Quantifying Dimensional Independence in Speech: An Information-Theoretic Framework for Disentangled Representation Learning
For speech researchers, this provides a principled method to quantify disentanglement in acoustic features, though it is an incremental application of existing MI estimation techniques to a new domain.
The paper introduces an information-theoretic framework to quantify cross-dimension statistical dependence in speech features, finding low cross-dimension mutual information (MI) across six corpora (bounds <0.15 nats) and higher Source-Filter MI (0.47 nats), with attribution analysis revealing source dominance for emotion (80%) and filter dominance for linguistic (60%) and pathological (58%) dimensions.
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.