A unified one-shot prosody and speaker conversion system with self-supervised discrete speech units

TL;DR

This paper introduces a unified one-shot voice conversion system that effectively models prosody and speaker attributes using self-supervised discrete speech units, improving naturalness and content preservation.

Contribution

It proposes a cascaded modular system leveraging self-supervised discrete speech units for better prosody and content modeling in voice conversion.

Findings

Outperforms previous methods in naturalness and intelligibility

Enhances speaker transferability and prosody transferability

Utilizes self-supervised discrete units for improved language content preservation

Abstract

We present a unified system to realize one-shot voice conversion (VC) on the pitch, rhythm, and speaker attributes. Existing works generally ignore the correlation between prosody and language content, leading to the degradation of naturalness in converted speech. Additionally, the lack of proper language features prevents these systems from accurately preserving language content after conversion. To address these issues, we devise a cascaded modular system leveraging self-supervised discrete speech units as language representation. These discrete units provide duration information essential for rhythm modeling. Our system first extracts utterance-level prosody and speaker representations from the raw waveform. Given the prosody representation, a prosody predictor estimates pitch, energy, and duration for each discrete unit in the utterance. A synthesizer further reconstructs speech based on the predicted prosody, speaker representation, and discrete units. Experiments show that our system outperforms previous approaches in naturalness, intelligibility, speaker transferability, and prosody transferability. Code and samples are publicly available.