arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2306.02405cs.CL
收藏

语音自监督离散表示的信息论分析

An Information-Theoretic Analysis of Self-supervised Discrete Representations of Speech

Saarland University(萨尔大学);Saarland Informatics Campus(萨尔信息学园区)

Badr M. Abdullah, Mohammed Maqsood Shaik, Bernd Möbius, Dietrich Klakow

AI总结:

针对自监督语音离散单元与音位类别关系不明的问题,本文提出信息论分布框架,分析 wav2vec 2.0 与 XLSR,揭示二者虽非一一对应但存在间接联系。

摘要:

语音的自监督表示学习通常包含一个量化步骤,将声学输入转换为离散单元。然而,目前仍不清楚应如何刻画这些离散单元与音位等抽象语音类别之间的关系。本文提出了一个信息论框架,将每个语音类别表示为离散单元上的一种分布。随后,我们将该框架应用于两种不同的自监督模型(即 wav2vec 2.0 和 XLSR),并以美式英语语音作为案例研究。研究表明,语音分布的熵反映了底层语音的变异性,语音上相似的声音呈现出相似的分布。尽管研究证实二者之间不存在直接的一一对应关系,但我们发现语音类别与离散单元之间存在一种有趣的间接关系。

英文摘要:

Self-supervised representation learning for speech often involves a quantization step that transforms the acoustic input into discrete units. However, it remains unclear how to characterize the relationship between these discrete units and abstract phonetic categories such as phonemes. In this paper, we develop an information-theoretic framework whereby we represent each phonetic category as a distribution over discrete units. We then apply our framework to two different self-supervised models (namely wav2vec 2.0 and XLSR) and use American English speech as a case study. Our study demonstrates that the entropy of phonetic distributions reflects the variability of the underlying speech sounds, with phonetically similar sounds exhibiting similar distributions. While our study confirms the lack of direct, one-to-one correspondence, we find an intriguing, indirect relationship between phonetic categories and discrete units.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