Classification Retrieval and Construction of Chinese Vocal Works Library Based on Machine Learning

Chinese Vocal Music Recognition (CVMR) is a method of recognizing voice features in Chinese music works library. Since Chinese vocal music contains delicate subtleties and complicated tone fluctuations, traditional voice recognition algorithms face difficulties in recognizing vocal parts correctly. The objective is to create algorithms that can precisely identify various vocals in Chinese music library. So, an Improved Wild Horse Optimization (IWHO) method is proposed for hyperparameter optimization of the Support Vector Machine (SVM) to achieve superior recognition accuracy. Greater recognition performance is achieved by optimizing the SVM parameters using the IWHO algorithm via the efficient navigation of the complicated search space. In this work, a Chinese voice music dataset named Dataset I is used and One Note detector approach is used to efficiently extract features. The experimental findings show that the proposed SVM-IWHO works better than conventional techniques such as Migration Learning (ML) and Optical Music Recognition (OMR) through obtaining a recognition accuracy of 99.87%.