Spatial Microphone Calibration – ICASSP 2020
Dickins Audio supported work and presentations at the International Conference for Acoustics and Speech Signal Processing 2020. The conference was set to be in Barcelona but with COVID it turned into a virtual presentation. Below you will find videos of the demo and paper.
We wanted to help introduce a very pragmatic and effective way of calibrating multi-channel spatial microphones. This is important in research applications where the precision of sound-field capture and reproduction is critical. The technique presented is one in use during the design, development and final calibrations of the speaker arrays that we ship. Also in the video, you can see a small dodecahedron 20 channel speaker array and the Dickins Audio 16 channel parabolic cluster microphone.
The main parameter to calibrate in a rigid microphone array is the different overall gain and frequency-dependent gain responses of each microphone element. Small variations in the microphone construction and electronics can cause variation of up to a few dB between each element. A well-constructed microphone array will tend to be consistent in physical geometry and the spatial response associated with that, so we desire a quick way to find and match the microphone levels in a spatially invariant or ‘diffuse field’ sense. A conventional and time-consuming way to do this is to take a large number of measurements in a controlled environment. In this work along with Dolby and The Australian National University, we demonstrate and show theoretically the ability to use a randomized movement of the microphone – hand waving – to obtain quickly a calibration of the microphones to within a fraction of a dB.
