Improving Robustness of LLM-based Speech Synthesis by Learning Monotonic Alignment

TL;DR

This paper enhances the robustness of LLM-based speech synthesis by enforcing monotonic alignment through CTC loss and attention priors, reducing errors like repetitions and misalignments.

Contribution

It introduces a novel guided attention training method that improves alignment robustness without adding new parameters in LLM-based TTS models.

Findings

Significant reduction in alignment errors and hallucinations.

Improved speech naturalness and consistency.

Enhanced robustness in multi-occurrence token scenarios.

Abstract

Large Language Model (LLM) based text-to-speech (TTS) systems have demonstrated remarkable capabilities in handling large speech datasets and generating natural speech for new speakers. However, LLM-based TTS models are not robust as the generated output can contain repeating words, missing words and mis-aligned speech (referred to as hallucinations or attention errors), especially when the text contains multiple occurrences of the same token. We examine these challenges in an encoder-decoder transformer model and find that certain cross-attention heads in such models implicitly learn the text and speech alignment when trained for predicting speech tokens for a given text. To make the alignment more robust, we propose techniques utilizing CTC loss and attention priors that encourage monotonic cross-attention over the text tokens. Our guided attention training technique does not introduce any new learnable parameters and significantly improves robustness of LLM-based TTS models.