Wendison

vqmivc

VQmivc is a model that performs one-shot voice conversion, meaning it can convert any input voice to any target voice. It is a neural network-based model that uses vector quantization and variational autoencoders to map the input voice to a learned speaker representation, and then generate the target voice using this representation. This allows for high-quality voice conversion between different speakers, without the need for a parallel dataset.

Read more