Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.05813 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912946489131008 |
|---|---|
| author | Sun, Jinuo Xiao, Yang Chung, Sung Kyun Hu, Qiuchi Huang, Gongping Holden, Eun-Jung Dang, Ting |
| author_facet | Sun, Jinuo Xiao, Yang Chung, Sung Kyun Hu, Qiuchi Huang, Gongping Holden, Eun-Jung Dang, Ting |
| contents | Accent variability remains a major errors in automatic speech recognition, yet most adaptation methods rely on parameter fine-tuning without understanding where accent information is encoded. We treat accent variation as an interpretable subspace in hidden representations and investigate whether it can be identified and controlled directly in activation space. We extract layer-wise encoder activations and estimate mean-shift directions capturing accent-induced representation shifts. By injecting these directions into individual layers and measuring how they align accented and standard embeddings, we derive a layer-wise accent sensitivity profile, revealing that accent information concentrates in a narrow band of middle encoder layers. Leveraging this structure, we further introduce parameter-free accent steering that modifies representations during inference without updating model weights. Experiments across eight accents show consistent word error rate reductions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_05813 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Activation Steering for Accent Adaptation in Speech Foundation Models Sun, Jinuo Xiao, Yang Chung, Sung Kyun Hu, Qiuchi Huang, Gongping Holden, Eun-Jung Dang, Ting Audio and Speech Processing Accent variability remains a major errors in automatic speech recognition, yet most adaptation methods rely on parameter fine-tuning without understanding where accent information is encoded. We treat accent variation as an interpretable subspace in hidden representations and investigate whether it can be identified and controlled directly in activation space. We extract layer-wise encoder activations and estimate mean-shift directions capturing accent-induced representation shifts. By injecting these directions into individual layers and measuring how they align accented and standard embeddings, we derive a layer-wise accent sensitivity profile, revealing that accent information concentrates in a narrow band of middle encoder layers. Leveraging this structure, we further introduce parameter-free accent steering that modifies representations during inference without updating model weights. Experiments across eight accents show consistent word error rate reductions. |
| title | Activation Steering for Accent Adaptation in Speech Foundation Models |
| topic | Audio and Speech Processing |
| url | https://arxiv.org/abs/2603.05813 |