Everyone-Can-Sing: Zero-Shot Singing Voice Synthesis and Conversion with Speech Reference

TL;DR

This paper introduces a unified zero-shot framework for singing voice synthesis and conversion that leverages pre-trained embeddings and diffusion models, enabling high-quality, controllable singing voice generation and cloning from speech references.

Contribution

A novel unified zero-shot framework for SVS and SVC using pre-trained embeddings and diffusion models, addressing cross-domain limitations and data scarcity.

Findings

Significant improvements in timbre similarity and musicality.

Effective singing voice cloning from speech references.

Insights into low-data music tasks like instrumental style transfer.

Abstract

We propose a unified framework for Singing Voice Synthesis (SVS) and Conversion (SVC), addressing the limitations of existing approaches in cross-domain SVS/SVC, poor output musicality, and scarcity of singing data. Our framework enables control over multiple aspects, including language content based on lyrics, performance attributes based on a musical score, singing style and vocal techniques based on a selector, and voice identity based on a speech sample. The proposed zero-shot learning paradigm consists of one SVS model and two SVC models, utilizing pre-trained content embeddings and a diffusion-based generator. The proposed framework is also trained on mixed datasets comprising both singing and speech audio, allowing singing voice cloning based on speech reference. Experiments show substantial improvements in timbre similarity and musicality over state-of-the-art baselines, providing insights into other low-data music tasks such as instrumental style transfer. Examples can be found at: everyone-can-sing.github.io.