Identity recognition model training method and device, and identity recognition method and device

Abstract

The embodiment of the invention provides an identity recognition model training method and device, and an identity recognition method and device. The identity recognition model training method comprises the steps of: obtaining a training audio data set, and carrying out feature extraction on the training audio data set to obtain a training feature set; inputting the training feature set into a content recognition model included in a to-be-trained model for iterative training, and inputting the training feature set into the trained content recognition model to output a content vector; and inputting the training feature set into a voiceprint recognition model included in the to-be-trained model for iterative training, inputting the training feature set into the trained voiceprint recognition model to output a voiceprint vector, inputting the content vector and the voiceprint vector into a classifier included in the to-be-trained model for iterative training, and obtaining an identity recognition model when the likelihood of the classifier is the maximum and the parameters are converged. By adopting the embodiment of the invention, the accuracy of an identity recognition result can be improved.

Description

technical field [0001] The embodiments of the present application relate to the technical field of machine learning, and in particular to an identity recognition model training method, identity recognition method and device. Background technique [0002] With the rapid development of science and technology, more and more manual repetitive mechanical labor is gradually replaced by artificial intelligence. As an important branch of artificial intelligence, machine learning has been widely used in machine translation, artificial intelligence customer service, text detection, voiceprint wake-up and other fields. [0003] In the field of voiceprint wake-up, identity recognition can be used to confirm whether the user to be recognized is a pre-registered user, and then operations such as terminal device wake-up or application deployment deployed in the terminal device can be performed based on the identity recognition results. In the prior art, a related voiceprint recognition al...

AI Technical Summary

Problems solved by technology

[0004] However, the processing process of the existing voiceprint recognition algorithm is generally to first compress the obtained text-related voiceprint vector sequence to obtain the voiceprint codebook set, and then based on the user's information received in the text-related voiceprint recognition scene. Voiceprint information, determine the Euclidean distance between the voiceprint information and the voiceprint codebook set, and determine the user's identification result based on the Euclidean distance, and only use the user's voiceprint information as a single dimension to determine the identification result, reducing the identity The recognition accuracy of the recognition results may cause false wake-up of terminal equipment or related applications

Images