Automatic Speech Recognition (ASR) has become a key technology for human--AI interaction. However, code-switching ASR (CS-ASR) remains particularly challenging due to the severe scarcity of multilingual CS speech resources across diverse language pairs. Existing approaches primarily improve CS-ASR performance through synthetic CS speech generation or pair-specific fine-tuning on limited bilingual datasets. Nevertheless, these approaches face an inherent scalability limitation, as support for CS must be developed separately for language pairs whose number grows combinatorially with the number of supported languages. In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. Our experiments show that merged bilingual CS-ASR models modestly generalize to unseen language pairs, suggesting limited transfer of bilingual CS capabilities across language pairs.</p>\n","updatedAt":"2026-06-05T04:43:10.735Z","author":{"_id":"66976a7fa7fd582ae754850e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/jQXfIqKh1FYGxjyFwSmSY.jpeg","fullname":"Gio Paik","name":"skyil7","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false,"primaryOrg":{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/4r_irltP9kCKVSVzlOj2j.jpeg","fullname":"Theta One AI","name":"thetaone-ai","type":"org","isHf":false,"details":"AI for Education and Communication","plan":"team"}}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8751175999641418},"editors":["skyil7"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/jQXfIqKh1FYGxjyFwSmSY.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.05846","authors":[{"_id":"6a2253aa3490a593e87b159f","name":"Gio Paik","hidden":false},{"_id":"6a2253aa3490a593e87b15a0","name":"Hyunseo Shin","hidden":false},{"_id":"6a2253aa3490a593e87b15a1","name":"Soungmin Lee","hidden":false}],"publishedAt":"2026-06-04T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"Towards Truly Multilingual ASR: Generalizing Code-Switching ASR to Unseen Language Pairs","submittedOnDailyBy":{"_id":"66976a7fa7fd582ae754850e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/jQXfIqKh1FYGxjyFwSmSY.jpeg","isPro":true,"fullname":"Gio Paik","user":"skyil7","type":"user","name":"skyil7"},"summary":"Automatic Speech Recognition (ASR) has become a key technology for human--AI interaction. However, code-switching ASR (CS-ASR) remains particularly challenging due to the severe scarcity of multilingual CS speech resources across diverse language pairs. Existing approaches primarily improve CS-ASR performance through synthetic CS speech generation or pair-specific fine-tuning on limited bilingual datasets. Nevertheless, these approaches face an inherent scalability limitation, as support for CS must be developed separately for language pairs whose number grows combinatorially with the number of supported languages. In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. Our experiments show that merged bilingual CS-ASR models modestly generalize to unseen language pairs, suggesting limited transfer of bilingual CS capabilities across language pairs.","upvotes":2,"discussionId":"6a2253aa3490a593e87b15a2","ai_summary":"Code-switching automatic speech recognition models show limited generalization across unseen language pairs despite attempts at model merging and domain generalization techniques.","ai_keywords":["automatic speech recognition","code-switching ASR","multilingual speech resources","synthetic CS speech generation","pair-specific fine-tuning","domain generalization","model merging"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"68b1775f3eded3884d27cbd3","name":"thetaone-ai","fullname":"Theta One AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/4r_irltP9kCKVSVzlOj2j.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"66976a7fa7fd582ae754850e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/jQXfIqKh1FYGxjyFwSmSY.jpeg","isPro":true,"fullname":"Gio Paik","user":"skyil7","type":"user"},{"_id":"65dc040952eca001fd0bb142","avatarUrl":"/avatars/cd8e54ceef7c9e4a3bb4b0900c47a8b6.svg","isPro":false,"fullname":"Mingxuan Xia","user":"MingxuanXia","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"68b1775f3eded3884d27cbd3","name":"thetaone-ai","fullname":"Theta One AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66976a7fa7fd582ae754850e/4r_irltP9kCKVSVzlOj2j.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.05846.md"}">
Towards Truly Multilingual ASR: Generalizing Code-Switching ASR to Unseen Language Pairs
Abstract
Code-switching automatic speech recognition models show limited generalization across unseen language pairs despite attempts at model merging and domain generalization techniques.
Automatic Speech Recognition (ASR) has become a key technology for human--AI interaction. However, code-switching ASR (CS-ASR) remains particularly challenging due to the severe scarcity of multilingual CS speech resources across diverse language pairs. Existing approaches primarily improve CS-ASR performance through synthetic CS speech generation or pair-specific fine-tuning on limited bilingual datasets. Nevertheless, these approaches face an inherent scalability limitation, as support for CS must be developed separately for language pairs whose number grows combinatorially with the number of supported languages. In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. Our experiments show that merged bilingual CS-ASR models modestly generalize to unseen language pairs, suggesting limited transfer of bilingual CS capabilities across language pairs.
Community
Automatic Speech Recognition (ASR) has become a key technology for human--AI interaction. However, code-switching ASR (CS-ASR) remains particularly challenging due to the severe scarcity of multilingual CS speech resources across diverse language pairs. Existing approaches primarily improve CS-ASR performance through synthetic CS speech generation or pair-specific fine-tuning on limited bilingual datasets. Nevertheless, these approaches face an inherent scalability limitation, as support for CS must be developed separately for language pairs whose number grows combinatorially with the number of supported languages. In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. Our experiments show that merged bilingual CS-ASR models modestly generalize to unseen language pairs, suggesting limited transfer of bilingual CS capabilities across language pairs.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.05846 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.05846 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.05846 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.