3
$\begingroup$

I'm working on a project with detecting killer whale vocalizations using PAMGuard, and am basing my approach off of the methodology in "Southern Resident Killer Whale (Orcinus orca) summer distribution and habitat use in the southern Salish Sea and the Swiftsure Bank area (2009 to 2020)". Namely, I'm attempting to use the same settings for PAMGuard as the authors did, but I'm having problems with the IIR Chebyshev filter.

enter image description here

I've added the filter module, but when I go to add my values for the frequencies, I get a popup that says "the low pass cut off frequency is too high". I've adjusted the values, and this happens with any value inputted, even when the low pass cutoff is as low as 1 Hz. The only time I don't get the popup is when it's at 0 Hz... and then if I have anything besides 0 Hz for the high-pass frequency cutoff, I get a popup that says

the high pass cut off frequency is too high.

This also occurs when I switch to an IIR Butterworth filter, since someone suggested trying that out. enter image description here

enter image description here

I'm very new to using PAMGuard, and I'm just baffled of what to do here. I've done this adjustment on the filter both before and after I loaded my audio files into the program just to see if it makes a difference (it doesn't). I read through the PAMGuard manual for how the filters work, and the only thing it says about restrictions is

For Butterworth, Chebychev and FIR Filters using the Window Method, select either a High pass, Band Pass, Band Stop or Low Pass response and the filter cut off frequencies in Hz. Note that these should be greater than 0 and less than half of the sample rate.

I downloaded PAMGuard in fall of last year, so this is the latest version as far as I know. The audio files were recorded in the Salish Sea using a SoundTrap hydrophone, with a sampling rate of 48 kHz and using a single channel.

I'd appreciate any help, I have no idea of what the underlying issue may be here.

$\endgroup$

2 Answers 2

2
$\begingroup$

You have first to select Data source in the Filter module tab.

Without doing that you will get the error you described.

$\endgroup$
3
  • $\begingroup$ Thank you, this worked! $\endgroup$ Mar 15, 2023 at 23:53
  • 1
    $\begingroup$ if it solved your problem, please accept the answer $\endgroup$
    – WMXZ
    Mar 16, 2023 at 6:13
  • $\begingroup$ Just did so! Apologies for the delay, I'm pretty new to this. $\endgroup$ Mar 27, 2023 at 6:19
1
$\begingroup$

It looks to me as though your data are sampled at 48kHz, so the maximum usable frequency in the data is 24kHz. The study you refer to sample data at 256kHz, so filtering data for them may have made sense. They searched for whistles up to 30kHz, which you clearly can't do with your data. Whether this matters or not, I cannot say, but you'll certainly have to modify some of the signal pre-processing stages to get data into a similar format and can't expect the same results with different equipment.

$\endgroup$
1
  • $\begingroup$ The main issue here ended up being that I didn't have select my audio file for "Data Source", as the selected answer above said. But you're correct, I do have to adjust my high pass parameters because of my hydrophone's lower sampling rate compared to the study. So far, my approach has been just to set the high pass as the Nyquist frequency, and that seems to be working well. $\endgroup$ Jun 13, 2023 at 2:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.