Voice Timbre Attribute Detection with Compact and Interpretable Training-Free Acoustic Parameters

TL;DR

This paper introduces a simple, interpretable acoustic parameter set for voice timbre attribute detection that rivals complex models, offering physical insight and low computational cost.

Contribution

The study demonstrates that a compact, training-free acoustic parameter set can effectively detect voice timbre attributes, outperforming traditional features and approaching advanced self-supervised models.

Findings

Outperforms conventional cepstral features in vTAD

Approaches state-of-the-art self-supervised models

Requires no trainable parameters and minimal computation

Abstract

Voice timbre attribute detection (vTAD) is the task of determining the relative intensity of timbre attributes between speech utterances. Voice timbre is a crucial yet inherently complex component of speech perception. While deep neural network (DNN) embeddings perform well in speaker modelling, they often act as black-box representations with limited physical interpretability and high computational cost. In this work, a compact acoustic parameter set is investigated for vTAD. The set captures important acoustic measures and their temporal dynamics which are found to be crucial in the task. Despite its simplicity, the acoustic parameter set is competitive, outperforming conventional cepstral features and supervised DNN embeddings, and approaching state-of-the-art self-supervised models. Importantly, the studied set require no trainable parameters, incur negligible computation, and offer explicit interpretability for analysing physical traits behind human timbre perception.