System and Method for Sound Translation Based on Generative Adversarial Networks

摘要

Acoustic modification systems and methods are disclosed. In this example,When transforming between mutually unpaired sounds, by applying training based on the self-attention module and characteristic loss function, it is easy to transform to simulate the target tone, but it is It is an object to provide an acoustic modification system and method in which the realism of the modified sound is increased by maintaining the same elements.