Distribution augmentation for low-resource expressive text-to-speech
This addresses data scarcity for TTS developers, but it is incremental as it builds on existing augmentation methods.
The paper tackles the problem of overfitting in low-resource text-to-speech by introducing a data augmentation technique that generates new training examples without extra data, resulting in improved speech quality and robustness across various datasets and models.
This paper presents a novel data augmentation technique for text-to-speech (TTS), that allows to generate new (text, audio) training examples without requiring any additional data. Our goal is to increase diversity of text conditionings available during training. This helps to reduce overfitting, especially in low-resource settings. Our method relies on substituting text and audio fragments in a way that preserves syntactical correctness. We take additional measures to ensure that synthesized speech does not contain artifacts caused by combining inconsistent audio samples. The perceptual evaluations show that our method improves speech quality over a number of datasets, speakers, and TTS architectures. We also demonstrate that it greatly improves robustness of attention-based TTS models.