noise at low frequencies

Hi,

I am a new user of audiomonth and new to bioacustics as well, so my question might be naive but I still hope somebody can help me. I want to use the audiomonth to monitor owls. The species of interest (Eagle Owl) calls in the range of 300-600 Hz. When doing some tests over the last days, I can clearly hear the calls on the audio, but when looking at the spectrograms the calls are hard or even impossible to detect among the noise signals. This is not the case for higher frequency calls by other species though as the noise is limited to below 600 Hz. As I will have thousands of hours of recording later, I was hoping to see the calls on the spectrograms, rather than needing to listen to everything. Is there any possibility to reduce the noise signature other than reducing the gain? Would it help to have microphone foam around the entire audiomoth?

Thanks

Arne

4 comments

4 Comments

Arne H

Jan 07, 2020

Thanks for your thoughts Alex. I have played around a bit, but haven't found the perfect solution yet. Will keep trying.

Regarding noise produced by wind: How efficient is it to put the entire audiomoth in some mic windscreen?

Thanks

Arne

Alex Rogers

Jan 06, 2020

You could try applying a high pass filter at 200Hz which should remove some of the low frequency noise. Also limiting the upper frequency range of the spectrogram might also help as then the colour map doesn't have to include the very low signal area at higher frequencies giving you more range in the frequencies that you are actually interested in.

Andy

Jan 03, 2020

Hi Arne, If you can hear them on playback they should be visible. You may need to increase the resolution of you spectrograms. What software are you using to view the audio and what sample rate are the recordings made at? Andy

Arne H

Jan 03, 2020

Replying to

Hi Andy,

thanks for your reply. They are kind of visible, but just. I sampled at 16000 Hz and use raven lite to analyse the spectrograms.

Here is a screenshot. In this sequence I can here 3 calls. The one at 10:36 is kind of visible, the one at 10:55 is more a guess and you would probably miss it when going through a long sequence

and there is one more at 11:05 which is basically undetectable in the spectrogram.