Research on Multi-Module Feature Fusion of Putonghua Based on Genetic Algorithm

In order to improve the accuracy of pronunciation and evaluation of Chinese Putonghua, it is necessary to deal with the multi-module feature fusion of Putonghua. A comparative method of pattern recognition is proposed based on genetic algorithm (GA) and high order spectrum feature fusion of Putonghua pronunciation in Chinese Putonghua pronunciation standardization. The acquisition model of Chinese Putonghua speech pronunciation signal is constructed, and the filter processing and adaptive feature matching of the collected Mandarin speech signal are carried out to extract the statistical feature quantity of Putonghua speech signal. The method of spectral feature analysis was used to fuse the pronunciation features of Putonghua, and the correlation was detected according to the action attributes of phonetic organs. The high order spectrum characteristics of Chinese Putonghua speech pronunciation signal are extracted, and the modeling and adaptive learning of Chinese Putonghua speech pronunciation feature are carried out by means of statistical average detection and genetic evolutionary algorithm, and the adaptive optimization is carried out according to the genetic evolution process. The multi-module feature fusion of Putonghua is realized, which provides a standardized contrast mode for pronunciation. The simulation results show that the method has high accuracy in speech pronunciation feature detection and pattern recognition, improves the ability of multi-module feature fusion of Putonghua, and is effective and reliable for acoustic modeling of Putonghua pronunciation features. It has good application value in guiding the study and training of Putonghua.