MegaTTS 3: Sparse Alignment Enhanced Latent Diffusion Transformer for Zero-Shot Speech Synthesis

TL;DR

MegaTTS 3 introduces a sparse alignment algorithm and advanced guidance strategies to significantly improve zero-shot speech synthesis quality, naturalness, and control, while reducing computational steps needed for high-quality output.

Contribution

The paper presents a novel sparse alignment method and guidance strategies that enhance zero-shot TTS performance and naturalness without relying on forced alignments.

Findings

Achieves state-of-the-art zero-shot TTS quality.

Supports flexible accent intensity control.

Generates high-quality speech with only 8 sampling steps.

Abstract

While recent zero-shot text-to-speech (TTS) models have significantly improved speech quality and expressiveness, mainstream systems still suffer from issues related to speech-text alignment modeling: 1) models without explicit speech-text alignment modeling exhibit less robustness, especially for hard sentences in practical applications; 2) predefined alignment-based models suffer from naturalness constraints of forced alignments. This paper introduces \textit{MegaTTS 3}, a TTS system featuring an innovative sparse alignment algorithm that guides the latent diffusion transformer (DiT). Specifically, we provide sparse alignment boundaries to MegaTTS 3 to reduce the difficulty of alignment without limiting the search space, thereby achieving high naturalness. Moreover, we employ a multi-condition classifier-free guidance strategy for accent intensity adjustment and adopt the piecewise rectified flow technique to accelerate the generation process. Experiments demonstrate that MegaTTS 3 achieves state-of-the-art zero-shot TTS speech quality and supports highly flexible control over accent intensity. Notably, our system can generate high-quality one-minute speech with only 8 sampling steps. Audio samples are available at https://sditdemo.github.io/sditdemo/.