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

Datasetespnet/yodas3Top pick

YODAS v3 100+ language speech corpus

Over 1.1 million hours of web audio in 100+ languages for ASR, TTS, and audio representation learning.

  • Audio → audio
  • Automatic speech recognition
  • Text → speech
253.4K 2631M–10M rows
Datasetespnet/yodas3.1Top pick

Audio corpus in 100+ languages

A massive dataset of over 1.1 million hours of audio across 100+ languages for speech tasks.

  • Audio → audio
  • Automatic speech recognition
  • Text → speech
230.8K 2761M–10M rows
Datasetgoogle/fleurs

FLEURS multilingual ASR benchmark

Speech data in 102 languages for training and evaluating ASR on parallel sentences.

  • Automatic speech recognition
  • Afrikaans
  • Amharic
  • Arabic
  • +9
220.2K 541100K–1M rows
Datasetehabnegm/100-hour-Egyptian-dataset-single-speakerTop pick

100 hours of spoken Egyptian Arabic (single speaker)

Spoken Egyptian Arabic dialect audio corpus with transcripts for TTS and ASR training.

  • Text → speech
  • Automatic speech recognition
  • Arabic
  • Egyptian Arabic
157.6K 1510K–100K rows
Datasetopenslr/librispeech_asr

LibriSpeech: 1000 hours of English speech

About 1000 hours of narrated English audiobooks with text for ASR training and evaluation.

  • Automatic speech recognition
  • Audio classification
  • English
130.7K 279100K–1M rows
Datasetlaion/small-overlapping-speech-bench

Overlapping speech recognition benchmark

A stress test for ASR models consisting of 100 audio clips where three people speak different languages simultaneously.

  • Automatic speech recognition
  • German
  • French
  • Spanish
  • +3
102.9K 2up to 1K rows
Datasetamphion/Emilia-DatasetTop pick

Emilia multilingual speech corpus

Multilingual audio and text for speech generation: 101k hours original and over 200k in Emilia-Large.

  • Text → speech
  • Automatic speech recognition
  • Chinese
  • English
  • Japanese
  • +3
89.7K 52010M–100M rows🔒 on request
Datasetdisco-eth/EuroSpeechTop pick

EuroSpeech multilingual parliamentary speech

Large corpus of aligned parliamentary speech in 22 European languages for ASR and TTS.

  • Automatic speech recognition
  • Text → speech
  • German
  • English
  • Maltese
  • +9
87.8K 10710M–100M rows
DatasetHebrish/Noisy-Condition-Sentences

Noisy condition speech recordings in English and Hebrew

An audio dataset of read-aloud sentences in noisy environments in English and Hebrew for ASR tasks.

  • Automatic speech recognition
  • English
  • Hebrew
67.8K 010K–100K rows🔒 on request
Datasetdanielrosehill/Read-Aloud-Voice-Sentences

Read-aloud sentence recordings for ASR personalization

Single-speaker audio recordings in English and Hebrew with text transcripts for speech recognition fine-tuning.

  • Automatic speech recognition
  • English
  • Hebrew
60.3K 010K–100K rows🔒 on request
Datasetzaibihassan/Quranic-Recitation-Data

Studio Quran audio recordings with word-level alignment

Arabic audio dataset from 128 reciters with word-level synchronization for ASR and TTS tasks.

  • Automatic speech recognition
  • Text → speech
  • Audio classification
  • Arabic
57.8K 9100K–1M rows
DatasetEthnic-Voices-Tz/ethnic-voices-tanzaniaTop pick

Speech and Texts in Six Tanzanian Languages

A corpus of 462 hours of audio recordings and texts in rare ethnic languages of Tanzania for training ASR and language models.

  • Automatic speech recognition
  • Text generation
54.4K 010K–100K rows
Datasettheusamaaslam/urdu-callsqa-private

Urdu phone call audio recordings for ASR

417 hours of annotated speech from Urdu phone calls for speech recognition and synthesis.

  • Automatic speech recognition
  • Text → speech
  • Urdu
51.4K 010K–100K rows🔒 on request
Datasetedinburghcstr/ami

AMI meeting corpus: 100 hours of audio

100 hours of meeting recordings with synchronized microphones, video, and text for ASR and meeting analysis.

  • Automatic speech recognition
  • English
48.5K 100100K–1M rows
DatasetMLCommons/peoples_speech

People's Speech: 30k hours of English speech

Large open ASR corpus: over 30,000 hours of transcribed English speech.

  • Automatic speech recognition
  • English
46.9K 3061M–10M rows
Datasetfacebook/voxpopuli

Multilingual European Parliament Speech Corpus

European Parliament speech audio and transcripts in many languages for ASR and representation learning.

  • Automatic speech recognition
  • English
  • German
  • French
  • +9
39.5K 1831M–10M rows
Datasetfacebook/multilingual_librispeech

Multilingual LibriSpeech (MLS) for Speech

LibriVox audiobooks in 8 languages with texts for ASR, TTS, and speech research.

  • Automatic speech recognition
  • Text → speech
  • Text → audio
  • German
  • Dutch
  • French
  • +5
38.9K 2011M–10M rows
Datasetfsicoli/common_voice_15_0

Multilingual Common Voice 15.0 speech corpus mirror

Unofficial export of the 15th version of the open Mozilla Common Voice dataset in dozens of languages for speech recognition.

  • Automatic speech recognition
  • Abkhazian
  • Afrikaans
  • Amharic
  • +9
38.2K 6100B–1T rows
Datasetanquachdev/en-vi-speech-20k

English-Vietnamese speech translation corpus (Hibiki-Zero)

Between 1M and 10M records for training simultaneous speech translation from Vietnamese to English, ASR, and TTS.

  • Translation
  • Automatic speech recognition
  • Text → speech
  • English
  • Vietnamese
37.9K 01M–10M rows🔒 on request
DatasetARTPARK-IISc/VaaniTop pick

Vaani Indian multilingual speech

31k hours of image-prompted spontaneous speech across 105 Indian languages, with transcripts for ASR and multimodal tasks.

  • Automatic speech recognition
  • Text → speech
  • Image → text
  • Nepali
  • Assamese
  • Malayalam
  • +9
37K 19810M–100M rows🔒 on request
DatasetHebrish/Yiddish-Sentences

English speech with Yiddish loanwords

A small audio and text dataset of English sentences containing Yiddish words for ASR tasks.

  • Automatic speech recognition
  • English
  • Yiddish
36.2K 0up to 1K rows🔒 on request
Datasetanyantudre/MooreSpeechCorporaV2

Cleaned Moore speech corpus for ASR and TTS

Between 100k and 1M audio-text pairs in Moore with verified alignment for ASR and TTS.

  • Automatic speech recognition
  • Text → speech
  • Mossi
  • French
35.6K 0100K–1M rows🔒 on request
Datasetespnet/yodas-granaryTop pick

YODAS-Granary European speech corpus for ASR and translation

A large multilingual audio dataset (10M-100M entries) with pseudo-labels for European speech recognition and translation.

  • Automatic speech recognition
  • Translation
  • Bulgarian
  • Czech
  • Danish
  • +9
35.1K 4310M–100M rows
DatasetQUD-Technologies/quranic-universal-ayahsTop pick

Quran audio recordings and timestamps

Arabic Quranic audio recordings with word-level timestamps and waqf markup.

  • Automatic speech recognition
  • Arabic
34.5K 14100K–1M rows🔒 on request