Abstract: We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks. Deep Voice lays the groundwork for truly end-to-end neural speech synthesis. The system comprises five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For the segmentation model, we propose a novel way of performing phoneme boundary detection with deep neural networks using connectionist temporal classification (CTC) loss. For the audio synthesis model, we implement a variant of WaveNet that requires fewer parameters and trains faster than the original. By using a neural network for each component, our system is simpler and more flexible than traditional text-to-speech systems, where each component requires laborious feature engineering and extensive domain expertise. Finally, we show that inference with our system can be performed faster than real time and describe optimized WaveNet inference kernels on both CPU and GPU that achieve up to 400x speedups over existing implementations.
| Comments: | Submitted to ICML 2017 |
| Subjects: | Computation and Language (cs.CL) ; Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE); Sound (cs.SD) |
| Cite as: | arXiv:1702.07825 [cs.CL] |
| (or arXiv:1702.07825v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.1702.07825 |
Focus to learn more
arXiv-issued DOI via DataCiteFrom: Andrew Gibiansky [view email]
[v1] Sat, 25 Feb 2017 03:11:04 UTC (123 KB)
[v2] Tue, 7 Mar 2017 23:09:23 UTC (123 KB)
View a PDF of the paper titled Deep Voice: Real-time Neural Text-to-Speech, by Sercan O. Arik and 11 other authors