Abstract
Judgments of annoyance caused by the sound of an idling diesel engine were determined by 80 subjects. The diesel-engine sounds were recorded and the spectra were subsequently modified electronically to produce various test signals. The subjects listened to eight different sounds in a paired-comparison procedure. Each sound signal was presented to the subjects at a time-averaged A weighted sound level of 80 dB. Two different prediction models of the annoyance response were developed by use of principal component analysis and partial least-squares regression. A new descriptor, the 'ear resonance range,' was discovered as a result of employing these analysis methods. The first prediction model, based on psychoacoustic measures, utilized the ear resonance range along with loudness and kurtosis to predict the annoyance judgments of the test signals. The second model utilized critical-band sound-pressure levels as the basis for the annoyance predictions. Both models were confirmed by internal and external validations and gave good predictions. The critical-band sound-pressure-level prediction model may provide a better means than the psychoacoustic-measure model to evaluate options for minimizing the annoyance of the sound from an idling diesel engine.