IVCGAN:An Improved GAN for Voice Conversion

In the task of voice conversion (VC), the voice of the source speaker is converted into the voice of the target speaker, but the voice content does not change in this task. There has been some research in voice conversion and some progress and results have been achieved. In this paper, we propose a new voice conversion network based on GAN network, which is a voice conversion technique that relies on non-parallel data and is capable of converting samples of arbitrary duration. This network is called IVCGAN. The network consists of a discriminator and a generator, in which the function of discriminator is to distinguish the real speech from the converted speech, and to classify the source speakers corresponding to the speech, while the function of generator is to deceive the discriminator. We tested the proposed model on a clear and clean voice data set, and the experimental results show that this method can effectively achieve voice conversion, and its performance is better than that of baseline.