Neural Voice Cloning with a Few Samples

TL;DR

This paper presents a neural voice cloning system that can generate personalized speech from only a few audio samples, using either speaker adaptation or speaker encoding methods, balancing quality and resource efficiency.

Contribution

It introduces and compares two novel approaches for neural voice cloning that require minimal samples, enabling high-quality personalized speech synthesis with different resource trade-offs.

Findings

Both methods achieve good naturalness and similarity with few samples.

Speaker adaptation yields better quality but is more resource-intensive.

Speaker encoding is faster and more resource-efficient, suitable for low-resource deployment.

Abstract

Voice cloning is a highly desired feature for personalized speech interfaces. Neural network based speech synthesis has been shown to generate high quality speech for a large number of speakers. In this paper, we introduce a neural voice cloning system that takes a few audio samples as input. We study two approaches: speaker adaptation and speaker encoding. Speaker adaptation is based on fine-tuning a multi-speaker generative model with a few cloning samples. Speaker encoding is based on training a separate model to directly infer a new speaker embedding from cloning audios and to be used with a multi-speaker generative model. In terms of naturalness of the speech and its similarity to original speaker, both approaches can achieve good performance, even with very few cloning audios. While speaker adaptation can achieve better naturalness and similarity, the cloning time or required memory for the speaker encoding approach is significantly less, making it favorable for low-resource deployment.