重大X投稿2026-09-28 03:26 Who wants more speech training data? YODAS v3 is now available @huggingface It’s 1.1M hours - the biggest audio dataset ever. And the first … 報道ソース Hugging Face2026-09-28 03:26