Hello to everyone, I have to clean my dataset of recordings concerning an African penguin colony inhabits the South African coast. In particular, since I have recordings with days of strong wind with clipped/noisy sounds, I need to detect and then remove all the files ruined by the wind (like the one attached) to create my final dataset.
Do you have some suggestions about the procedure? I did not find package or suggested procedure to fix this problem, I tried this methods:
1) to adapt HardRain package for strong wind but did not work (https://www.sciencedirect.com/science/article/pii/S1470160X19307873)
2) I tried also to use WindNoiseDetection (https://github.com/kenders2000/WindNoiseDetection) but did not work well
3) now I'm trying to use MonitoR package (https://cran.r-project.org/web/packages/monitoR/index.html) to detect all the windy event.
However, since this is a big and recurring problem of wild recordings it's strange that there are not automated and functioning packages of detection. Maybe I didn't search in the right way.
13 June 2023 10:42pm
Hi Francesca,
You pose a great question and my first piece of advice would be to cross post it to the bioacoustics stack exchange. Lot's of other people would benefit from this knowledge.
The way I would approach this problem is to measure the ambient noise across your data and then consider the ambient noise level at which 100% (or whatever makes the most sense to you) of your calls would be masked. Then exclude those data from your analysis. The benefit of this approach is that you can go back and either chose different thresholds or use ambient noise level as an effect for your statistical models.
Now doing that is the challenge, especially in R. To that end I've been working on a project do ambient noise monitoring in R that may suit your needs. I'm actively working on this so I'd be willing to help you out a bit if it's something you might find useful.
Good luck!
Best,
Kait
22 June 2023 8:17pm
Hi Francesca!
If you are just looking to detect the penguin vocalizations themselves, you might try Arbimon's Pattern Matching tool. You could make multiple templates of the penguin calls (with/without overlap with wind) to try and capture the variation that wind adds in. You'll probably still get a bunch of false positives, but the manual validation page makes it easy to quickly mark these and filter out.
I agree that it is a huge problem and definitely something that more work should be dedicated to! It's something I've also struggled with in my own PAM fieldwork in Madagascar.
30 June 2023 12:39pm
Hi Francesca,
Sorry to hear HardRain wouldn't work - what went wrong? I'd been meaning to look at adapting it for wind myself.
There are two other approaches that I know of that look effective, but very involved. One by Juodakis et al., https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13928
and one by Cook et al.,
https://doi.org/10.1121/2.0001526
but I don't have first-hand experience of either.
Cheers,
Ollie
Lars Holst Hansen
Aarhus University
1 July 2023 12:46am
Interesting subject!
I work on the tundra in Northeast Greenland and we get some quite high winds.
I am right now testing wrapping our AudioMoths with a simple windjammer. It will not work in the strongest winds, but in many cases it will reduce clipping.
I am using the Røde Dead Kitten which seem to fit perfectly! Only issue is that the last batch I got had a much tighter elastic band (perhaps something went wrong in production or they changed it).
Cheers,
Lars
Dr. Kaitlin Palmer