Time domain sound visualization using complex envelope of acoustic holography

Abstract

Acoustic holography is a method to predict sound field in the region of interest by measuring sound pressure or particle velocity at a plane (hologram plane). In particular, time-domain sound visualization implemented by acoustic holography enables ones to observe the predicted sound field as time goes by. The time domain visualization is useful for showing a sound field generated from sound sources which generate transient or non-stationary sound. For instance, sound from an automotive engine during run-up test or an impact sound from machinery, etc. The time domain visualization using holographic reconstruction usually takes not only several steps but also a great deal of data on time and space to obtain the reconstructed sound pressure field with respect to time. In addition, the sound pressure field often contains much more information than we want to observe. The sound visualization is frequently used as a means to provide where noise sources are, which is required to deal with noise source. The sound visualization will be useful if it shows only the information we want to see as fast as possible. Various ways to represent the information we want to see can be considered: for instance, pointing the maximum pressure point, drawing intensity flow or energy propagation, etc. The objective of this research is to define a means of representing the information we want to see (where noise sources are and how they radiate) and also to develop a visualization procedure which shows the information as fast as possible using the method. This leads us to raise two problems associated with the time domain sound visualization. One is processing time issue, and the other is the analysis problem [Y.-H. Kim, “Can we hear the shape of a noise source?” Proceedings of the 18th International Congress on Acoustics, 3357-3370 (2004).]. The former issue is caused by large number of data than we need, which means that we can reduce processing time by leaving only the data we n...