{"success":true,"data":{"author":"ESPnet / CMU WAVLab","bucketAccessType":"public","bucketName":"espnet/yodas3","bucketPrefix":"","category":"Speech & Audio","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:17.721Z","description":"YODAS v3 is a large web-crawled speech dataset of over 1.1 million hours of audio in over 100 languages, all originally released under a CC-BY-3.0 licence, published by the ESPnet team as the third version of YODAS (YouTube-Oriented Dataset for Audio and Speech). The videos were crawled as a distinct set from YODAS v1 and v2, so there is no overlap with those releases. For each audio file it provides segment-level transcripts where available, English translations where available, language IDs, audio length and effective bandwidth. It is intended for automatic speech recognition, speech translation, text-to-speech and audio representation learning at foundation-model scale. This repository holds a data folder with one subfolder per language code (104 of them, for example data/en with audio and metadata subfolders) (measured 2026-11-14; the Hub reports about 56 TB). Stream it with the datasets library or read the files directly from the mounted repo; see the card for per-language hours. Described in an Interspeech 2026 paper. The publisher states the audio was released under CC BY 3.0; attribute the dataset and the original creators.","displayTitle":"YODAS v3: 1.1-million-hour multilingual web speech corpus","id":"8f123c47-5cb4-502f-9a88-545700d04797","license":"CC-BY-3.0","licenseEvidenceUrl":"https://huggingface.co/datasets/espnet/yodas3","objectCountEstimate":13373,"probeStatus":"ok","referenceUrl":"https://www.isca-archive.org/interspeech_2026/chen26d_interspeech.html","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"espnet/yodas3\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"espnet/yodas3\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=espnet, license:cc-by-3.0, size_categories:1M<n<10M, format:parquet, downloads=259034"],"format":"mixed","generatedAt":"2026-11-14T14:15:20.205Z","layout":{"example":"data/ab/metadata/0000.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"5041523115041548155a4c1502150012000028b52ffd20242101002000000032626536363530383739666239626664303061343736623434366465383064611500151215242c15021510150615061c36002820326265363635303837396662396266643030613437366234343664653830646118203262653636353038373966623962666430306134373662343436646538306461111100000028b52ffd20094900000200000002010102001504151015224c1502150012000028b52ffd200841000004000000303030301500151215242c15021510150615061c3600280430303030180430303030111100000028b52ffd20094900000200000002010102001504155c156e4c1502150012000028b52ffd202e7101002a0000005b7b227374617274223a302c226475726174696f6e223a323030302c2274657874223a22555341227d5d1500151215242c15021510150615061c3600282a5b7b227374617274223a302c226475726174696f6e223a323030302c2274657874223a22555341227d5d182a5b7b227374617274223a302c226475726174696f6e223a323030302c2274657874223a22555341227d5d111100000028b52ffd20094900000200000002010102001504150015124c1500150012000028b52ffd20000100001500150e15202c15021510150615061c360200000028b52ffd20073900000200000002","encoding":"hex","key":"data/ab/metadata/0000.parquet","truncatedAt":512}],"sampleObjects":["data/ab/audio/0000.tar","data/ab/metadata/0000.parquet","data/af/audio/0000.tar","data/af/audio/0001.tar","data/af/metadata/0000.parquet"],"sidecars":[{"bytes":"---\nlicense: cc-by-3.0\ntask_categories:\n- audio-to-audio\n- automatic-speech-recognition\n- text-to-speech\n- translation\ndataset_info:\n- config_name: preview\n  features:\n  - name: audio\n    dtype: audio\n  - name: lang\n    dtype: string\n  - name: id\n    dtype: string\n  - name: shard\n    dtype: string\n  - name: transcript\n    dtype: string\n  - name: translation_en\n    dtype: string\n  - name: transcript_lang\n    dtype: string\n  - name: locale_lang\n    dtype: string\n  - name: length\n    dtype: float64\n  - name: audio_bitrate\n    dtype: float64\n  - name: max_freq_hz\n    dtype: float64\n  - name: estimated_bandwidth_hz\n    dtype: float64\n  - name: channels\n    dtype: int32\n  - name: n_distinct_channels\n    dtype: int32\n  - name: