\n\t\t\n\t\n\t\n\t\tWhisper: Revolutionizing Speech Recognition with Weak Supervision\n\t\n\n

\n\n

\n\t\n\t\t\n\t\n\t\n\t\tLinks 🔗:\n\t\n

👉 Subscribe: https://www.youtube.com/@Arxflix
👉 Twitter: https://x.com/arxflix
👉 LMNT (Partner): https://lmnt.com/

\n","updatedAt":"2024-06-09T07:23:19.745Z","author":{"_id":"6186ddf6a7717cb375090c01","avatarUrl":"/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg","fullname":"Julien BLANCHON","name":"blanchon","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":176,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.532935619354248},"editors":["blanchon"],"editorAvatarUrls":["/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg"],"reactions":[],"isReport":false}},{"id":"66ca8c0ddc312dec4337bbb8","author":{"_id":"66755810653173016769aa85","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66755810653173016769aa85/Kum5PT-VqJcXQ8UwqUSW-.png","fullname":"bmmfr","name":"vbmhrr","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"createdAt":"2024-08-25T01:42:37.000Z","type":"comment","data":{"edited":true,"hidden":true,"hiddenBy":"","hiddenReason":"Off-Topic","latest":{"raw":"This comment has been hidden","html":"This comment has been hidden","updatedAt":"2025-03-12T08:24:06.320Z","author":{"_id":"66755810653173016769aa85","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66755810653173016769aa85/Kum5PT-VqJcXQ8UwqUSW-.png","fullname":"bmmfr","name":"vbmhrr","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"editors":[],"editorAvatarUrls":[],"reactions":[]}},{"id":"66ca8c1de0cdc14fc0629fa0","author":{"_id":"66755810653173016769aa85","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66755810653173016769aa85/Kum5PT-VqJcXQ8UwqUSW-.png","fullname":"bmmfr","name":"vbmhrr","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"createdAt":"2024-08-25T01:42:53.000Z","type":"comment","data":{"edited":true,"hidden":true,"hiddenBy":"","latest":{"raw":"This comment has been hidden","html":"This comment has been hidden","updatedAt":"2024-08-26T06:44:18.821Z","author":{"_id":"66755810653173016769aa85","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66755810653173016769aa85/Kum5PT-VqJcXQ8UwqUSW-.png","fullname":"bmmfr","name":"vbmhrr","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"editors":[],"editorAvatarUrls":[],"reactions":[]}},{"id":"66ff8d3151e6d8bfb21a2c0c","author":{"_id":"65881053c4b20046638b1397","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65881053c4b20046638b1397/KuF67dIMju8vEWuWjpZ8Q.jpeg","fullname":"erol ural","name":"ucanzii","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":2,"isUserFollowing":false},"createdAt":"2024-10-04T06:37:37.000Z","type":"comment","data":{"edited":true,"hidden":true,"hiddenBy":"","latest":{"raw":"This comment has been hidden","html":"This comment has been hidden","updatedAt":"2024-11-11T09:52:33.796Z","author":{"_id":"65881053c4b20046638b1397","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65881053c4b20046638b1397/KuF67dIMju8vEWuWjpZ8Q.jpeg","fullname":"erol ural","name":"ucanzii","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":2,"isUserFollowing":false}},"numEdits":0,"editors":[],"editorAvatarUrls":[],"reactions":[]}},{"id":"67edc4316a9812afb63f2fe0","author":{"_id":"67e527e5fe1f5acc682cab94","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/IdtMDXkwRTTf5QbGYEJ3T.png","fullname":"GUY NTUMBA CHISSUNGU","name":"AIGLE243","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"createdAt":"2025-04-02T23:11:45.000Z","type":"comment","data":{"edited":true,"hidden":true,"hiddenBy":"","hiddenReason":"Off-Topic","latest":{"raw":"This comment has been hidden","html":"This comment has been hidden","updatedAt":"2025-12-04T08:48:48.148Z","author":{"_id":"67e527e5fe1f5acc682cab94","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/IdtMDXkwRTTf5QbGYEJ3T.png","fullname":"GUY NTUMBA CHISSUNGU","name":"AIGLE243","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":1,"editors":[],"editorAvatarUrls":[],"reactions":[]}},{"id":"68b84825b1c73413f5bf109d","author":{"_id":"682a4c10ba968b2e325d9178","avatarUrl":"/avatars/6e3c01fae5d5a5a8bc4ade90a67cc659.svg","fullname":"hsn abodzal","name":"Hasasn76","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"createdAt":"2025-09-03T13:52:37.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"https://termius.com/download","html":"

