Sound Quality Index development for the boomiong noise of automotive sound using artificial nural network information theory

초록

The recent noise control technology for the passenger car can reduce the A-weighted noise level in the car compartment to as low as possible. Unfortunately, this technology often brings out sound quality problems inside of cars since A-weight sound level will not tell the whole story as far as the customer is concerned [1]. Therefore, sound quality is becoming increasingly important as a part of vehicle design and many research papers on booming sound have been published [2, 3, 4,5]. Booming sound is one of the most important interior sounds of passenger cars. A few papers [2,3] are concerned with sound quality analysis of booming sound, whereas almost all of the research papers on booming noise are interested in the control of A-weighted sound level. Sound quality analysis basically involves using tedious subjective evaluations. In order to link these subjective evaluations to objective evaluations, psychoacoustics has been used [6]. Subjective parameters being used in psychoacoustics are sound metrics such as loudness, sharpness, roughness and fluctuation. According to most research results, this relationship between sound metrics and subjective evaluation has nonlinear characteristics and is very complex. In order to estimate this nonlinear relationship, the artificial neural network information technology (ANNIT) has been applied in the field of sound quality [7]. In the present paper, the ANNIT has been employed to derive the sound quality index for the booming sound of a passenger car. Results from this research are used for development of the sound quality index for booming sound of mass produced passenger vehicles.