When Forgetting Looks Like Improvement: Metric Masking in Streaming Diarizer Adaptation and the Price of Rehearsal

Computer Science > Computation and Language arXiv:2610.08828 (cs) [Submitted on 26 Sep 2026] Title: When Forgetting Looks Like Improvement: Metric Masking in Streaming Diarizer Adaptation and the Price of Rehearsal Authors: Mo Yu, Yang Liu, Jing Qian.

计算机科学 > 计算与语言 arXiv:2610.08828 (cs) [提交于 2026 年 9 月 26 日] 标题：当遗忘看起来像改进：流式说话人日志自适应中的指标掩盖与复述的代价 作者：Mo Yu, Yang Liu, Jing Qian。

Abstract: Small-data adaptation can improve speech detection while degrading speaker attribution. We study this discrepancy in a released streaming diarizer adapted on 7.5 h of two-party conversation and evaluated across six corpora.

摘要：小数据自适应可以改善语音检测，但会降低说话人归属的准确性。我们研究了这一差异，对象是一个在 7.5 小时双人对话上进行自适应并在六个语料库中进行评估的已发布流式说话人日志系统。

Adaptation substantially improves in-domain diarization performance and transfers to an independent corpus. However, this improvement is not consistent across evaluation scenarios as the additional confusion is mainly associated with impaired temporal identity consistency rather than speaker-count errors.

自适应显著提高了域内日志性能，并可迁移至独立语料库。然而，这种改进在不同评估场景中并不一致，因为额外的混淆主要与时间身份一致性的受损有关，而非说话人数量错误。

A local-remapping diagnostic reveals different patterns of identity degradation across corpora, indicating that adaptation may alter how streaming models maintain speaker assignments over time. Rehearsal reduces the observed degradation but reduces the cross-domain transfer performance.

局部重映射诊断揭示了跨语料库中身份退化的不同模式，表明自适应可能会改变流式模型随时间维持说话人分配的方式。复述减少了观察到的退化，但降低了跨域迁移性能。

These results highlight the need to jointly evaluate detection accuracy, identity consistency, and retention behavior when adapting streaming diarization systems.

这些结果强调了在自适应流式说话人日志系统时，需要联合评估检测准确性、身份一致性和保留行为。