SNAC: Speaker-normalized affine coupling layer in flow-based architecture for zero-shot multi-speaker text-to-speech

TL;DR

This paper introduces SNAC, a speaker-normalized affine coupling layer that improves zero-shot multi-speaker TTS by enhancing speaker similarity and speech quality for unseen speakers.

Contribution

The paper proposes a novel speaker-normalized affine coupling layer that explicitly normalizes inputs based on speaker embeddings, enabling better zero-shot speaker adaptation in flow-based TTS models.

Findings

Achieved state-of-the-art speaker similarity in ZSM-TTS

Improved speech quality over previous methods

Effective normalization-based conditioning for unseen speakers

Abstract

Zero-shot multi-speaker text-to-speech (ZSM-TTS) models aim to generate a speech sample with the voice characteristic of an unseen speaker. The main challenge of ZSM-TTS is to increase the overall speaker similarity for unseen speakers. One of the most successful speaker conditioning methods for flow-based multi-speaker text-to-speech (TTS) models is to utilize the functions which predict the scale and bias parameters of the affine coupling layers according to the given speaker embedding vector. In this letter, we improve on the previous speaker conditioning method by introducing a speaker-normalized affine coupling (SNAC) layer which allows for unseen speaker speech synthesis in a zero-shot manner leveraging a normalization-based conditioning technique. The newly designed coupling layer explicitly normalizes the input by the parameters predicted from a speaker embedding vector while training, enabling an inverse process of denormalizing for a new speaker embedding at inference. The proposed conditioning scheme yields the state-of-the-art performance in terms of the speech quality and speaker similarity in a ZSM-TTS setting.