Speech Preprocessing Method Based on Transmission Wave and Fluid Theory

ABSTRACT

With the continuous development of social economy, voice audio has become one of the important ways of streaming media and is widely used in many fields. However, it is very important how to process speech. In view of this requirement and limitation, this paper introduces transmission wave and fluid theory, tries to comb the corresponding noise reduction processing in the early stage, and by continuing the noise sample recognition, improves the effectiveness of speech processing; that is, the corresponding fluid theory algorithm for deep neural network training, using transmission wave technology to realize quantitative identification and final calculation of speech recognition model. Simulation results show that the transmission wave and fluid theory are effective and improve the recognition accuracy compared with other traditional methods.