fakes.

  • Audio classification
  • Automatic speech recognition
  • Text → audio
  • Arabic
10.2K 910K–100K rows🔒 on request
DatasetMLCommons/ml_spoken_words

50-language spoken words corpus

Large-scale set of one-second audio recordings of individual words in 50 languages for keyword spotting.

  • Audio classification
  • Arabic
  • Assamese
  • Breton
  • +9
9.8K 3710M–100M rows
Datasetfacebook/omnilingual-asr-corpusTop pick

Meta ASR corpus in 348 rare languages

Spontaneous speech recordings and transcriptions for 348 languages for ASR training and language identification.

  • Automatic speech recognition
  • Audio classification
  • aae
  • aal
  • aao
  • +9
9.5K 222100K–1M rows
DatasetLoie/VGGSound

Audiovisual YouTube sound dataset

Short audio clips from YouTube videos across more than 310 classes for sound recognition.

  • Audio classification
7.3K 58up to 1K rows
Datasetbasis-ai/basis-global-emotionsTop pick

Multilingual emotional speech dataset

Over 525 hours of audio with expressed emotions and speaking styles across 108 locales for voice synthesis and analysis.

  • Text → speech
  • Audio classification
  • Afrikaans
  • Akan
  • Amharic
  • +9
6.7K 56100K–1M rows🔒 on request
Datasetedinburghcstr/edacc

English dialogues with diverse accents

40 hours of paired English conversations with diverse accents for ASR training and evaluation.

  • Automatic speech recognition
  • Audio classification
6.6K 1810K–100K rows
Datasetlangswap/dialogs-ru-emotional-conversationsTop pick

Studio emotional dialogues in Russian

20.6 hours of studio recordings of live theatrical actor dialogues for emotional TTS and ASR.

  • Text → speech
  • Automatic speech recognition
  • Audio classification
  • Russian
5.6K 27up to 1K rows
DatasetHFUT-MMH-Lab/multimodal_mental_health_dataset_MMH

Multimodal dataset for mental disorder diagnosis

Participant audio, video, and texts in English and Chinese with clinically confirmed diagnoses for distinguishing disorders.

  • Audio classification
  • Video classification
  • Text classification
  • English
  • Chinese
5.3K 510K–100K rows
Datasetamazon-agi/SIFT-50MTop pick

50 million instructions for speech language models from Amazon

Large-scale multilingual corpus based on 14,000 hours of audio for training speech understanding and generation.

  • Audio text → text
  • Audio classification
  • Text → speech
  • English
  • German
  • French
  • +2
4.3K 3910M–100M rows
DatasetFBK-MT/Speech-MASSIVE

Multilingual spoken language understanding corpus

Spoken version of the MASSIVE text corpus for ASR, intent detection, and slot filling tasks.

  • Audio classification
  • Text classification
  • Zero shot classification
  • Arabic
  • German
  • Spanish
  • +9
3.6K 5710K–100K rows