English Pronunciation Transformation Text Model Based on Decision Tree Pattern Recognition Algorithm

With the development of various technologies, the speech recognition system is gradually improved to meet the various requirements. In this study, the focus is on English pronunciation transformation text model based on decision tree and pattern recognition algorithm. In the proposed algorithm, 3 aspects are considered. Aspect 1: the decision tree algorithm is improved, this study considers the multivariate joint partitioning can improve the overall performance of the integrator by bringing more diverse individuals into the integrator, then the targeted normal form distance function is considered for the better distance measurement. Aspect 2: the speech signal features are designed to make the better signal classification. Aspect 3: the novel noise removal algorithm is proposed to make the audio-text transformation are efficient. In the experiment section, the accuracy is selected as the measurement and the models such as Neural Networks (NN) and Support Vector Machine (SVM) are considered. The accuracy of the proposed algorithm reaches more than 99% on average.