An Investigation of Reprogramming for Cross-Language Adaptation in Speaker Verification Systems
Li, Chiu, Lee · eess.AS,cs.SD · 2026-09-05 · 原文
Language mismatch is among the most common and challenging domain mismatches in deploying speaker verification (SV) systems. Adversarial reprogramming has shown promising results in cross-language adaptation for SV. The reprogramming is implemented by padding learnable parameters on the two sides of input speech signals. In this paper, we investigate the relationship between the number of padded parameters and the performance of the reprogrammed models. Sufficient experiments are conducted with different scales of SV models and datasets. The results demonstrate that reprogramming consistently improves the performance of cross-language SV, while the improvement is saturated or even degraded when using larger padding lengths. The performance is mainly determined by the capacity of the original SV models instead of the number of padded parameters. The SV models with larger scales have higher upper bounds in performance and can endure longer padding without performance degradation.
讲义
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
Language mismatch is among the most common and challenging domain mismatches in deploying speaker verification (SV) systems.
Adversarial reprogramming has shown promising results in cross-language adaptation for SV.
2. 领域脉络
本文类目：eess.AS、cs.SD，属于其所在研究脉络的最新进展。
3. 机制拆解
The reprogramming is implemented by padding learnable parameters on the two sides of input speech signals.
In this paper, we investigate the relationship between the number of padded parameters and the performance of the reprogrammed models.
Sufficient experiments are conducted with different scales of SV models and datasets.
4. 证据与数字
摘要未给出量化结果——留意原文的实验与数据。
5. 反例与边界
摘要未声明局限与反例——这是需要警惕的信号，精读时先问边界。
6. 跨领域连接与意外收获
横跨 2 个类目（eess.AS、cs.SD），关注其在你兴趣板块间的迁移面。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。