Cross-lingual Text-To-Speech with Flow-based Voice Conversion for Improved Pronunciation

TL;DR

This paper introduces a flow-based voice conversion integrated with a non-attentive Tacotron model for improved cross-lingual TTS, enhancing pronunciation accuracy and speaker similarity, especially in low-resource settings.

Contribution

It proposes a novel end-to-end cross-lingual TTS framework combining flow-based voice conversion with disentangled representations, enabling better pronunciation and speaker consistency.

Findings

Improved pronunciation accuracy in cross-lingual synthesis.

Effective voice conversion with minimal data (average 7.5 minutes per speaker).

Outperforms baseline methods in objective and subjective evaluations.

Abstract

This paper presents a method for end-to-end cross-lingual text-to-speech (TTS) which aims to preserve the target language's pronunciation regardless of the original speaker's language. The model used is based on a non-attentive Tacotron architecture, where the decoder has been replaced with a normalizing flow network conditioned on the speaker identity, allowing both TTS and voice conversion (VC) to be performed by the same model due to the inherent linguistic content and speaker identity disentanglement. When used in a cross-lingual setting, acoustic features are initially produced with a native speaker of the target language and then voice conversion is applied by the same model in order to convert these features to the target speaker's voice. We verify through objective and subjective evaluations that our method can have benefits compared to baseline cross-lingual synthesis. By including speakers averaging 7.5 minutes of speech, we also present positive results on low-resource scenarios.