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
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.
Answered by Gemini 4 · the chat is kept only in this tab
Over 1.1 million hours of web audio in 100+ languages for ASR, TTS, and audio representation learning.
A massive dataset of over 1.1 million hours of audio across 100+ languages for speech tasks.
Speech data in 102 languages for training and evaluating ASR on parallel sentences.
4,000+ hours of audio and video of in-person interactions by 4,000+ participants for AI research.
Spoken Egyptian Arabic dialect audio corpus with transcripts for TTS and ASR training.
About 1000 hours of narrated English audiobooks with text for ASR training and evaluation.
Synchronized video, audio, and 3D data from real home environments for embodied AI training.
A large multimodal corpus of spatial audio recordings with refined English and Chinese text annotations.
Over 10,000 hours of VTLA data with tactile sensors for robot training.
Multimodal set of 4,307 characters with AI images, definitions, and pinyin audio for generative and language models.
A stress test for ASR models consisting of 100 audio clips where three people speak different languages simultaneously.
Multilingual audio and text for speech generation: 101k hours original and over 200k in Emilia-Large.
Large corpus of aligned parliamentary speech in 22 European languages for ASR and TTS.
An audio and text benchmark in English and Korean for evaluating neural network roles in music production.
200 hours of synchronized first-person recordings with RGB, LiDAR depth, audio, and text for embodied AI and robotics.
Speech recordings of a single speaker with an Irish accent mixing English and Hebrew for speech synthesis training.
An audio dataset of read-aloud sentences in noisy environments in English and Hebrew for ASR tasks.
Mirror of the well-known FSD50K dataset for recognizing and classifying everyday and natural sounds.
About 41 million speech editing pairs in English and Chinese with refined transcripts.
A large neutral Mandarin speech corpus from 218 speakers with gender, age, and accent annotations for TTS training.
Single-speaker audio recordings in English and Hebrew with text transcripts for speech recognition fine-tuning.
A collection of measured room impulse responses with narrowband frequency-dependent truncation at a 48 kHz sampling rate.
Camera, lidar, radar, and microphone recordings for detecting and tracking miniature UAVs.
Arabic audio dataset from 128 reciters with word-level synchronization for ASR and TTS tasks.