Deep Denoising Auto-encoder for Statistical Speech Synthesis
Sound
2015-06-18 v1 Machine Learning
Abstract
This paper proposes a deep denoising auto-encoder technique to extract better acoustic features for speech synthesis. The technique allows us to automatically extract low-dimensional features from high dimensional spectral features in a non-linear, data-driven, unsupervised way. We compared the new stochastic feature extractor with conventional mel-cepstral analysis in analysis-by-synthesis and text-to-speech experiments. Our results confirm that the proposed method increases the quality of synthetic speech in both experiments.
Cite
@article{arxiv.1506.05268,
title = {Deep Denoising Auto-encoder for Statistical Speech Synthesis},
author = {Zhenzhou Wu and Shinji Takaki and Junichi Yamagishi},
journal= {arXiv preprint arXiv:1506.05268},
year = {2015}
}