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 2109.08710eess.AScs.CLcs.PFcs.SD
收藏

设备上神经语音合成

On-device neural speech synthesis

Apple(苹果公司)

Sivanand Achanta, Albert Antony, Ladan Golipour, Jiangchuan Li, Tuomo Raitio, Ramya Rasipuram, Francesco Rossi, Jennifer Shi, Jaimin Upadhyay, David Winarsky, Hepeng Zhang

摘要:

文本到语音(TTS)合成的最新进展,如Tacotron和WaveRNN,使得有可能通过将两个组件耦合在一起来构建一个完全基于神经网络的TTS系统。这样的系统在概念上是简单的,因为它只需要字素或音素输入,使用梅尔频谱图作为中间特征,并直接生成语音样本。该系统实现了等于或接近自然语音的质量。然而,系统的高计算成本和鲁棒性问题限制了它们在现实世界的语音合成应用和产品中的使用。在本文中,我们提出了关键的建模改进和优化策略,使部署这些模型,不仅在GPU服务器上,而且在移动设备上。该系统可以生成高质量的24 kHz语音,速度比服务器上的实时速度快5倍,比移动设备上的实时速度快3倍。

英文摘要:

Recent advances in text-to-speech (TTS) synthesis, such as Tacotron and WaveRNN, have made it possible to construct a fully neural network based TTS system, by coupling the two components together. Such a system is conceptually simple as it only takes grapheme or phoneme input, uses Mel-spectrogram as an intermediate feature, and directly generates speech samples. The system achieves quality equal or close to natural speech. However, the high computational cost of the system and issues with robustness have limited their usage in real-world speech synthesis applications and products. In this paper, we present key modeling improvements and optimization strategies that enable deploying these models, not only on GPU servers, but also on mobile devices. The proposed system can generate high-quality 24 kHz speech at 5x faster than real time on server and 3x faster than real time on mobile devices.

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。

↑