TTS-by-TTS 2: Data-selective augmentation for neural speech synthesis using ranking support vector machine with variational autoencoder

TL;DR

This paper introduces a data-selective augmentation method for neural speech synthesis that uses a variational autoencoder and ranking support vector machine to choose synthetic data similar to real recordings, improving TTS quality.

Contribution

The study proposes a novel approach combining VAEs and RankSVM to effectively select synthetic data that enhances neural TTS training.

Findings

Selected synthetic data improves TTS quality

Method outperforms conventional data selection techniques

Significant subjective and objective quality gains achieved

Abstract

Recent advances in synthetic speech quality have enabled us to train text-to-speech (TTS) systems by using synthetic corpora. However, merely increasing the amount of synthetic data is not always advantageous for improving training efficiency. Our aim in this study is to selectively choose synthetic data that are beneficial to the training process. In the proposed method, we first adopt a variational autoencoder whose posterior distribution is utilized to extract latent features representing acoustic similarity between the recorded and synthetic corpora. By using those learned features, we then train a ranking support vector machine (RankSVM) that is well known for effectively ranking relative attributes among binary classes. By setting the recorded and synthetic ones as two opposite classes, RankSVM is used to determine how the synthesized speech is acoustically similar to the recorded data. Then, synthetic TTS data, whose distribution is close to the recorded data, are selected from large-scale synthetic corpora. By using these data for retraining the TTS model, the synthetic quality can be significantly improved. Objective and subjective evaluation results show the superiority of the proposed method over the conventional methods.