arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.08480eess.AScs.SD
收藏

s-Transformer:用于鲁棒神经语音合成的分段变换器

s-Transformer: Segment-Transformer for Robust Neural Speech Synthesis

Microsoft (China) Corporation(微软(中国)有限公司)

Xi Wang, Huaiping Ming, Lei He, Frank K. Soong

摘要:

神经元端到端的文本到语音转换(TTS)采用递归模型(如Tacotron)或注意力模型(如Transformer)对语音进行表征,显著提高了语音合成的性能。然而,它仍然是非常具有挑战性的,以处理不同的句子长度,特别是对于长句,序列模型的有效上下文长度的限制。我们提出了一种新的段变压器(S-变压器),它的语音模型在段级别的递归是通过缓存存储器的编码器和解码器的重用。扩展的存储器可以捕获长范围的上下文,同时,编码器和解码器的注意力集中在更容易处理的片段上。此外,我们采用了一个修改的相对位置自我注意泛化序列长度超出一个可能看不见的训练数据。通过将所提出的s-Transformer与标准的Transformer进行比较,在短句上,两者都实现了相同的MOS分数4.29,这与记录的4.32非常接近;在长句上的相似分数为4.22 vs 4.2,并且在MOS中的增益为0.2的超长句子上显著更好。由于缓存的内存会随着时间的推移而更新,因此s-Transformer可以长时间生成相当自然和连贯的语音。

英文摘要:

Neural end-to-end text-to-speech (TTS) , which adopts either a recurrent model, e.g. Tacotron, or an attention one, e.g. Transformer, to characterize a speech utterance, has achieved significant improvement of speech synthesis. However, it is still very challenging to deal with different sentence lengths, particularly, for long sentences where sequence model has limitation of the effective context length. We propose a novel segment-Transformer (s-Transformer), which models speech at segment level where recurrence is reused via cached memories for both the encoder and decoder. Long-range contexts can be captured by the extended memory, meanwhile, the encoder-decoder attention on segment which is much easier to handle. In addition, we employ a modified relative positional self attention to generalize sequence length beyond a period possibly unseen in the training data. By comparing the proposed s-Transformer with the standard Transformer, on short sentences, both achieve the same MOS scores of 4.29, which is very close to 4.32 by the recordings; similar scores of 4.22 vs 4.2 on long sentences, and significantly better for extra-long sentences with a gain of 0.2 in MOS. Since the cached memory is updated with time, the s-Transformer generates rather natural and coherent speech for a long period of time.

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。

↑