arXiv cs.CL· Minu Kim, Jihwan Lee, David R. Mortensen, Shrikanth Narayanan·· 10 天前AI 评分21
如何消除自监督语音表征中的口音-语言混淆以提升语言识别
Mitigating Accent-Language Confusion in Self-Supervised Speech Representations for Language Identification
AI 导读
针对自监督语音表征微调的 LID 模型常把非母语（L2）口音误判为说话人母语（L1）的问题，研究者提出一种几何投影方法，仅用母语语音估计 L1 偏置方向，并在冻结的 LID 头之前将其移除。在五个 MMS-LID 模型和非母语语料上，该方法显著提升了对 L2 口音语音的目标语言识别，同时保持母语语音预测不变，且无需 L2 训练数据或模型适配。
正文 · 原文
Abstract:Spoken language identification (LID) aims to recognize the target language regardless of accent. In practice, however, LID models fine-tuned from self-supervised speech representations frequently confuse accents with languages, misclassifying non-native (L2) speech as the speaker's first language (L1). We show that non-native speech representations lie between native target-language and native L1 poles, causing systematic misclassification. To address this, we introduce a geometric projection that estimates an L1-bias direction solely from native speech and removes it before the frozen LID head. Across five MMS-LID models and non-native corpora, this projection substantially improves target language identification for L2-accented speech while preserving predictions for native speech. These results show that accent-induced L1 bias can be corrected directly within the representation space without L2 training data or model adaptation.
| Comments: | Submitted to ICASSP 2027 |
| Subjects: | Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD) |
| Cite as: | arXiv:2610.09486 [eess.AS] |
| (or arXiv:2610.09486v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09486 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Minu Kim [view email]
[v1]
Wed, 7 Oct 2026 05:42:40 UTC (351 KB)
来源：arXiv cs.CL · arxiv.org