Prosodic ABX: A Language-Agnostic Method for Measuring Prosodic Contrast in Speech Representations

TL;DR

This paper introduces prosodic ABX, a novel, language-agnostic method to measure prosodic contrast in speech representations, using minimal pairs without explicit labels, and demonstrates its effectiveness across multiple languages.

Contribution

The authors develop and release a new prosodic ABX framework and datasets to evaluate prosodic contrast in self-supervised speech models across languages.

Findings

Model and layer rankings are consistent across conditions.

The method effectively measures prosodic contrast in English, Japanese, and Mandarin.

The approach requires only a few examples and no explicit labels.

Abstract

Speech representations from self-supervised speech models (S3Ms) are known to be sensitive to phonemic contrasts, but their sensitivity to prosodic contrasts has not been directly measured. The ABX discrimination task has been used to measure phonemic contrast in S3M representations via minimal pairs. We introduce prosodic ABX, an extension of this framework to evaluate prosodic contrast with only a handful of examples and no explicit labels. Also, we build and release a dataset of English and Japanese minimal pairs and use it along with a Mandarin dataset to evaluate contrast in English stress, Japanese pitch accent, and Mandarin tone. Finally, we show that model and layer rankings are often preserved across several experimental conditions, making it practical for low-resource settings.