When i made measurment with acoustic timing reference, difference between R and L always about 315-320 ms. May be it is DSP processing of amplifier or slow android sound system in the car?..The problem is a peak in the measurements around 327 ms before the true peak, sometimes the early peak is larger than the later peak and gets chosen instead. Because of the left hand window width both peaks are included in the response in that case, and the large time delay between them results in the small frequency spacing of the ripple. Does the system have a lot of latency? FIR processing, for example?
View attachment 37083
Looking at the frequency response of the early peak it might be crosstalk in the interface.
View attachment 37084