Speaker identification method, device, equipment, storage medium and program product

Abstract

The invention provides a speaker recognition method and device, equipment, a storage medium and a program product, and relates to the field of artificial intelligence, in particular to the fields of speech recognition, deep learning, big data, cloud computing and the like. The specific implementation scheme is as follows: obtaining an initial model, wherein the initial model comprises a feature extraction network; acquiring sample features of a source domain sample audio and a target domain sample audio, wherein the source domain sample audio comprises a speaker label and a domain label, and the target domain sample audio comprises a domain label; extracting sample features of the source domain sample audio and the target domain sample audio frame by frame based on a feature extraction network to obtain source domain clause features and target domain clause features; and training an initial model by using the source domain clause features and the target domain clause features to generate a speaker recognition model, the speaker recognition model being used for recognizing a speaker of the to-be-recognized audio of the target domain. According to the technology disclosed by the invention, the training efficiency of the speaker recognition model can be improved, and the training period is shortened.

Description

technical field [0001] The present disclosure relates to the technical field of artificial intelligence, in particular to the fields of speech recognition, big data, deep learning and cloud computing. Background technique [0002] In related technologies, in the application of speaker recognition models in different fields, due to differences in data characteristics in different fields, such as data encoding, data content, data channels, and data dimensions, etc., speaker recognition models are used in cross-field There is a problem of poor recognition effect in the application. [0003] Based on this, it is necessary to optimize the interlocutor recognition model. Usually, it is necessary to collect a large amount of labeled data related to the target field, retrain the model and verify it, which has defects such as long training period and high labor cost. Contents of the invention [0004] The present disclosure provides a method, device, device, storage medium and pro...

AI Technical Summary

Problems solved by technology

[0002] In related technologies, in the application of speaker recognition models in different fields, due to differences in data characteristics in different fields, such as data encoding, data content, data channels, and data dimensions, the speaker recognition model is used in cross-field There is a problem with poor recognition effect in the application

[0003] Based on this, it is necessary to optimize the interlocutor recognition model. Usually, it is necessary to collect a large amount of labeled data related to the target field, retrain the model and verify it, and there are defects such as long training period and high labor cost.

Images