Model training method and device for voice processing, medium and program product

The invention discloses a model training method and device for voice processing, a medium and a program product, and relates to the technical field of artificial intelligence. The method comprises the steps that a first voice sample and a second voice sample are input into a voice processing model to be processed, second generated voice is obtained, and the voice processing model extracts and decodes the characteristics of voice content in the first generated voice and the characteristics of timbre in the first voice sample to obtain the second generated voice, the first generated voice is obtained by extracting and decoding the features of the voice content in the first voice sample and the features of the tone in the second voice sample by the voice processing model; and updating model parameters of the voice processing model through the difference between the first voice sample and the second generated voice. 本申请公开了一种用于语音处理的模型训练方法、设备、介质及程序产品，涉及人工智能技术领域。该方法包括：将第一语音样本和第二语音样本输入语音处理模型进行处理，获得第二生成语音，第二生成语音由语音处理模