Zero-Shot Sing Voice Conversion: built upon clustering-based phoneme representations

TL;DR

This paper introduces a zero-shot singing voice conversion method using clustering-based phoneme representations, enabling effective manipulation of voice characteristics and improving sound quality and timbre accuracy across diverse datasets.

Contribution

It presents a novel clustering-based phoneme representation for zero-shot singing voice conversion, enhancing content-timbre separation and voice manipulation capabilities.

Findings

Datasets with fewer recordings per artist show more timbre leakage.

Model significantly improves sound quality and timbre accuracy.

Advances zero-shot SVC and emphasizes future work on discrete speech representation.

Abstract

This study presents an innovative Zero-Shot any-to-any Singing Voice Conversion (SVC) method, leveraging a novel clustering-based phoneme representation to effectively separate content, timbre, and singing style. This approach enables precise voice characteristic manipulation. We discovered that datasets with fewer recordings per artist are more susceptible to timbre leakage. Extensive testing on over 10,000 hours of singing and user feedback revealed our model significantly improves sound quality and timbre accuracy, aligning with our objectives and advancing voice conversion technology. Furthermore, this research advances zero-shot SVC and sets the stage for future work on discrete speech representation, emphasizing the preservation of rhyme.