Advancing Zero-shot Text-to-Speech Intelligibility across Diverse Domains via Preference Alignment

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Xueyao, Wang, Yuancheng, Wang, Chaoren, Li, Ziniu, Chen, Zhuo, Wu, Zhizheng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913880445288448
author Zhang, Xueyao
Wang, Yuancheng
Wang, Chaoren
Li, Ziniu
Chen, Zhuo
Wu, Zhizheng
author_facet Zhang, Xueyao
Wang, Yuancheng
Wang, Chaoren
Li, Ziniu
Chen, Zhuo
Wu, Zhizheng
contents Modern zero-shot text-to-speech (TTS) systems, despite using extensive pre-training, often struggle in challenging scenarios such as tongue twisters, repeated words, code-switching, and cross-lingual synthesis, leading to intelligibility issues. To address these limitations, this paper leverages preference alignment techniques, which enable targeted construction of out-of-pretraining-distribution data to enhance performance. We introduce a new dataset, named the Intelligibility Preference Speech Dataset (INTP), and extend the Direct Preference Optimization (DPO) framework to accommodate diverse TTS architectures. After INTP alignment, in addition to intelligibility, we observe overall improvements including naturalness, similarity, and audio quality for multiple TTS models across diverse domains. Based on that, we also verify the weak-to-strong generalization ability of INTP for more intelligible models such as CosyVoice 2 and Ints. Moreover, we showcase the potential for further improvements through iterative alignment based on Ints. Audio samples are available at https://intalign.github.io/.
format Preprint
id arxiv_https___arxiv_org_abs_2505_04113
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Advancing Zero-shot Text-to-Speech Intelligibility across Diverse Domains via Preference Alignment
Zhang, Xueyao
Wang, Yuancheng
Wang, Chaoren
Li, Ziniu
Chen, Zhuo
Wu, Zhizheng
Sound
Audio and Speech Processing
Modern zero-shot text-to-speech (TTS) systems, despite using extensive pre-training, often struggle in challenging scenarios such as tongue twisters, repeated words, code-switching, and cross-lingual synthesis, leading to intelligibility issues. To address these limitations, this paper leverages preference alignment techniques, which enable targeted construction of out-of-pretraining-distribution data to enhance performance. We introduce a new dataset, named the Intelligibility Preference Speech Dataset (INTP), and extend the Direct Preference Optimization (DPO) framework to accommodate diverse TTS architectures. After INTP alignment, in addition to intelligibility, we observe overall improvements including naturalness, similarity, and audio quality for multiple TTS models across diverse domains. Based on that, we also verify the weak-to-strong generalization ability of INTP for more intelligible models such as CosyVoice 2 and Ints. Moreover, we showcase the potential for further improvements through iterative alignment based on Ints. Audio samples are available at https://intalign.github.io/.
title Advancing Zero-shot Text-to-Speech Intelligibility across Diverse Domains via Preference Alignment
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2505.04113