Speaker recognition method combining Gaussian mixture model and quantum neural network

Abstract

The invention provides a speaker recognition method combining a Gaussian mixture model and a quantum neural network. The method provided by the invention comprises the following steps: at the training stage, framing input training voice signals, extracting characteristic parameters and generating characteristic parameter vectors; then using a K mean value method and an EM (expectation-maximization) algorithm to obtain the Gaussian mixture model parameters of the characteristic parameter vectors, and finally utilizing the Gaussian mixture model parameters of all the training voice signals to train the quantum neural network; and at the recognition stage, obtaining the Gaussian mixture model parameters of a recognized speaker, then inputting the model parameters into the trained neural network, and obtaining a recognition result. The speaker recognition method is applicable to recognition of the speaker under the condition of less sample data and unbalanced sample data, and simultaneously the capability of the quantum neural network which can carries out effective judgment on voice data with cross data and fuzzy boundary of the speaker is utilized, so that the correct recognition rate of a system can be improved.

Description

technical field [0001] The invention relates to a speaker identification method, in particular to a speaker identification method combining a Gaussian mixture model and a quantum neural network. Background technique [0002] At present, the methods adopted by the speaker recognition system mainly include the following: [0003] 1. A speaker recognition method based on Vector Quantization (VQ). In the training phase, this method first extracts the feature vector from the training speech, and then uses this feature vector to generate a speaker template through clustering; when recognizing, first extracts the feature vector of the speech to be recognized, and then calculates this feature vector and the existing ones in the system in turn. template distance, and select the speaker corresponding to the template with the smallest distance as the result of this recognition. Each template only describes the statistical distribution of the speaker's speech feature vector in the vec...

AI Technical Summary

Problems solved by technology

Due to the randomness of the distribution of the signal in the vector space, there is a certain error in the division of the input feature vector space by the learned neural network, which affects the accuracy of recognition.

Images