arXiv cs.CL· Zhuoyan Tao, Aditya Kommineni, Shrikanth Narayanan·· 10 小时前AI 评分18
Off-target Posterior Mass：一种免训练的语音语种分割微调探针收益诊断指标
Off-target posterior mass: A training-free diagnostic for fine-tuning spoken language diarization probes
AI 导读
研究提出 Off-target Posterior Mass（OTPM），一种基于后验语种识别概率计算的免训练诊断指标，用于预测语音语种分割（SLD）微调带来的性能增益。OTPM 与探针增益呈正相关，在未见过的合成双语数据上相关系数为 0.56，在真实语料上为 0.23。
正文 · 原文
Abstract:Spoken Language Diarization (SLD) models provide frame-level language information of a speech recording. SLD modeling pipeline includes feature extraction from pre-trained speech models followed by clustering. Fine-tuning pre-trained speech models has shown performance gains for SLD with its magnitude being dataset dependent. Off-target Posterior Mass (OTPM), computed based on posterior language identification probabilities is proposed as a diagnostic metric to forecast performance gains from SLD fine-tuning. OTPM obtains positive correlation with probe gains, 0.56 on unseen synthetic bilingual data and 0.23 on real-world corpora. These results indicate promising potential of OTPM as a diagnostic metric for SLD.
| Comments: | 10 pages, 2 figures, 3 tables. Submitted to JASA Express Letters |
| Subjects: | Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD) |
| Cite as: | arXiv:2611.13430 [eess.AS] |
| (or arXiv:2611.13430v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13430 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhuoyan Tao [view email]
[v1]
Tue, 10 Nov 2026 01:15:31 UTC (105 KB)
来源：arXiv cs.CL · arxiv.org