Sonic booms, both conventional N-waves and shaped low booms, present a sound-pressure wavernon the ground with most of the energy concentrated between 4 and 15 Hz. These frequenciesrnmatch the natural structural frequencies of a typical residential house, causing structural motionrnthat can rattle windows. The resulting high-frequency interior noise is a major contributor tornhuman annoyance from sonic booms. Earlier a large database of residential house responses wasrncollected experimentally, during which window rattle was reproduced by affixing low-frequencyrnloudspeakers to the outside of windows instrumented with accelerometers and microphones.rnRattle was detected in the recorded waveforms by a spectrogram-based method. It was observed,rnhowever, that human listeners present during these tests heard rattles that were not detected byrnthis method. The present work applies a detection algorithm based on the shift-invariant discreternwavelet transform to these recordings. The wavelet-based method is found to have better rattledetectionrnsensitivity than the earlier spectrogram-based method, and also reveals the detailedrntime history of individual rattle impacts. Comparing these impact signatures betweenrnmicrophone and accelerometer locations can provide additional physical insight into mechanismsrnof rattle-sound radiation.
展开▼