← 研踪 PaperTrailarXiv:2611.09313 · v1
Asymmetric Correspondence Learning for Lexical and Style Embeddings: Interpreting Latent Space via Acoustic Features
首次提交 2026-11-06 19:04
版本更新 2026-11-06 19:04
雷达观测 2026-11-09 11:19
摘要
Self-supervised speech embeddings jointly encode lexical, speaker, and prosodic information, complicating feature isolation. We introduce Asymmetric Correspondence Learning (ACL), distilling HuBERT features into separate Acoustic Word Embeddings (AWEs) and Acoustic Style Embeddings (ASEs) via a split-latent correspondence bottleneck. Through probing and clustering evaluations on single-word utterances, we demonstrate that substantial variance from extrinsic sources remains encoded within each latent space, despite linear probing indicating the information has been stripped. Whilst the AWE space exhibits cohesive lexical clustering (AP0.90, Fisher6.28), linear probing on the ASE space achieves 97.9\% speaker accuracy across a diffuse topology (Fisher1.01), yet preserves substantial lexical variance (Fisher0.86) that remains linearly unrecoverable (0.17\%). To explain these findings, we apply group-level permutation feature importance over eGeMAPS acoustic features, finding that AWEs partition by vocal tract filters and ASEs by dynamics. These findings suggest that standard diagnostic benchmarks may be insufficient for verifying feature isolation, as substantial speech structure can be retained even when it remains entirely hidden from probes.
作者备注:Accepted to Spoken Language Technology 2026. 6 pages
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-06 19:04 · 最新观测版本
Asymmetric Correspondence Learning for Lexical and Style Embeddings: Interpreting Latent Space via Acoustic Features
Self-supervised speech embeddings jointly encode lexical, speaker, and prosodic information, complicating feature isolation. We introduce Asymmetric Correspondence Learning (ACL), distilling HuBERT features into separate Acoustic Word Embeddings (AWEs) and Acoustic Style Embeddings (ASEs) via a split-latent correspondence bottleneck. Through probing and clustering evaluations on single-word utterances, we demonstrate that substantial variance from extrinsic sources remains encoded within each latent space, despite linear probing indicating the information has been stripped. Whilst the AWE space exhibits cohesive lexical clustering (AP0.90, Fisher6.28), linear probing on the ASE space achieves 97.9\% speaker accuracy across a diffuse topology (Fisher1.01), yet preserves substantial lexical variance (Fisher0.86) that remains linearly unrecoverable (0.17\%). To explain these findings, we apply group-level permutation feature importance over eGeMAPS acoustic features, finding that AWEs partition by vocal tract filters and ASEs by dynamics. These findings suggest that standard diagnostic benchmarks may be insufficient for verifying feature isolation, as substantial speech structure can be retained even when it remains entirely hidden from probes.
查看此版本解读 →查看此版本原文 ↗