Self-Supervised Singing Voice Pre-Training towards Speech-to-Singing Conversion

TL;DR

This paper introduces SVPT, a self-supervised pre-training approach for speech-to-singing conversion that addresses data scarcity and alignment challenges, enabling zero-shot conversion and improving singing voice synthesis.

Contribution

It proposes a novel self-supervised pre-training method that leverages unpaired data and in-context learning for improved speech-to-singing conversion and synthesis.

Findings

Significant improvements in STS and SVS performance

Effective zero-shot conversion capability

Mitigation of data scarcity issues

Abstract

Speech-to-singing voice conversion (STS) task always suffers from data scarcity, because it requires paired speech and singing data. Compounding this issue are the challenges of content-pitch alignment and the suboptimal quality of generated outputs, presenting significant hurdles in STS research. This paper presents SVPT, an STS approach boosted by a self-supervised singing voice pre-training model. We leverage spoken language model techniques to tackle the rhythm alignment problem and the in-context learning capability to achieve zero-shot conversion. We adopt discrete-unit random resampling and pitch corruption strategies, enabling training with unpaired singing data and thus mitigating the issue of data scarcity. SVPT also serves as an effective backbone for singing voice synthesis (SVS), offering insights into scaling up SVS models. Experimental results indicate that SVPT delivers notable improvements in both STS and SVS endeavors. Audio samples are available at https://speech2sing.github.io.