Abnormal Speech Recognition Method Based on Double Input Mutual Interference Convolutional Neural Network

Abstract

The present invention provides a method for recognizing abnormal speech in a double-input mutual interference convolutional neural network, comprising: S1. collecting a speech signal, and performing segmentation and preprocessing on the speech signal to obtain a speech sample; S2. constructing a double-input mutual interference convolution Neural network, the double-input mutual interference convolutional neural network includes a first convolution unit, a second convolution unit, a feature fusion unit, a fully connected unit and a classification output unit; the first convolution unit has 5 layers of convolution Kernel, the second convolution unit has 7 layers of convolution kernels, the first convolution unit and the second convolution unit input the same voice sample, the first convolution unit and the second convolution unit to the feature fusion unit Output feature extraction results, the feature fusion unit performs fusion processing on the feature extraction results and outputs to the fully connected unit classification output unit; the classification output unit performs classification and recognition output abnormal voice according to the processed feature extraction results output by the fully connected unit, through The invention can accurately identify the abnormal voice in the voice signal from the human body, thereby ensuring the recognition accuracy and high sensitivity.

Description

technical field [0001] The invention relates to a speech recognition method, in particular to an abnormal speech recognition method based on a double-input mutual interference convolutional neural network. Background technique [0002] Throat disease will lead to vocal cord dysfunction, resulting in abnormal speech, so it is very difficult to realize the abnormal recognition of speech signals issued by the human body. [0003] In the prior art, the following methods are adopted for the identification of abnormal speech signals: abnormal speech recognition based on phoneme spectrum, identification based on the metric method of mutual information, false neighbor score and Lyapunov spectrum, but the accuracy of these methods is poor, Although the prior art also proposes a computer-based analysis for further studies, its accuracy and sensitivity are relatively poor. [0004] Therefore, in order to solve the above technical problems, it is urgent to propose a new technical means...

AI Technical Summary

Problems solved by technology

[0002] Throat disease will lead to vocal cord dysfunction, which will cause abnormal speech, so how to realize the abnormal recognition of speech signals from the human body is extremely difficult

[0003] In the prior art, the following methods are adopted for the recognition of abnormal speech signals: abnormal speech recognition based on phoneme spectrum, recognition based on mutual information, false neighbor score and Lyapunov spectrum measurement method, but the accuracy of these methods is poor in recognition, Although the prior art also proposes a computer-based analysis of entering higher education, its accuracy and sensitivity are relatively poor

Images