Voice conversion method and device, equipment and storage medium

The embodiment of the invention provides a voice conversion method. The voice conversion method comprises the following steps: extracting a bottleneck feature, a quantitative representation feature and a fundamental frequency feature of a source voice; based on the features and the identification representation of the target sounder, obtaining the voice of the target sounder by using a voice conversion model; wherein a first coding network included in the voice conversion model is used for coding the bottleneck feature into a semantic feature, a second coding network is used for coding the quantitative representation feature into a rhythm feature, and a decoding network is used for outputting the Mel spectrum of the voice of the target sounder based on the semantic feature, the rhythm feature, the fundamental frequency feature and the identification representation of the target sounder. By applying the technical scheme provided by the embodiment of the invention, when the voice conversion is carried out, the