Exact Prosody Cloning in Zero-Shot Multispeaker Text-to-Speech

TL;DR

This paper presents a novel zero-shot multispeaker TTS method that independently clones speaker identity and prosody using utterance normalization, a lightweight aligner, and quick fine-tuning, achieving high-quality, high-similarity results.

Contribution

It introduces a combined approach for prosody and speaker cloning in zero-shot TTS with a lightweight aligner and rapid fine-tuning, enhancing flexibility and quality.

Findings

High similarity to original voice and prosody demonstrated

Independent cloning of voice and prosody without quality loss

Fast fine-tuning within seconds for individual samples

Abstract

The cloning of a speaker's voice using an untranscribed reference sample is one of the great advances of modern neural text-to-speech (TTS) methods. Approaches for mimicking the prosody of a transcribed reference audio have also been proposed recently. In this work, we bring these two tasks together for the first time through utterance level normalization in conjunction with an utterance level speaker embedding. We further introduce a lightweight aligner for extracting fine-grained prosodic features, that can be finetuned on individual samples within seconds. We show that it is possible to clone the voice of a speaker as well as the prosody of a spoken reference independently without any degradation in quality and high similarity to both original voice and prosody, as our objective evaluation and human study show. All of our code and trained models are available, alongside static and interactive demos.