arXiv 2409.157592024-12-20cs.SDeess.AS

VoiceGuider：通过自引导增强参数高效说话人自适应文本转语音的域外性能

VoiceGuider: Enhancing Out-of-Domain Performance in Parameter-Efficient Speaker-Adaptive Text-to-Speech via Autoguidance

Seoul National University（首尔大学）；Samsung Electronics Co., Ltd（三星电子株式会社）

Jiheum Yeom, Heeseung Kim, Jooyoung Choi, Che Hyun Lee, Nohil Park, Sungroh Yoon

备注：: IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2025, Demo Page: https://voiceguider.github.io/

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出VoiceGuider，通过自引导机制强化基于LoRA的参数高效说话人自适应文本转语音系统，有效缩小与全量微调模型的性能差距，显著提升极端域外说话人的自适应表现。

摘要：

当通过LoRA将参数高效微调应用于说话人自适应文本转语音模型时，与全量微调相比，自适应性能可能会下降，尤其是对于域外说话人。为此，我们提出VoiceGuider，这是一种通过自引导强化的参数高效说话人自适应文本转语音系统，旨在增强说话人自适应性能，缩小与全量微调模型的差距。我们深入探索了多种强化自引导的方法，最终找到了最优策略。VoiceGuider因此展现出稳健的自适应性能，特别是在极端域外语音数据上。我们在演示页面提供了音频样本。

英文摘要：

When applying parameter-efficient finetuning via LoRA onto speaker adaptive text-to-speech models, adaptation performance may decline compared to full-finetuned counterparts, especially for out-of-domain speakers. Here, we propose VoiceGuider, a parameter-efficient speaker adaptive text-to-speech system reinforced with autoguidance to enhance the speaker adaptation performance, reducing the gap against full-finetuned models. We carefully explore various ways of strengthening autoguidance, ultimately finding the optimal strategy. VoiceGuider as a result shows robust adaptation performance especially on extreme out-of-domain speech data. We provide audible samples in our demo page.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。