Improving Audio Codec-based Zero-Shot Text-to-Speech Synthesis with Multi-Modal Context and Large Language Model

TL;DR

This paper introduces a novel multi-modal context-enhanced TTS model that leverages large language models and audio codecs to improve zero-shot speech synthesis, especially for longer contexts like audiobooks and conversations.

Contribution

It proposes a multi-modal context-enhanced Qformer and integrates a pretrained LLM with SoundStorm to improve context utilization, audio quality, and speaker similarity in zero-shot TTS.

Findings

Outperforms baseline models in various context TTS scenarios.

Enhances audio quality and speaker similarity.

Effectively leverages longer context information.

Abstract

Recent advances in large language models (LLMs) and development of audio codecs greatly propel the zero-shot TTS. They can synthesize personalized speech with only a 3-second speech of an unseen speaker as acoustic prompt. However, they only support short speech prompts and cannot leverage longer context information, as required in audiobook and conversational TTS scenarios. In this paper, we introduce a novel audio codec-based TTS model to adapt context features with multiple enhancements. Inspired by the success of Qformer, we propose a multi-modal context-enhanced Qformer (MMCE-Qformer) to utilize additional multi-modal context information. Besides, we adapt a pretrained LLM to leverage its understanding ability to predict semantic tokens, and use a SoundStorm to generate acoustic tokens thereby enhancing audio quality and speaker similarity. The extensive objective and subjective evaluations show that our proposed method outperforms baselines across various context TTS scenarios.