{hhr, 12-may-2018}
We have encountered two issues so far. First, there is is a "short-circuiting" between senders and receivers when using unshielded speaker cable for the receivers. Not only do we register -20 dB cross-talk, but also the Doppler shifts appear no longer on the sonogram; if we change to a shielded microphone cable, cross-talk falls to -35 dB (cable 1) or -50 dB (cable 2), and Doppler shift is correctly registered. The second problem are spikes in the mic signal (depicted); using the MOTU instead of the Focusrite audio interface seem to fix this problem, although further investigation is needed (we should check the Focusrite's buffer sizes etc. to see if it can be rectified with that interface).
{hhr, 14-may-2018}
I added Constant-Q filter to FScape (to be included in v2.15.1), also some other necessary UGens (RotateWindow, RotateFlipMatrix). This allows me to create a high-res image of the Doppler analysis. Next step will be median filtering or smoothing, and subtracting "calibrated" background signal.
Linear spectrum, using the unshielded cable for the microphone. While the hum is not a problem, as it is very low at the high frequencies, it may well be the case that cross-talk between speaker and mic cables are so high that we cannot discount it, and therefore decided to order shielded microphone cable, even if it will probably be less visually appealing.
One afternoon chasing these strange spikes or singal break-downs, probably coming from some sort of condensor overloading / discharging, perhaps in the ADC. In the end, it seems to come from the Focusrite USB audio interface, as the MOTU 1248 does not exhibit these; still to test, though, is if the different space where the sound was recorded, was making any difference (interference).
After obtaining a "clean" signal, we can clearly observe Doppler shifts in front of the sensors; interestingly, the fundamental frequencies (here shifted down to correspond with 200 Hz) remains always there and relatively strong, which probably means that there are always enough reflections from static objects, and that we'll have to filter that out if we only want to look at the glissandi.
