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.
Large collection of Korean audio recordings (thousands of hours) with metadata, tags, and instructions for speech synthesis.
Over 366 million audio clips with captions used to train LAION CLAP2 models.
Over 133,000 hours of voiced books across 86 languages for training speech recognition and synthesis systems.
Up to 950K hours of open speech across 24 official EU languages with automatic transcripts.
Largest open audio-text corpus in Persian, collected from audiobooks for speech synthesis and recognition.
A cleaned multimodal corpus for joint speech and video generation of talking persons in Chinese and English.
English aspect lists and free-form music descriptions by musicians for 5,521 examples.
A large multi-task streaming audio dataset for training audio LLMs and voice agents.
Metadata of nearly 660k Suno AI songs with multilingual descriptions and audio modality.
English annotations of speaker and speech characteristics for training controllable text-to-speech (TTS) synthesis.
A multimodal corpus of millions of examples for training universal query-based audio separation systems.
Pairwise audio comparisons from 15 TTS models in English for RLHF and quality evaluation.