Detecting the start of the impulse responses for lateral parameters

In a lateral calculation based on omni and figure-of-8 microphones, the start is detected from the primary (omni) impulse response before filtering. It is further assumed that the two channels are synchronized and the same sample number is taken as the start of the figure-of-8 mike. Automatic detection of the start of such a figure-of-8 mike is virtually impossible, especially when its null is pointed towards the source (because the reflections may be strong compared to the direct sound).

It is not recommended to use any spacing between the microphones. Even 15 cm (6 inches) can mean a significant difference in the impulse responses due to reflection interference. If scattering from mike housing is a concern, remove one mike while measuring with the other. In other words, do not measure both channels simultaneously. But make sure that you are measuring the correct system delay (initial time delay) if measuring two 1 channel measurements instead of one 2 channel measurement.

A time difference of ~0.5 ms will probably in itself not mean much, in general. But there may be cases with strong reflections close to the early window boundary where it will lead to an additional bias in the results. Therefore avoid this difference if possible.