Tibetan-TTS:Low-Resource Tibetan Speech Synthesis with Large Model Adaptation

TL;DR

This paper introduces a large-model Tibetan TTS system that effectively addresses low-resource challenges, dialectal variation, and complex text-pronunciation mapping, achieving high-quality speech synthesis.

Contribution

It presents the first large-model Tibetan TTS system utilizing data enhancement, text representation adaptation, and cross-lingual training for low-resource settings.

Findings

Achieved MOS scores of 4.28 and 4.35 for syllable-level and BPE-based systems.

Pronunciation accuracy of 97.6% and 96.6%, outperforming commercial TTS.

Demonstrated the effectiveness of large-model adaptation for Tibetan speech synthesis.

Abstract

Tibetan text-to-speech (TTS) has long been challenged by scarce speech resources, significant dialectal variation, and the complex mapping between written text and spoken pronunciation. To address these issues, this work presents, to the best of our knowledge, the first large-model-based Tibetan TTS system in the industry, built upon a large speech synthesis model developed by Xingchen AGI Lab. The proposed system integrates data quality enhancement, Tibetan-oriented text representation and tokenizer adaptation, and cross-lingual adaptive training for low-resource Tibetan speech synthesis. Experimental results show that the system can generate stable, natural, and intelligible Tibetan speech under low-resource conditions. In subjective evaluation, the MOS scores of the syllable-level and BPE-based systems reach 4.28 and 4.35, while their pronunciation accuracies reach 97.6% and 96.6%, respectively, outperforming an external commercial Tibetan TTS interface. These results demonstrate that combining a large-model backbone with Tibetan-oriented text representation adaptation and cross-lingual adaptive training enables highly usable low-resource Tibetan speech synthesis, and also provides a technical foundation for future unified multi-dialect Tibetan speech synthesis.