arXiv cs.CL· Michael Whealing, Robert Gaizauskas, Thomas Hain·· 11 小时前AI 评分20
非对称对应学习 ACL：将 HuBERT 特征蒸馏为词汇与风格嵌入，并用声学特征解读潜在空间
Asymmetric Correspondence Learning for Lexical and Style Embeddings: Interpreting Latent Space via Acoustic Features
AI 导读
研究提出非对称对应学习（ACL），通过分离潜在空间的对应瓶颈将 HuBERT 特征蒸馏为声学词嵌入（AWE）和声学风格嵌入（ASE）。在单词语音上，AWE 空间呈现紧密的词汇聚类（AP≈0.90，Fisher≈6.28），ASE 空间线性探测说话人准确率达 97.9%，但仍保留大量线性不可恢复的词汇方差（Fisher≈0.86，0.17%）。
正文 · 原文
Abstract: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 (AP$\approx$0.90, Fisher$\approx$6.28), linear probing on the ASE space achieves 97.9\% speaker accuracy across a diffuse topology (Fisher$\approx$1.01), yet preserves substantial lexical variance (Fisher$\approx$0.86) 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 $F_0$ 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.
| Comments: | Accepted to Spoken Language Technology 2026. 6 pages |
| Subjects: | Computation and Language (cs.CL); Sound (cs.SD) |
| Cite as: | arXiv:2611.09313 [cs.CL] |
| (or arXiv:2611.09313v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.09313 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Michael Whealing Mr [view email]
[v1]
Fri, 6 Nov 2026 11:04:58 UTC (1,119 KB)
来源：arXiv cs.CL · arxiv.org