Fast, Compact, and High Quality LSTM-RNN Based Statistical Parametric Speech Synthesizers for Mobile Devices

Heiga Zen, Yannis Agiomyrgiannakis, Niels Egberts, Fergus Henderson, Przemysław Szczepaniak


Acoustic models based on long short-term memory recurrent neural networks (LSTM-RNNs) were applied to statistical parametric speech synthesis (SPSS) and showed significant improvements in naturalness and latency over those based on hidden Markov models (HMMs). This paper describes further optimizations of LSTM-RNN-based SPSS for deployment on mobile devices; weight quantization, multi-frame inference, and robust inference using an ϵ-contaminated Gaussian loss function. Experimental results in subjective listening tests show that these optimizations can make LSTM-RNN-based SPSS comparable to HMM-based SPSS in runtime speed while maintaining naturalness. Evaluations between LSTM-RNN-based SPSS and HMM-driven unit selection speech synthesis are also presented.


DOI: 10.21437/Interspeech.2016-522

Cite as

Zen, H., Agiomyrgiannakis, Y., Egberts, N., Henderson, F., Szczepaniak, P. (2016) Fast, Compact, and High Quality LSTM-RNN Based Statistical Parametric Speech Synthesizers for Mobile Devices. Proc. Interspeech 2016, 2273-2277.

Bibtex
@inproceedings{Zen+2016,
author={Heiga Zen and Yannis Agiomyrgiannakis and Niels Egberts and Fergus Henderson and Przemysław Szczepaniak},
title={Fast, Compact, and High Quality LSTM-RNN Based Statistical Parametric Speech Synthesizers for Mobile Devices},
year=2016,
booktitle={Interspeech 2016},
doi={10.21437/Interspeech.2016-522},
url={http://dx.doi.org/10.21437/Interspeech.2016-522},
pages={2273--2277}
}