Design of Chinese Speech Intelligibility Evaluation System Based on Machine Learning Algorithm

Abstract

With the acceleration of globalization and the popularization of computer technique, the links between different parts of the world are getting closer and closer, and the translation between different languages has become a key research topic of domestic and foreign scholars. Speech recognition is to identify the original speech signal into the corresponding text or other forms of message that can be processed by the computer. Speech recognition technique is an important research direction in the domain of artificial intelligence, which has high research value and commercial value. The method adopted in this paper is machine learning algorithm, and the deep learning technique among machine learning algorithms is selected. In this paper, the influence of high-order harmonic distortion, high frequency and mixed elements on the quality of speech signals caused by human vocal organs and speech signal acquisition equipment is studied. Windowing and framing are performed, so that the whole speech waveform will be divided into many small speech segments with overlapping 25ms, and then appropriate acoustic feature extraction algorithm is used to extract the corresponding acoustic features from the 25ms speech segments. After research, this method ameliorates the efficiency of speech recognition, which is higher than that of traditional methods.