Recent approaches in text-to-speech (TTS) synthesis employ neural network strategies to vocode perceptually-informed spectrogram representations directly into listenable waveforms. Such vocoding procedures create a computational bottleneck in modern TTS pipelines. We propose an alternative approach which utilizes generative adversarial networks (GANs) to learn mappings from perceptually-informed spectrograms to simple magnitude spectrograms which can be heuristically vocoded. Through a user study, we show that our approach significantly outperforms na\"ive vocoding strategies while being hundreds of times faster than neural network vocoders used in state-of-the-art TTS systems. We also show that our method can be used to achieve state-of-the-art results in unsupervised synthesis of individual words of speech.
@article{arxiv.1904.07944,
title = {Expediting TTS Synthesis with Adversarial Vocoding},
author = {Paarth Neekhara and Chris Donahue and Miller Puckette and Shlomo Dubnov and Julian McAuley},
journal= {arXiv preprint arXiv:1904.07944},
year = {2019}
}
Comments
Published as a conference paper at INTERSPEECH 2019