A multilingual training strategy for low resource Text to Speech

TL;DR

This paper explores using social media data and cross-lingual transfer learning to improve low-resource Text to Speech systems through multilingual pre-training, demonstrating enhanced speech quality.

Contribution

It introduces a novel approach combining social media data and multilingual transfer learning to improve low-resource TTS models, showing the effectiveness of multilingual pre-training.

Findings

Multilingual pre-training improves speech intelligibility and naturalness.

Social media data can be effectively used for low-resource TTS datasets.

Cross-lingual transfer learning benefits low-resource language TTS.

Abstract

Recent speech technologies have led to produce high quality synthesised speech due to recent advances in neural Text to Speech (TTS). However, such TTS models depend on extensive amounts of data that can be costly to produce and is hardly scalable to all existing languages, especially that seldom attention is given to low resource languages. With techniques such as knowledge transfer, the burden of creating datasets can be alleviated. In this paper, we therefore investigate two aspects; firstly, whether data from social media can be used for a small TTS dataset construction, and secondly whether cross lingual transfer learning (TL) for a low resource language can work with this type of data. In this aspect, we specifically assess to what extent multilingual modeling can be leveraged as an alternative to training on monolingual corporas. To do so, we explore how data from foreign languages may be selected and pooled to train a TTS model for a target low resource language. Our findings show that multilingual pre-training is better than monolingual pre-training at increasing the intelligibility and naturalness of the generated speech.