Forum Comments

Sound triggering
In Device Support
Harry J Lehto
Jun 12, 2018
I have been looking at spectrograms [will call these spectra for short] from 4 Audiomoths mainly at 384kHz but also lower sampling rates of 48 and 96kHz. The audiomoths have been physically far from each. I have placed them in vertical and horisontal positions. Furthermore the locations are such that no other source of electricity is not within 500m and most often not within 4km. At low sampling rates the spectra are quite clean, and make nice recordings. The sensitivity of the instrument at high frequenscies is good. At high sampling rates there are some problems. The good thing is that most of the instrument based issues show as vertical or horisontal bands (or their mix) only. I have not noted interference signals with siginificant "tilted" what you expected from most (European) bat calls, so in principle they are separable. However, the horisontal, internal (and external) electric like signals in the spectrum can mimic horseshoe bats and similar sounds. The most consistent non-thermal instrument noise at 384kHz is a wide spectrum buzz repeating at a rate of about 25 times/sec (which is close to the rate in a grasshopper warbler song!). This signal is mostly quite weak at audible frequencies at 384kHz sampling, but in one audiomoth it is clear all the way to ~0kHz. The same signal is present at 96KHz sampling but at at rate of about 6/sec, and appers significantly weaker. In time steps the interval of the buzzes seems to be 2^14 points = 16384 (the size of the write block?) independent of sampling. The reduction or abolition of this noise signal would clean up the spectrum significantly, and make noise triggering truly feasible. Another noise type present in all recordings and all my devices is a strong horizontal tone at ~20kHz which lasts about the same time as the previous signal, but shows many quite strong harmonics. This or its harmonics would cause most likely an extra trigger. Fortunately it occurs long intervals 2 to 22 seconds. The timing and the exact frequency varies a bit. The signal can recall a noctule bat in priciple. This sound appears to be repeted occasionally but normally followed by a loud "2^14 point type " session for a varying length of time and even up to 15 seconds.
0
0
Harry J Lehto
More actions