Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech
This work addresses the challenge of producing high-quality, diverse speech synthesis for TTS applications, though it is incremental as it builds on existing end-to-end and two-stage models.
The authors tackled the problem of generating natural-sounding audio in end-to-end text-to-speech (TTS) by proposing a parallel method that uses variational inference with normalizing flows, adversarial training, and a stochastic duration predictor. Their method achieved a mean opinion score (MOS) comparable to ground truth on the LJ Speech dataset, outperforming existing TTS systems.
Several recent end-to-end text-to-speech (TTS) models enabling single-stage training and parallel sampling have been proposed, but their sample quality does not match that of two-stage TTS systems. In this work, we present a parallel end-to-end TTS method that generates more natural sounding audio than current two-stage models. Our method adopts variational inference augmented with normalizing flows and an adversarial training process, which improves the expressive power of generative modeling. We also propose a stochastic duration predictor to synthesize speech with diverse rhythms from input text. With the uncertainty modeling over latent variables and the stochastic duration predictor, our method expresses the natural one-to-many relationship in which a text input can be spoken in multiple ways with different pitches and rhythms. A subjective human evaluation (mean opinion score, or MOS) on the LJ Speech, a single speaker dataset, shows that our method outperforms the best publicly available TTS systems and achieves a MOS comparable to ground truth.