Towards General-Purpose Text-Instruction-Guided Voice Conversion

TL;DR

This paper presents a versatile voice conversion model guided by text instructions, enabling more controllable and specific speech modifications without relying on reference utterances, advancing the flexibility of voice conversion technology.

Contribution

The proposed neural codec language model uniquely incorporates text instructions as style prompts, allowing end-to-end control of speech prosody and emotion in voice conversion.

Findings

Model effectively interprets text instructions to modify speech

Achieves high-quality voice conversion without reference utterances

Demonstrates strong understanding of style prompts

Abstract

This paper introduces a novel voice conversion (VC) model, guided by text instructions such as "articulate slowly with a deep tone" or "speak in a cheerful boyish voice". Unlike traditional methods that rely on reference utterances to determine the attributes of the converted speech, our model adds versatility and specificity to voice conversion. The proposed VC model is a neural codec language model which processes a sequence of discrete codes, resulting in the code sequence of converted speech. It utilizes text instructions as style prompts to modify the prosody and emotional information of the given speech. In contrast to previous approaches, which often rely on employing separate encoders like prosody and content encoders to handle different aspects of the source speech, our model handles various information of speech in an end-to-end manner. Experiments have demonstrated the impressive capabilities of our model in comprehending instructions and delivering reasonable results.