https://termius.com/download

\n","updatedAt":"2025-09-03T13:52:37.558Z","author":{"_id":"682a4c10ba968b2e325d9178","avatarUrl":"/avatars/6e3c01fae5d5a5a8bc4ade90a67cc659.svg","fullname":"hsn abodzal","name":"Hasasn76","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7070923447608948},"editors":["Hasasn76"],"editorAvatarUrls":["/avatars/6e3c01fae5d5a5a8bc4ade90a67cc659.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2212.04356","authors":[{"_id":"6411c77f6b75ddced3890ceb","name":"Alec Radford","hidden":false},{"_id":"6411c77f6b75ddced3890cec","name":"Jong Wook Kim","hidden":false},{"_id":"6411c77f6b75ddced3890ced","name":"Tao Xu","hidden":false},{"_id":"6411c77f6b75ddced3890cee","name":"Greg Brockman","hidden":false},{"_id":"6411c77f6b75ddced3890cef","name":"Christine McLeavey","hidden":false},{"_id":"6411c77f6b75ddced3890cf0","name":"Ilya Sutskever","hidden":false}],"publishedAt":"2022-12-06T18:46:04.000Z","title":"Robust Speech Recognition via Large-Scale Weak Supervision","summary":"We study the capabilities of speech processing systems trained simply to\npredict large amounts of transcripts of audio on the internet. When scaled to\n680,000 hours of multilingual and multitask supervision, the resulting models\ngeneralize well to standard benchmarks and are often competitive with prior\nfully supervised results but in a zero-shot transfer setting without the need\nfor any fine-tuning. When compared to humans, the models approach their\naccuracy and robustness. We are releasing models and inference code to serve as\na foundation for further work on robust speech processing.","upvotes":49,"discussionId":"641192363ea54b1aa7e2f439","ai_summary":"Large-scale speech models trained on extensive multilingual audio data achieve high accuracy and robustness in zero-shot tasks without fine-tuning.","ai_keywords":["speech processing systems","zero-shot transfer","multilingual","multitask supervision","audio data"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"620630b603825909dcbeba35","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620630b603825909dcbeba35/vNlOtJqdcP3vpAfeHfNvP.jpeg","isPro":true,"fullname":"Aaron C Wacker","user":"awacke1","type":"user"},{"_id":"63358001686c20e55973298d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1665133631770-63358001686c20e55973298d.png","isPro":false,"fullname":"Mathias Nielsen","user":"mathiasn1","type":"user"},{"_id":"65010ddf1e14749e840ace51","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/9sr55FlCsLH7d7f7s-_pI.png","isPro":false,"fullname":"Andrew Zhuang","user":"youraveragedev","type":"user"},{"_id":"644accc5958b779698092c4f","avatarUrl":"/avatars/973db41f23cd375c588e7828c5efe5ce.svg","isPro":false,"fullname":"Pierric","user":"pierrci","type":"user"},{"_id":"5fcfb7c407408029ba3577e2","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1608146735109-5fcfb7c407408029ba3577e2.png","isPro":false,"fullname":"Simon Brandeis","user":"sbrandeis","type":"user"},{"_id":"63d9ce35aefaa735258607e2","avatarUrl":"/avatars/b93d4acb519a8bebe14494ebe113b0b8.svg","isPro":false,"fullname":"Yiming Zheng","user":"ZYM666","type":"user"},{"_id":"63d406b8ff1384ce6c6007fe","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674839690440-noauth.png","isPro":false,"fullname":"Shan KS","user":"ShanmugaKS","type":"user"},{"_id":"64c7ea7db8685df800204a74","avatarUrl":"/avatars/1ab5ec264b1421fbaa44e1b2b58cf09b.svg","isPro":false,"fullname":"Yash Pethe","user":"fierce74","type":"user"},{"_id":"63a41c04de134926a2b5817c","avatarUrl":"/avatars/04d1003c35d27594370d96bcb15df766.svg","isPro":false,"fullname":"Durgesh Patel","user":"Durgesh123","type":"user"},{"_id":"64a7e6332431ebd64dd3d54e","avatarUrl":"/avatars/4e788c18e5ab52f19a27f6e524f61eed.svg","isPro":false,"fullname":"adnan","user":"adnankarim","type":"user"},{"_id":"6538119803519fddb4a17e10","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6538119803519fddb4a17e10/ffJMkdx-rM7VvLTCM6ri_.jpeg","isPro":false,"fullname":"samusenps","user":"samusenps","type":"user"},{"_id":"62661fbd5909b93b24d9901f","avatarUrl":"/avatars/92bddaeae968ab78ff462ab2a313e1c8.svg","isPro":false,"fullname":"Jaehee Lee","user":"ij5","type":"user"}],"acceptLanguages":["*"]}">

Papers

arxiv:2212.04356

Robust Speech Recognition via Large-Scale Weak Supervision

Published on Dec 6, 2022

Upvote

Authors:

Abstract

Large-scale speech models trained on extensive multilingual audio data achieve high accuracy and robustness in zero-shot tasks without fine-tuning.

AI-generated summary

We study the capabilities of speech processing systems trained simply to predict large amounts of transcripts of audio on the internet. When scaled to 680,000 hours of multilingual and multitask supervision, the resulting models generalize well to standard benchmarks and are often competitive with prior fully supervised results but in a zero-shot transfer setting without the need for any fine-tuning. When compared to humans, the models approach their accuracy and robustness. We are releasing models and inference code to serve as a foundation for further work on robust speech processing.

View arXiv page View PDF Add to collection