面向自回归语音生成的并行合成方法
Parallel Synthesis for Autoregressive Speech Generation
AI总结：
本文提出频域自回归生成（FAR）和逐位自回归生成（BAR）两种并行合成方法，将自回归迭代次数从与话语长度成正比改为与子带数/位数成正比，显著提升推理效率，无需GPU即可实现比实时更快的语音合成，并在MUSHRA和泛化能力上优于基线声码器。
摘要：
自回归神经声码器在文本到语音和语音转换等语音合成任务中取得了卓越性能。自回归声码器以先前时间步的样本为条件预测某个时间步的样本。尽管它能合成自然的人类语音，但迭代式生成不可避免地使合成时间与话语长度成正比，导致效率低下。许多工作致力于并行生成整个语音序列，并提出了基于GAN、基于流和基于分数的声码器。本文为自回归生成提出了一种新思路。所提模型不再按时间序列迭代预测样本，而是执行频域自回归生成（FAR）和逐位自回归生成（BAR）来合成语音。在FAR中，语音话语被分割为多个频率子带，每个子带以先前生成的子带为条件进行生成。类似地，在BAR中，8位量化信号从第一位开始迭代生成。通过重新设计自回归方法，使其在时间域以外的域中进行计算，所提模型中的迭代次数不再与话语长度成正比，而是与子带数/位数成正比，显著提高了推理效率。此外，采用后置滤波器从输出后验分布中采样信号；其训练目标根据所提方法的特点设计。实验结果表明，所提模型无需GPU加速即可实现比实时更快的语音合成。与基线声码器相比，所提模型取得了更好的MUSHRA结果，并展现出对未见说话人和44 kHz语音的良好泛化能力。
英文摘要：
Autoregressive neural vocoders have achieved outstanding performance in speech synthesis tasks such as text-to-speech and voice conversion. An autoregressive vocoder predicts a sample at some time step conditioned on those at previous time steps. Though it synthesizes natural human speech, the iterative generation inevitably makes the synthesis time proportional to the utterance length, leading to low efficiency. Many works were dedicated to generating the whole speech sequence in parallel and proposed GAN-based, flow-based, and score-based vocoders. This paper proposed a new thought for the autoregressive generation. Instead of iteratively predicting samples in a time sequence, the proposed model performs frequency-wise autoregressive generation (FAR) and bit-wise autoregressive generation (BAR) to synthesize speech. In FAR, a speech utterance is split into frequency subbands, and a subband is generated conditioned on the previously generated one. Similarly, in BAR, an 8-bit quantized signal is generated iteratively from the first bit. By redesigning the autoregressive method to compute in domains other than the time domain, the number of iterations in the proposed model is no longer proportional to the utterance length but to the number of subbands/bits, significantly increasing inference efficiency. Besides, a post-filter is employed to sample signals from output posteriors; its training objective is designed based on the characteristics of the proposed methods. Experimental results show that the proposed model can synthesize speech faster than real-time without GPU acceleration. Compared with baseline vocoders, the proposed model achieves better MUSHRA results and shows good generalization ability for unseen speakers and 44 kHz speech.
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。
