Standard pronunciation generation method and related device

Abstract

An embodiment of the present invention provides a method for generating standard pronunciation and related devices. The method for generating standard pronunciation includes: obtaining a standard text and user audio corresponding to the standard text; obtaining each standard text feature vector according to the standard text; The user audio obtains each user audio frame feature vector, and obtains each user acoustic feature vector of the user according to each of the user audio frame feature vectors and predetermined reference identity feature vectors; the user acoustic feature vectors corresponding to each other and The standard text feature vectors are fused to obtain each fused feature vector; and the standard pronunciation corresponding to the standard text is obtained according to each fused feature vector. The standard pronunciation generation method and related device provided by the embodiments of the present invention can generate standard pronunciation with accurate user characteristics, so as to improve the efficiency of language learning for users.

Description

technical field [0001] The embodiments of the present invention relate to the field of computers, and in particular, to a standard pronunciation generation method and related devices. Background technique [0002] With the development of computer technology and Internet technology, users can use the language teaching system for language learning and pronunciation correction. [0003] First, the user pronounces according to the given text, and then the language teaching system recognizes and evaluates the user's pronunciation, outputs the evaluation result, and also gives a standard pronunciation audio corresponding to the given text, and the user can follow up the standard pronunciation audio Read to correct your pronunciation. [0004] However, when language learning is carried out through the above methods, most of the standard pronunciations in the language teaching system are obtained by manual recording or by means of language synthesis model generation. For the manual...

AI Technical Summary

Problems solved by technology

[0004] However, when language learning is carried out through the above methods, most of the standard pronunciations in the language teaching system are obtained by manual recording or by means of language synthesis model generation. For the manual recording method, due to the rich variety and huge amount of language learning texts, manual recording is difficult. The cost is very high, and with the update of the teaching content, the standard pronunciation needs to be re-recorded, and the time cost is also high; for the speech synthesis method, the existing speech synthesis model can only synthesize standard audio with a fixed speaker's timbre, and cannot The current users make changes, and due to the differences in pronunciation characteristics, pitches, and ranges of each person, the imitation effect of users based on standard pronunciation is not good, which affects learning efficiency and learning experience to a certain extent.

Images