Evaluation Index for Sporty Engine Sound Reflecting Evaluators’ Tastes, Developed Using K-means Cluster Analysis

Abstract

Developments in vehicle technology and accompanying improvements in NVH performance have led to increased consumer demand for high sound quality, such as a “sporty” engine sound. As sporty sound is subjective, this study sought to express its meaning quantitatively and to develop a model that accommodates the differences in individuals’ tastes. Engine sounds of four vehicles at wide-open throttle were recorded and the signal was modulated via filtering. Acoustic and psychoacoustic parameters of the samples produced were calculated, and the preferences for sportiness were identified through jury testing. Using K-means clustering, factor analysis, and multiple linear regression, a sound quality index and a model reflecting differences in evaluators’ tastes were developed. The index was retested using new evaluators and new samples, demonstrating its reliability by high correlation. This sound quality evaluation index is useful for producing highly accurate results and reflecting the opinions of groups expressing a variety of commonalities.