preview_audio_seconds\n    dtype: float64\nconfigs:\n- config_name: \"preview\"\n  default: true\n  data_files:\n  - split: train\n    path: data/preview/*.parquet\n- config_name: \"ab_metadata\"\n  data_files:\n  - split: train\n    path: data/ab/metadata/*.parquet\n- config_name: \"af_metadata\"\n  data_files:\n  - split: train\n    path: data/af/metadata/*.parquet\n- config_name: \"am_metadata\"\n  data_files:\n  - split: train\n    path: data/am/metadata/*.parquet\n- config_name: \"ar_metadata\"\n  data_files:\n  - split: train\n    path: data/ar/metadata/*.parquet\n- config_name: \"as_metadata\"\n  data_files:\n  - split: train\n    path: data/as/metadata/*.parquet\n- config_name: \"ay_metadata\"\n  data_files:\n  - split: train\n    path: data/ay/metadata/*.parquet\n- config_name: \"az_metadata\"\n  data_files:\n  - split: train\n    path: data/az/metadata/*.parquet\n- config_name: \"be_metadata\"\n  data_files:\n  - split: train\n    path: data/be/metadata/*.parquet\n- config_name: \"bg_metadata\"\n  data_files:\n  - split: train\n    path: data/bg/metadata/*.parquet\n- config_name: \"bn_metadata\"\n  data_files:\n  - split: train\n    path: data/bn/metadata/*.parquet\n- config_name: \"bo_metadata\"\n  data_files:\n  - split: train\n    path: data/bo/metadata/*.parquet\n- config_name: \"bs_metadata\"\n  data_files:\n  - split: train\n    path: data/bs/metadata/*.parquet\n- config_name: \"ca_metadata\"\n  data_files:\n  - split: train\n    path: data/ca/metadata/*.parquet\n- config_name: \"cr_metadata\"\n  data_files:\n  - split: train\n    path: data/cr/metadata/*.parquet\n- config_name: \"cs_metadata\"\n  data_files:\n  - split: train\n    path: data/cs/metadata/*.parquet\n- config_name: \"cy_metadata\"\n  data_files:\n  - split: train\n    path: data/cy/metadata/*.parquet\n- config_name: \"da_metadata\"\n  data_files:\n  - split: train\n    path: data/da/metadata/*.parquet\n- config_name: \"de_metadata\"\n  data_files:\n  - split: train\n    path: data/de/metadata/*.parquet\n- config_name: \"el_metadata\"\n  data_files:\n  - split: train\n    path: data/el/metadata/*.parquet\n- config_name: \"en_metadata\"\n  data_files:\n  - split: train\n    path: data/en/metadata/*.parquet\n- config_name: \"eo_metadata\"\n  data_files:\n  - split: train\n    path: data/eo/metadata/*.parquet\n- config_name: \"es_metadata\"\n  data_files:\n  - split: train\n    path: data/es/metadata/*.parquet\n- config_name: \"et_metadata\"\n  data_files:\n  - split: train\n    path: data/et/metadata/*.parquet\n- config_name: \"eu_metadata\"\n  data_files:\n  - split: train\n    path: data/eu/metadata/*.parquet\n- config_name: \"fa_metadata\"\n  data_files:\n  - split: train\n    path: data/fa/metadata/*.parquet\n- config_name: \"fi_metadata\"\n  data_files:\n  - split: train\n    path: data/fi/metadata/*.parquet\n- config_name: \"fil_metadata\"\n  data_files:\n  - split: train\n    path: data/fil/metadata/*.parquet\n- config_name: \"fo_metadata\"\n  data_files:\n  - split: train\n    path: data/fo/metadata/*.parquet\n- config_name: \"fr_metadata\"\n  data_files:\n  - split: train\n    path: data/fr/metadata/*.parquet\n- config_name: \"gl_metadata\"\n  data_files:\n  - split: train\n    path: data/gl/metadata/*.parquet\n- config_name: \"gn_metadata\"\n  data_files:\n  - split: train\n    path: data/gn/metadata/*.parquet\n- config_name: \"gu_metadata\"\n  data_files:\n  - split: train\n    path: data/gu/metadata/*.parquet\n- config_name: \"ha_metadata\"\n  data_files:\n  - split: train\n    path: data/ha/metadata/*.parquet\n- config","encoding":"utf8","key":"README.md","name":"README.md","size":22928,"truncated":true}],"sourceObjectKey":"data/ab/metadata/0000.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"espnet/yodas3\")\nprint(ds)","starterSamplePath":"data/ab/metadata/0000.parquet","totalObjects":13373},"summary":"YODAS v3 is a large web-crawled speech dataset of over 1.1 million hours of audio in over 100 languages, all originally released under a CC-BY-3.0 licence, published by the ESPnet…","title":"YODAS v3: 1.1-million-hour multilingual web speech corpus","updatedAt":"2026-11-14T17:36:26.031Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/espnet/yodas3"}}