Perception of tones in machinery noise and its influence on annoyance.

摘要

Earth moving machinery noise is usually very loud and contains strong tonal components. People judge the quality of the sound by using various attributes of the sound and annoyance is one of the most important issues in sound quality. The objective of this research was to identify how tonal characteristics of sounds influence people's responses and to develop an annoyance model. Recordings inside cabs of large earth moving machinery were analyzed and a sound simulation process was developed to create sounds with similar characteristics but where individual component characteristics could be varied. When multiple tones are present within a critical band they are heard as a single tone with amplitude and frequency modulation. For these cases, it is unclear how to calculate many of the existing tonal metrics. By removing the trackable portion of the frequency variation, established tonal prominence metrics could be used on the modified signal to predict the perceived tonalness of the sound. Often machinery sounds also contain families of harmonic complex tones caused by various rotating components. A series of experiments was conducted to evaluate tonal prominence of harmonic complex tones. Aures Tonality was found to overpredict tonalness of harmonic complex tones and modifications were proposed to improve its performance. Annoyance caused by this type of machinery noise was studied by conducting four subjective tests. Sounds with a range of tonalness and normalized loudness levels were used in the tests. Tonalness increased annoyance but the effect was limited compared to the effect of loudness on annoyance. Tone penalties currently used overpredicted the impact of the tonal presence on annoyance. Stimulus duration affected the responses to the least and the most tonal sounds but did not have much effect on other sounds where tonal components were clearly present. A linear model composed of modified Aures Tonality, Fluctuation Strength and Sharpness was developed. The Psychoacoustic Annoyance model proposed by Fastl was modified to reflect the effect of tonalness of sounds, but this modified model was less accurate than the linear models.