Abstract: From the perspective of subjective and objective evaluation, physical evaluation models of sound quality are established for the rapidly accelerating intake noise in competitive vehicles. By testing the rapidly accelerating intake noise in competitive vehicles, the grade scoring method is used to perform subjective evaluation of the acquired sound samples and the typical objective parameters of their sound qualities are calculated. Correlation analysis and principal component analysis are used to study the relevance of the obtained subjective and objective indicators, and the back propagation (BP) neural network model is established with the results of correlation analysis and principal component analysis. The analysis results show that when there is information overlap in objective parameters, the principal component analysis can better reflect the correlation between the sound sample indicators to simplify the neural network input and ensure prediction accuracy.