arXiv cs.CL· Shashi Kumar, Esa\'u Villatoro-Tello, Thibault Ba\~neras-Roux, Sergio Burdisso, Petr Motlicek, Kadri Hacioglu, Andreas Stolcke·· 12 小时前AI 评分18
面向因子化 Transducer ASR 纯文本域适应的残差专家路由
Residual Expert Routing for Text-Only Domain Adaptation of Factorized Transducer ASR
AI 导读
针对因子化 Transducer ASR 仅用目标域文本微调非空白预测器会提升目标域却损害源域的问题，研究者提出保守残差路由：把适配后的预测器作为基础预测器的残差修正，由 16k 参数的门控在每个前缀决定接纳多少，门控关闭即完全还原基础模型。
正文 · 原文
Abstract:Fine-tuning the non-blank predictor of a factorized transducer on target-domain text improves recognition on that domain but degrades performance on the source domain. A single interpolation weight between base and adapted predictors cannot resolve this: the adapted predictor helps at some prefixes and hurts at others, often within one utterance. We propose conservative residual routing: the adapted predictor becomes a residual correction to the base, and a 16k-parameter gate decides at each prefix how much of it to admit. Closing the gate reproduces the base exactly. On a 40-hour conversational corpus, routing retains essentially all of the text-only gain on two held-out domains (19.89 to 16.74, 20.91 to 16.21 WER) while leaving source-domain WER unchanged (12.90 to 12.93); the adapted predictor alone costs 1.08 WER. Routing needs no per-domain tuning on target-domain audio. Analysis reveals the gate does not behave as a simple domain-level switch: it modulates intervention through the disagreement between the two predictors, suppressing it in-domain and amplifying it out-of-domain.
| Comments: | Paper submitted to ICASSP 2027 (Under review) |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.32637 [cs.CL] |
| (or arXiv:2610.32637v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.32637 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Esaú Villatoro-Tello [view email]
[v1]
Fri, 23 Oct 2026 13:48:33 UTC (44 KB)
来源：arXiv cs.CL · arxiv.org