TTS-by-TTS: TTS-driven Data Augmentation for Fast and High-Quality Speech Synthesis
This work addresses a specific bottleneck in fast speech synthesis for TTS applications, but it is incremental as it builds on existing non-autoregressive models like FastSpeech 2.
The paper tackled the problem of low-quality speech synthesis in non-autoregressive TTS systems when training data is insufficient, by proposing a TTS-driven data augmentation method that uses an autoregressive TTS to generate synthetic data, resulting in a 40% improvement in mean opinion score to 3.74.
In this paper, we propose a text-to-speech (TTS)-driven data augmentation method for improving the quality of a non-autoregressive (AR) TTS system. Recently proposed non-AR models, such as FastSpeech 2, have successfully achieved fast speech synthesis system. However, their quality is not satisfactory, especially when the amount of training data is insufficient. To address this problem, we propose an effective data augmentation method using a well-designed AR TTS system. In this method, large-scale synthetic corpora including text-waveform pairs with phoneme duration are generated by the AR TTS system and then used to train the target non-AR model. Perceptual listening test results showed that the proposed method significantly improved the quality of the non-AR TTS system. In particular, we augmented five hours of a training database to 179 hours of a synthetic one. Using these databases, our TTS system consisting of a FastSpeech 2 acoustic model with a Parallel WaveGAN vocoder achieved a mean opinion score of 3.74, which is 40% higher than that achieved by the conventional method.