FMSD-TTS: Few-shot Multi-Speaker Multi-Dialect Text-to-Speech Synthesis for \"U-Tsang, Amdo and Kham Speech Dataset Generation

TL;DR

FMSD-TTS is a novel few-shot multi-dialect TTS framework for Tibetan that synthesizes high-quality speech across dialects using limited data, with new modules and publicly released resources.

Contribution

The paper introduces a new few-shot TTS system for Tibetan dialects, including novel modules, a large synthetic speech corpus, and an open-source evaluation toolkit.

Findings

FMSD-TTS outperforms baselines in dialectal expressiveness and speaker similarity.

Synthesized speech is validated through dialect conversion tasks.

The system effectively captures fine-grained dialectal and speaker variations.

Abstract

Tibetan is a low-resource language with minimal parallel speech corpora spanning its three major dialects-\"U-Tsang, Amdo, and Kham-limiting progress in speech modeling. To address this issue, we propose FMSD-TTS, a few-shot, multi-speaker, multi-dialect text-to-speech framework that synthesizes parallel dialectal speech from limited reference audio and explicit dialect labels. Our method features a novel speaker-dialect fusion module and a Dialect-Specialized Dynamic Routing Network (DSDR-Net) to capture fine-grained acoustic and linguistic variations across dialects while preserving speaker identity. Extensive objective and subjective evaluations demonstrate that FMSD-TTS significantly outperforms baselines in both dialectal expressiveness and speaker similarity. We further validate the quality and utility of the synthesized speech through a challenging speech-to-speech dialect conversion task. Our contributions include: (1) a novel few-shot TTS system tailored for Tibetan multi-dialect speech synthesis, (2) the public release of a large-scale synthetic Tibetan speech corpus generated by FMSD-TTS, and (3) an open-source evaluation toolkit for standardized assessment of speaker similarity, dialect consistency, and audio quality.