Quantifying Dimensional Independence in Speech: An Information-Theoretic Framework for Disentangled Representation Learning

TL;DR

This paper introduces an information-theoretic framework to quantify statistical dependence among speech features, revealing weak coupling across dimensions and source/filter dominance for different speech attributes.

Contribution

The authors develop a novel MI-based method to measure cross-dimension dependence in speech features, providing a principled approach for disentangled representation learning.

Findings

Cross-dimension MI is low (<0.15 nats), indicating weak dependence.

Source--Filter MI is higher (0.47 nats), showing stronger coupling.

Source dominates emotional dimensions, while filter dominates linguistic and pathological ones.

Abstract

Speech signals encode emotional, linguistic, and pathological information within a shared acoustic channel; however, disentanglement is typically assessed indirectly through downstream task performance. We introduce an information-theoretic framework to quantify cross-dimension statistical dependence in handcrafted acoustic features by integrating bounded neural mutual information (MI) estimation with non-parametric validation. Across six corpora, cross-dimension MI remains low, with tight estimation bounds ($< 0.15$ nats), indicating weak statistical coupling in the data considered, whereas Source--Filter MI is substantially higher (0.47 nats). Attribution analysis, defined as the proportion of total MI attributable to source versus filter components, reveals source dominance for emotional dimensions (80\%) and filter dominance for linguistic and pathological dimensions (60\% and 58\%, respectively). These findings provide a principled framework for quantifying dimensional independence in speech.