Voice synthesis model training method, and voice synthesis method and device

Abstract

The embodiment of the invention provides a voice synthesis model training method, and a voice synthesis method and device. The voice synthesis model training method comprises the steps: a first character vector sequence of a Chinese character sentence corresponding to encoding training is obtained; the first character vector sequence is encoded through an encoding module, and a first linguistic encoding characteristic is obtained; the first linguistic encoding characteristic is subjected to linguistic characteristic decoding through a linguistic characteristic decoding module, and a linguistic decoding characteristic is obtained; and according to the linguistic characteristic loss between the linguistic decoding characteristic and a reference linguistic decoding characteristic, model parameters of the encoding module of a voice synthesis model are adjusted till the linguistic characteristic loss meets a linguistic characteristic loss threshold value, and the trained encoding module of the voice synthesis model is obtained. According to the voice synthesis model training method, and the voice synthesis method and device provided by the embodiment of the invention, the complexity of voice synthesis can be lowered, meanwhile, the training accuracy of an encoder is improved, and then the effect of synthesized voice is guaranteed.

Description

technical field [0001] The embodiments of the present invention relate to the field of computers, and in particular to a speech synthesis model training method, device, device and storage medium, and a speech synthesis method, device, device and storage medium. Background technique [0002] With the development of artificial intelligence technology, people pay more and more attention to speech synthesis technology. Using speech synthesis technology and speech recognition technology, on the one hand, computers and other equipment can generate spoken language that people can understand; Human-machine voice communication can be realized. [0003] In order to realize speech synthesis, the traditional parametric speech synthesis method can be used, which is divided into multiple parts such as linguistic feature prediction, duration prediction, and acoustic feature prediction. It is necessary to build and train models for each part to realize speech synthesis. [0004] It can be ...

AI Technical Summary

Problems solved by technology

[0004] It can be seen that the above method requires the design and training of multiple models, so the training process is very complicated, and the effect of the synthesized speech obtained is not ideal, and the sound quality is flat

Images