Microphone arrays

Sound propagation in free and bounded space.

A satisfactory computation of the propagation of sound is yet an unsolved problem, and many application fields need this kind of calculation.

The frequency range of the acoustic signal goes from 20Hz to 20KHz, exactly three decades both in frequency and wavelength.
This means that no method of approximation can fulfill the needs of the whole frequency range. Geometrical approximation can be used for high frequencies (with negligible wavelength / dimension ratios). On the other hand, even at the lowest frequencies the concentrated-constants approximations can hardly be used.

A rigorous, full ondulatory approach has instead too high computational complexity (proportional to the cube of the shortest wavelength implied).  An alternative approach is the BEM (Boundary Element Method), in wich integration is made only over  sources, reducing thus the dimensionality of the problem.

This field, the "Concert-Hall problem", is still a research field for acousticians and musical researchers.

Our contribution

We developed a method and a SW for prediction and auralisation (in deferred time) based on geometrical approximation (Virtual Source Modeling). This method can help the ondulatory methods, reducing the frequency range for which they must be used.

A geometrical method gives a lot of interesting information about spatialisation cues, which are tied more to the high frequency side of the sound, the lowest frequencies being more or less "delocalised", just by reason of its "full ondulatory" propagation.

The work was the subject of the Thesis of Marco Giordano (Lorenzo Seno supervisor) at the Department of Engineering of Rome's "La Sapienza" University. It became also a paper at the 25° Symposium of the Italian Acoustic Association (Associazione Italiana di Acustica) Perugia, June 1996.