A Vector Quantized Variational Autoencoder (VQ-VAE) Autoregressive Neural $F_0$ Model for Statistical Parametric Speech Synthesis

Recurrent neural networks (RNNs) can predict fundamental frequency (F
₀
) for statistical parametric speech synthesis systems, given linguistic features as input. However, these models assume conditional independence between consecutive F
₀
values, given the RNN state. In a previous study, we proposed autoregressive (AR) neural F
₀
models to capture the causal dependency of successive F
₀
values. In subjective evaluations, a deep AR model (DAR) outperformed an RNN. Here, we propose a Vector Quantized Variational Autoencoder (VQ-VAE) neural F
₀
model that is both more efficient and more interpretable than the DAR. This model has two stages: one uses the VQ-VAE framework to learn a latent code for the F
₀
contour of each linguistic unit, and other learns to map from linguistic features to latent codes. In contrast to the DAR and RNN, which process the input linguistic features frame-by-frame, the new model converts one linguistic feature vector into one latent code for each linguistic unit. The new model achieves better objective scores than the DAR, has a smaller memory footprint and is computationally faster. Visualization of the latent codes for phones and moras reveals that each latent code represents an F
₀
shape for a linguistic unit.