Few-Shot Cross-Lingual TTS Using Transferable Phoneme Embedding

TL;DR

This paper introduces a transferable phoneme embedding framework for cross-lingual TTS that enables high-quality speech synthesis in unseen languages with very limited data, leveraging a phoneme-based model and self-supervised features.

Contribution

The paper proposes a novel framework combining a phoneme-based TTS model and a codebook module for effective cross-lingual transfer in few-shot settings.

Findings

Achieves intelligible speech synthesis with only 4 utterances in unseen languages.

Utilizes phoneme-level self-supervised features to improve speech quality.

Naive transfer learning fails under extremely few-shot conditions, motivating the new approach.

Abstract

This paper studies a transferable phoneme embedding framework that aims to deal with the cross-lingual text-to-speech (TTS) problem under the few-shot setting. Transfer learning is a common approach when it comes to few-shot learning since training from scratch on few-shot training data is bound to overfit. Still, we find that the naive transfer learning approach fails to adapt to unseen languages under extremely few-shot settings, where less than 8 minutes of data is provided. We deal with the problem by proposing a framework that consists of a phoneme-based TTS model and a codebook module to project phonemes from different languages into a learned latent space. Furthermore, by utilizing phoneme-level averaged self-supervised learned features, we effectively improve the quality of synthesized speeches. Experiments show that using 4 utterances, which is about 30 seconds of data, is enough to synthesize intelligible speech when adapting to an unseen language using our framework.