{"success":true,"data":{"author":"Xinjian Li, Shinnosuke Takamichi, Takaaki Saeki et al. (CMU WAVLab, ESPnet)","bucketAccessType":"public","bucketName":"espnet/yodas2","bucketPrefix":"","category":"Speech & Audio","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:17.686Z","description":"YODAS (YouTube-Oriented Dataset for Audio and Speech) is a large-scale multilingual speech dataset of over 500,000 hours in more than 100 languages, collected only from YouTube videos carrying a Creative Commons licence. It has labelled subsets with manual or automatic subtitles for supervised ASR training and unlabelled subsets for self-supervised learning. YODAS2 provides the same data as espnet/yodas in long-form, video-level format: audio is not segmented and is encoded at 24 kHz instead of 16 kHz, and each record carries the whole video's audio, its duration and a list of utterances with text and start and end timestamps. It is one of the largest open multilingual ASR corpora. This repository holds a data folder with one subfolder per language shard (230 of them, such as en000, each with audio, duration and text subfolders) and a SQLite metadata file (measured 2026-11-14; the Hub reports about 60 TB). Load a subset with datasets.load_dataset('espnet/yodas2', 'en000', streaming=True). The YODAS paper lists all subsets as CC BY 3.0, the licence of the source videos; attribute the dataset and, where needed, the original channels.","displayTitle":"YODAS2: YouTube-Oriented Dataset for Audio and Speech (long-form)","id":"afb85e44-0aa4-5778-82a8-8e10bb8e69f7","license":"CC-BY-3.0","licenseEvidenceUrl":"https://arxiv.org/pdf/2406.00899","objectCountEstimate":100000,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/2406.00899","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"espnet/yodas2\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"espnet/yodas2\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=espnet, license:cc-by-3.0, downloads=176499"],"format":"mixed","generatedAt":"2026-11-14T14:15:26.768Z","layout":{"example":"data/aa000/audio/00000000.tar.gz","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/aa000/audio/00000000.tar.gz","truncatedAt":1536}],"sampleObjects":["data/aa000/audio/00000000.tar.gz","data/aa000/audio/00000001.tar.gz","data/aa000/duration/00000000.txt","data/aa000/duration/00000001.txt","data/aa000/text/00000000.json"],"sidecars":[{"bytes":"---\nlicense: cc-by-3.0\n---\n\nYODAS2 is the long-form dataset from YODAS dataset.\n\nIt provides the same dataset as [espnet/yodas](https://huggingface.co/datasets/espnet/yodas)  but YODAS2 has the following new features:\n\n- formatted in the long-form (video-level) where audios are not segmented.\n- audios are encoded using higher sampling rates (i.e. 24k)\n\nFor detailed information about YODAS dataset, please refer to [our paper](https://arxiv.org/abs/2406.00899) and the [espnet/yodas repo](https://huggingface.co/datasets/espnet/yodas).\n\n## Usage:\n\n\nEach data point corresponds to an entire video on YouTube, it contains the following fields:\n\n- video_id: unique id of this video (note this id is not the video_id in Youtube)\n- duration: total duration in seconds of this video\n- audio\n  - path: local path to wav file if in standard mode, otherwise empty in the streaming mode\n  - sampling_rate: fixed to be 24k. (note that the sampling rate in `espnet/yodas` is 16k)\n  - array: wav samples in float\n- utterances\n  - utt_id: unique id of this utterance\n  - text: transcription of this utterance\n  - start: start timestamp in seconds of this utterance\n  - end: end timestamp in seconds of this utterance\n\nYODAS2 also supports two modes:\n\n**standard mode**: each subset will be downloaded to the local dish before first iterating. \n\n```python\nfrom datasets import load_dataset\n\n# Note this will take very long time to download and preprocess\n# you can try small subset for testing purpose\nds = load_dataset('espnet/yodas2', 'en000')\nprint(next(iter(ds['train'])))\n```\n\n**streaming mode** most of the files will be streamed instead of downloaded to your local deivce. It can be used to inspect this dataset quickly.\n\n```python\nfrom datasets import load_dataset\n\n# this streaming loading will finish quickly\nds = load_dataset('espnet/yodas2', 'en000', streaming=True)\n```\n\n\n## Reference\n\n```\n@inproceedings{li2023yodas,\n  title={Yodas: Youtube-Oriented Dataset for Audio and Speech},\n  author={Li, Xinjian and Takamichi, Shinnosuke and Saeki, Takaaki and Chen, William and Shiota, Sayaka and Watanabe, Shinji},\n  booktitle={2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)},\n  pages={1--8},\n  year={2023},\n  organization={IEEE}\n}\n```\n\n\n## Contact\n\nIf you have any questions, feel free to contact us at the following email address.\n\nWe made sure that our dataset only consisted of videos with CC licenses during our downloading. But in case you find your video unintentionally included in our dataset and would like to delete it, you can send a delete request to the following email.\n\nRemove the parenthesis `()` from the following email address\n\n`(lixinjian)(1217)@gmail.com`\n\n\n","encoding":"utf8","key":"README.md","name":"README.md","size":2700,"truncated":false}],"sourceObjectKey":"data/aa000/audio/00000000.tar.gz","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"espnet/yodas2\", filename=\"data/aa000/audio/00000000.tar.gz\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"data/aa000/duration/00000000.txt","totalObjects":100000},"summary":"YODAS (YouTube-Oriented Dataset for Audio and Speech) is a large-scale multilingual speech dataset of over 500,000 hours in more than 100 languages, collected only from YouTube…","title":"YODAS2: YouTube-Oriented Dataset for Audio and Speech (long-form)","updatedAt":"2026-11-14T17:33:19.416Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/espnet/yodas2"}}