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

Datasetopedromartins/ASR-datasets-ptbr

Unified Brazilian Portuguese corpus for ASR

Large collection of audio recordings with transcriptions in Brazilian Portuguese for training speech recognition systems.

  • Automatic speech recognition
  • Portuguese
869 141M–10M rows
DatasetHKUSTAudio/Audio-FLAN-DatasetTop pick

Multitask audio dataset for instruction-tuning

Massive Audio-FLAN corpus in English and Chinese for speech, music, and sound understanding and generation.

  • Text → speech
  • Text → audio
  • Automatic speech recognition
  • English
  • Chinese
863 4710M–100M rows🔒 on request
Datasetpnnbao-ump/VieNeu-TTS-140h

140 hours of Vietnamese speech with phonemic notation

A benchmark dataset of 74,858 Vietnamese audio recordings with IPA transcriptions for speech synthesis and recognition.

  • Text → speech
  • Automatic speech recognition
  • Vietnamese
856 4010K–100K rows🔒 on request
Datasetdoof-ferb/fpt_fosd

Unofficial mirror of FPT Open Speech Dataset

Vietnamese audio dataset for automatic speech recognition and synthesis.

  • Automatic speech recognition
  • Text → speech
  • Vietnamese
812 610K–100K rows
DatasetJzuluaga/atco2_corpus_1h

ATCO2 1-hour air traffic control test corpus

A small audio test corpus of air traffic control radio communications with English transcriptions for ASR verification.

  • Automatic speech recognition
  • English
788 13up to 1K rows
DatasetASLP-lab/MS-Bench

Multilingual multi-speaker conversational ASR benchmark

32.63 hours of multi-voice recordings in 6 languages with speech overlap for testing recognition systems.

  • Automatic speech recognition
  • Chinese
  • English
  • Portuguese
  • +4
772 3up to 1K rows
Datasetreazon-research/reazonspeech

Japanese speech from television broadcasts

35,000+ hours of Japanese speech from television broadcasts for training and evaluating ASR models.

  • Automatic speech recognition
  • Japanese
699 13910M–100M rows🔒 on request
Datasetamphion/Emilia-NV

Chinese speech with non-verbal sound annotations

Mandarin Chinese audio and text with word-level annotations of sighs, laughter, and interjections for ASR and TTS.

  • Text → speech
  • Automatic speech recognition
  • Chinese
625 56100K–1M rows🔒 on request
DatasetDr-AliGomaa/ar-eg-dataset

Egyptian Arabic speech corpus

An audio dataset of Egyptian Arabic lectures with transcripts for speech recognition.

  • Automatic speech recognition
  • Translation
  • Arabic
605 21K–10K rows
Datasetfacebook/covost2

Multilingual corpus for speech translation and ASR

A large audio corpus based on Mozilla Common Voice for speech recognition and translation between English and dozens of languages.

  • Automatic speech recognition
  • Arabic
  • Catalan
  • Welsh
  • +9
600 52100K–1M rows
DatasetJaimeMaLangue/JML-fon-speech-v1

Fon language speech corpus

Fon language speech data corpus with linguist-verified transcriptions for speech technologies.

  • Automatic speech recognition
  • Fon
564 64100K–1M rows
Datasetopenslr/openslr

Speech corpus from the OpenSLR archive

A small multilingual audio dataset from the OpenSLR project for automatic speech recognition (ASR) tasks.

  • Automatic speech recognition
  • Afrikaans
  • Bangla
  • Catalan
  • +9
550 321K–10K rows
Datasethalabi2016/arabic_speech_corpus

Studio Arabic speech corpus (Damascus)

Studio-quality Arabic speech recordings (Levantine dialect) for speech recognition and synthesis tasks.

  • Automatic speech recognition
  • Arabic
488 371K–10K rows
DatasetBAAI/ChildMandarinTop pick

Mandarin Chinese child speech (3-5 years)

Open audio dataset of speech from 3-5 year old children in Mandarin Chinese for ASR and speaker verification tasks.

  • Automatic speech recognition
  • Chinese
481 5010K–100K rows🔒 on request
Datasetlitagin/Galgame_Speech_ASR_16kHz

Large Japanese speech corpus from visual novels

From 1 to 10 million pairs of 16 kHz audio and Japanese transcriptions for ASR tasks.

  • Automatic speech recognition
  • Japanese
470 481M–10M rows
Datasetespnet/ace-kising-segments

ACE-KiSing singing voice segments

Vocal audio recordings and lyrics in Chinese and English for singing generation and recognition.

  • Text → audio
  • Audio → audio
  • Automatic speech recognition
  • Chinese
  • English
451 810K–100K rows
DatasetMohamedRashad/arabic-english-code-switching

ASR audio dataset with mixed Arabic-English speech

Speech recordings with Arabic-English code-switching for training speech recognition systems.

  • Automatic speech recognition
  • Arabic
  • English
395 3810K–100K rows
DatasetMohamedRashad/Quran-Recitations

Quran recitation audio with verse texts

A large Arabic corpus (100k-1M records) of paired Quranic texts and audio recordings by various reciters for ASR and TTS.

  • Automatic speech recognition
  • Text → speech
  • Arabic
392 63100K–1M rows
Datasetdoof-ferb/infore1_25hours

InfoRe Vietnamese audio dataset (25 hours)

Vietnamese audio recordings and texts for training speech recognition and synthesis.

  • Automatic speech recognition
  • Text → speech
  • Vietnamese
392 1610K–100K rows
DatasetFBK-MT/Speech-MASSIVE-test

Test split of the multilingual SLU dataset

Test split of a multilingual spoken corpus for speech recognition and intent understanding.

  • Audio classification
  • Text classification
  • Zero shot classification
  • Arabic
  • German
  • Spanish
  • +9
377 1110K–100K rows🔒 on request