Voice data generation method and device, computer equipment and storage medium

Abstract

The invention discloses a voice data generation method and device, computer equipment and a storage medium, relates to the technical field of artificial intelligence, and mainly aims to generate morevoice data of a few sample users based on the sample voice data of the few sample users so as to balance the voice data of different users in a sample library. The method comprises the steps: acquiring sample voice data of a target user; performing feature extraction on the sample voice data to obtain voice features corresponding to the sample voice data; calculating an attention score corresponding to the sample voice data according to the voice features corresponding to the sample voice data and a pre-constructed embedded matrix, wherein the embedded matrix is obtained by training the samplevoice data; and based on the attention score, determining verification voice data except the target user sample voice data. A machine learning technology is adopted, and the method is mainly suitablefor generating voice data.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a voice data generation method, device, computer equipment and storage medium. Background technique [0002] In the speech recognition technology, for the classifier, if the difference in the amount of speech data of different users in the sample library is too large, it will be difficult for the classifier to perform well, which will affect the accuracy of classifier recognition. Therefore, in order to ensure speech recognition The accuracy rate needs to ensure the balance of the sample speech data. [0003] At present, for unbalanced sample voice data, the majority of sample voice data is usually eliminated by under-sampling, or the voice data of a small number of sample users are artificially collected for data supplementation. However, if part of the voice data in most samples is eliminated, valuable user information is likely to be lost, thereby affecting th...

AI Technical Summary

Problems solved by technology

However, if part of the voice data in most samples is eliminated, valuable user information is likely to be lost, thereby affecting the accuracy of the user's voice recognition. In addition, the way of supplementing voice data is limited by user privacy and security factors. It is difficult to obtain a large amount of voice data from a small number of sample users, and this method of artificially collecting voice data is inconvenient to operate

Images