Effect of long-term time-varying loudness and duration on subjects’ ratings of startle evoked by shaped sonic booms and impulsive sounds

摘要

Statistics of Moore and Glasberg’s time-varying loudness and Zwicker’s sharpness have been found to accurately predict startle and annoyance evoked by low amplitude sonic booms. In a previous study by the authors where subjects judged the startle evoked by these sounds and boom-like amplitude modulated noise, regression models consisting of a statistic of maximum loudness (short-term or long-term), maximum sharpness and maximum loudness derivative predicted subject ratings the best. However, while most of the metric values of that stimulus set were weakly correlated, short-term and long-term loudness were highly correlated with each other. In addition, as the signals used in this dataset were very similar in duration (< 400 ms), it was unclear how well the models developed would predict startle evoked by longer sounds, such as sonic booms with echoes. To investigate these issues, another experiment was conducted where subjects were asked to judge the startle, annoyance and acceptability evoked by a series of sonic booms (with and without echoes), boom-like noises, distant gunfire, door slams and blast noise. The performance of these metrics as well as previously developed startle models will also be discussed.