arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.06320eess.AScs.AIcs.CLcs.SD
收藏

TTS-Transducer:基于神经转换器的端到端语音合成

TTS-Transducer: End-to-End Speech Synthesis with Neural Transducer

NVIDIA(英伟达)

Vladimir Bataev, Subhankar Ghosh, Vitaly Lavrukhin, Jason Li

AI总结:

本研究针对音频编解码器多码本多标记预测的挑战,提出TTS-Transducer端到端语音合成架构,融合神经转换器(免显式时长预测)与非自回归Transformer,实现高效语音生成,性能优于当代系统。

摘要:

本研究提出TTS-Transducer——一种新颖的文本转语音架构,它融合了音频编解码器模型与神经转换器的优势。在语音识别中以高质量和鲁棒性著称的转换器被用于学习单调对齐,从而避免使用显式时长预测器。神经音频编解码器能高效地将音频压缩为离散编码,为将文本建模方法应用于语音生成提供了可能。然而,具有残差量化器的音频编解码器模型要求每帧从多个码本预测多个标记,这一复杂性构成了重大挑战。该系统首先利用转换器架构学习文本标记与第一码本语音编解码标记间的单调对齐,再通过非自回归Transformer利用转换器损失提取的对齐信息预测剩余编码,且整个系统采用端到端训练。实验表明,TTS-Transducer是当代TTS系统的竞争性鲁棒替代方案。

英文摘要:

This work introduces TTS-Transducer - a novel architecture for text-to-speech, leveraging the strengths of audio codec models and neural transducers. Transducers, renowned for their superior quality and robustness in speech recognition, are employed to learn monotonic alignments and allow for avoiding using explicit duration predictors. Neural audio codecs efficiently compress audio into discrete codes, revealing the possibility of applying text modeling approaches to speech generation. However, the complexity of predicting multiple tokens per frame from several codebooks, as necessitated by audio codec models with residual quantizers, poses a significant challenge. The proposed system first uses a transducer architecture to learn monotonic alignments between tokenized text and speech codec tokens for the first codebook. Next, a non-autoregressive Transformer predicts the remaining codes using the alignment extracted from transducer loss. The proposed system is trained end-to-end. We show that TTS-Transducer is a competitive and robust alternative to contemporary TTS systems.

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。

↑