Exploring VQ-VAE with Prosody Parameters for Speaker Anonymization

TL;DR

This paper presents a novel VQ-VAE-based speaker anonymization method that disentangles speech components to modify speaker identity while preserving linguistic and emotional content, improving emotional preservation over baselines.

Contribution

Introduces an end-to-end VQ-VAE model with separate embeddings for content, prosody, and speaker identity, enabling nuanced speaker anonymization with emotional preservation.

Findings

Outperforms baselines in emotional content preservation

Limited performance on some voice privacy tasks

Highlights need for further improvements

Abstract

Human speech conveys prosody, linguistic content, and speaker identity. This article investigates a novel speaker anonymization approach using an end-to-end network based on a Vector-Quantized Variational Auto-Encoder (VQ-VAE) to deal with these speech components. This approach is designed to disentangle these components to specifically target and modify the speaker identity while preserving the linguistic and emotionalcontent. To do so, three separate branches compute embeddings for content, prosody, and speaker identity respectively. During synthesis, taking these embeddings, the decoder of the proposed architecture is conditioned on both speaker and prosody information, allowing for capturing more nuanced emotional states and precise adjustments to speaker identification. Findings indicate that this method outperforms most baseline techniques in preserving emotional information. However, it exhibits more limited performance on other voice privacy tasks, emphasizing the need for further improvements.