Speech synthesis method and speech synthesis device

Abstract

The invention discloses a speech synthesis method and device. The method includes: determining the language type to which the sentence text information to be synthesized belongs, wherein the language type includes a first language type and a second language type; determining the first basic model corresponding to the first language type, and determining the corresponding language type of the second language type The second basic model: determine the target timbre, and perform adaptive transformation on the first basic model and the second basic model respectively according to the target timbre, and treat the synthetic sentence text information according to the first basic model and the second basic model after the adaptive transformation Perform training to generate corresponding spectral parameters and fundamental frequency parameters; adjust the fundamental frequency parameters of the first language type and the second language type according to the target timbre; according to the spectral parameters of the first language type and the spectral parameters of the second language type , the adjusted fundamental frequency parameters of the first language type, and the adjusted fundamental frequency parameters of the second language type to synthesize the target speech.

Description

technical field [0001] The invention relates to the technical field of speech synthesis, in particular to a speech synthesis method and a speech synthesis device. Background technique [0002] With the development of speech synthesis technology and popularization of applications, speech synthesis services are being accepted and used by more and more users. Among users of speech synthesis services, a large part are bilingual or multilingual users, and speech synthesis is increasingly applied to multilingual content occasions. Therefore, there is a demand for multilingual speech synthesis, among which the mixed reading of Chinese and English is the most common. Users generally require intelligibility for multilingual speech synthesis, followed by accurate, natural and uniform timbre. Under the condition that the current speech synthesis technology has basically solved the intelligibility, how to synthesize natural, accurate and uniform multilingual speech has become a techni...

AI Technical Summary

Problems solved by technology

[0004] However, the current problems are: (1) the method of using data of different native speakers for different languages ​​will cause the problem of inconsistent synthesized timbres, which will affect the naturalness and user experience of speech synthesis; (2) the use of multilingual In the method of speaker data, most of the speakers are not idiomatic in languages ​​other than their native language, and have an accent, which is quite different from that of native speakers, which reduces user experience. Pronunciation is not standard enough, and the speakers who are standard in multiple languages ​​are usually professionals, and the cost of data collection is high

Images