Eagle owl #113
-
|
I an engaged in a Eurasian Eagle Owl project in Sweden. In my survey for breeding birds I use audio recording devices (AudioMoth) and about 200 files á 160 Mb need to be analysed. For this purpose I am using Audacity, where I analyse the frequency-time signal between 200 -800 Hz known to contain the call. However, Chirpity would be very helpful if I manage to make it work! |
Beta Was this translation helpful? Give feedback.
Replies: 6 comments
-
|
Hi @larsgunnarn, this is perhaps because you are using the default 'nocturnal birds' list. When I created this list for the Chirpity model, it was intended to filter out all species save for nocturnal migrants. I didn't consider the Eagle Owl to be a migrant species. This is perhaps debateable, and it is certainly a bird that vocalises at night. Anyway, if you use the Birds list, the Owl detections are shown: 
It's possible you will get even better results if you do not first apply the cut filters. I'm not sure, but it's worth a try. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Matt,
Thanks for the previous hint. I have attached another file with a recorded
Eagle-Owl. Chirpity does not recognize the Eagle-Owl although I am using
the Bird setting. The recorded Eagle-Owl sound is weak, but it can be heard
repeatedly.
Do you have any hint on how to make Chiripity recognise it?
Regards,
Larsgunnar
…_________________
*Larsgunnar Nilsson+46-705-233539*
Den sön 31 mars 2024 kl 11:01 skrev Mattk70 ***@***.***>:
Hi @larsgunnarn <https://github.com/larsgunnarn>, this is perhaps because
you are using the default 'nocturnal birds' list. When I created this list
for the Chirpity model, it was intended to filter out all species save for
nocturnal *migrants*. I didn't consider the Eagle Owl to be a migrant
species. This is perhaps debateable, and it is certainly a bird that
vocalises at night. Anyway, if you use the Birds list, the Owl is detected.
It's possible you will get even better results if you do not first apply
the cut filters. I'm not sure, but it's worth a try.
—
Reply to this email directly, view it on GitHub
<#86 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/A2HUH2TJUF25FFEDQ4TOQ7LY27GE5AVCNFSM6AAAAABFQGVGB2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMRYGYYDOOJQGI>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Hi @larsgunnarn, that file isn't attached. WIthout it, I can only suggest you try any or all of the following:
|
Beta Was this translation helpful? Give feedback.
-
|
Thanks!
Yet another question:
The Eagle-Owl has several different sounds, for example
- Male call ("oohu",frequency about 350 Hz, )
- Female call ("oohu", frequency above 400 Hz)
- Femail call ("kvek-kvek")
In my recording, XC361012, you can hear both the male's and the female's
"oohu"-sounds. In the recording XC461798, the male's "oohu" call is
followed by the femails "kvek-kvek".
If I use a custom list in Chirpity with just the Eagle-owl named, would all
the different calls be recognised?
//Larsgunnar
…_________________
*Larsgunnar Nilsson+46-705-233539*
Den mån 1 apr. 2024 kl 19:21 skrev Mattk70 ***@***.***>:
Hi @larsgunnarn <https://github.com/larsgunnarn>, that file isn't
attached. WIthout it, I can only suggest you try any or all of the
following:
1. Use a custom list with just Eagle-Owl on it.
2. Drop the detection threshold down to 5%
3. Experiment with the audio filters. Before doing thie, enable the
advanced option in settings to 'send filtered audio for analysis'.
4. Try both Chirpity and BirdNET models. One may be better than the
other.
—
Reply to this email directly, view it on GitHub
<#86 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/A2HUH2QE44664EWNY5B3Q5TY3GJSBAVCNFSM6AAAAABFQGVGB2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMZQGE4TGNRSGI>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
I ran the file through Chirpity here and all the calls seemed to be recognised in the first file. The model doesn't discern the sex of the caller - you won't get male detectinos or female detections, just Owl detections. The best results were with the raw audio - no filters, however, enabling context mode did pick out an additional call at the very beginning of the file: In the second file, the 'kvek-kvek' calls aren't recognised as Eagle-Owl. It associated that sound more with the Long-Eared Owl. I'm not sure whether LEO is an Owl you also get in this habitat, but you could add LEO to your custom list and check these detections for LEO vs. Eagle-Owl: 
|
Beta Was this translation helpful? Give feedback.
-
|
With the second file, the BirdNET model does appear to recognise the female calls quite well. |
Beta Was this translation helpful? Give feedback.
I ran the file through Chirpity here and all the calls seemed to be recognised in the first file. The model doesn't discern the sex of the caller - you won't get male detectinos or female detections, just Owl detections. The best results were with the raw audio - no filters, however, enabling context mode did pick out an additional call at the very beginning of the file:
In the second file, the 'kvek-kvek' calls aren't recognised as Eagle-Owl. It associated that sound more with the Long-Eared Owl. I'm not sure whether LEO is an Owl you also get in this habitat, but you could add LEO to your custom list and check these detections for LEO vs. Eagle-Owl: