Voice command identification method, computer device and computer-readable storage medium

Abstract

The invention provides a voice command identification method, a computer device and a computer-readable storage medium. The method comprises the steps of acquiring voice data, extracting characteristic information of the voice data, and identifying the characteristic information through a convolutional neural network, wherein the convolutional neural network is a convolutional neural network separable in depth, and traditional three-dimensional convolution is replaced by one-time channel convolution and one-time 1*1 convolution; obtaining the decimal digit of an output value of each layer of the convolutional neural network and outputting an output value of the layer according to the corresponding decimal digit preset by the layer, wherein the output values of all the layers share one decimal place which can be preset, and convolution and offset operations are completed according to a fixed-point number; obtaining a calculation result of the neural network to serve as a recognition result of the voice data. The invention also provides a computer device for implementing the method and a computer-readable storage medium. The method can simplify the calculation amount of the neural network in a voice recognition process, increase the voice recognition speed and reduce the cost for hardware realization.

Description

technical field [0001] The technical field of speech recognition of the present invention is, in particular, a speech command recognition method based on a convolutional neural network, and also relates to a computer device and a computer-readable storage medium for realizing the method. Background technique [0002] With the development of intelligent recognition technology, more and more intelligent recognition realizes functions such as image recognition and voice recognition. Image recognition and voice recognition both analyze and recognize the input data, so as to judge whether the input image is similar to the target image, or whether the input voice is similar to the target voice, so as to recognize the content of the input image, or recognize Out of the voice content. [0003] Speech recognition technology usually requires powerful hardware computing capabilities. In recent years, with the improvement of hardware computing capabilities and the decline of production...

AI Technical Summary

Problems solved by technology

However, the deep neural network has the following disadvantages in the application of speech recognition: because the deep neural network architecture does not target the highly correlated features of speech data in the time domain and frequency domain, it results in a large amount of calculation and the recognition accuracy is not ideal

The cyclic neural network is suitable for long-term speech recognition. For short-term instruction word recognition, it is difficult to show superiority compared with the convolutional neural network. Moreover, the hardware architecture of the cyclic neural network is relatively complex, and the implementation cost is high, which is not conducive to low-cost cost of smart devices used in

Images