Search Results

While driving a passenger car, a driver can hear many sorts of sounds inside of the car. Among these sounds, booming and rumbling sounds are classified as the dominant sound characteristics of passenger cars. A sound quality index evaluating the quality of these two sounds objectively is therefore required and is developed by using an artificial neural network (ANN) in the present paper. Throughout this research, the booming sound and rumbling sound were found to effectively relate the loudness, sharpness and roughness. The booming sound qualities and rumbling sound qualities for interior sounds were subjectively evaluated by 21 persons for the target of the ANN. After the ANN was trained, the two outputs of this ANN were used for the booming index and rumbling index, respectively. These outputs were tested in the evaluation of the sound quality of the interior sounds which were measured inside of the sixteen passenger cars.

This paper presents a new booming index, which is developed by using psychoacoustics theory and neural network theory. The input of neural network is sound metrics for interior noise signal, which replace of the auditory system of a human. The neural network replace of the neuron structure of human' brain. The 150 sounds for the training of neural network or for optimization of the weights of the neuron have been synthesized by using a reference sound signal measured on the drive seat. The correlation for booming sounds between objective values evaluated by the trained neural network system and the subjective values evaluated by the 21 persons are well corresponded.

Rumbling sound is one of the most important sound qualities in a passenger car. In previous work, an objective evaluation method for rumbling sound was developed based on the principal component. In the present paper, the rumbling sound was found to effectively relate not only the loudness but also roughness. The last two subjective parameters are sound metrics in psychoacoustics. Principal rumble component, roughness and loudness were used as the sound metrics for the development of the rumbling index to evaluate the rumbling sound objectively. The relationship between rumbling index and these sound metrics is identified by an artificial neural network (ANN). Interior sounds of 14 passenger cars were measured, and 21 persons subjectively evaluated the rumbling sound qualities of these interior sounds. Throughout this research, it was found that the results of these evaluations and the output of a neural network have high correlation.