Reducing over-smoothness in speech synthesis using Generative Adversarial Networks
This addresses unnatural speech synthesis for applications like text-to-speech systems, but appears incremental as it applies existing GAN techniques to a known bottleneck.
The paper tackled the problem of over-smoothness in synthetic speech, which reduces naturalness, by using a GAN-based image-to-image translation method on mel-spectrograms, resulting in greatly reduced over-smoothness and more realistic speech.
Speech synthesis is widely used in many practical applications. In recent years, speech synthesis technology has developed rapidly. However, one of the reasons why synthetic speech is unnatural is that it often has over-smoothness. In order to improve the naturalness of synthetic speech, we first extract the mel-spectrogram of speech and convert it into a real image, then take the over-smooth mel-spectrogram image as input, and use image-to-image translation Generative Adversarial Networks(GANs) framework to generate a more realistic mel-spectrogram. Finally, the results show that this method greatly reduces the over-smoothness of synthesized speech and is more close to the mel-spectrogram of real speech.