Automatic Bird counting

I want to do the following:
Record birds 24/7 in 3 files of 8 hours
Offer these files to audacity, one at the time.
Identify the clips with sounds, hopefully birds.
Save the clips with sounds to different files, with a time tag in the name.

Then offer the files to birdnet for further identification of the birds.
The results should be a list with which birds flew when over my parabolic microphone.

I’m new to audacity and birdnet. I program in python, amongst others.
How can I do this, has anyone already experience in this area?

Possibly this :confused:

Birdnet has the best a.i based technology for bird recognition. There are several apps that make use of that technology. Like Raven and Merlin, I believe. I don’t expect problems in that realm. The preparation of the recording by audacity is what me concerns the most. A typical track contains 7.8 hours of silence or rubbish noise, and 20 minutes of bird sounds, divided over different groups of birds that are flying over the parabolic microphone at differenttimes. I want to isolate the groups, save them as separate audio files, with the time of recording as name tag.
At the same time, I want to get rid of the silence and the rubbish.

The separate files with bird sounds are then offered to birdnet. That has nothing to do with audacity.

Silence finder can label regions where the volume of the sound is above threshold.
But it has no intelligence: it will detect dogs barking, people speaking, etc.

[ Raven may have already solved that problem in their “Automated event detection” ].

Isolating the harmonic content could help discriminate bird-song from say wind & rain, (which don’t have any).

A high-pass filter at ~250Hz will exclude some non-bird noise …

You may very well be quite right. I will have to experiment with that first. Thanks!

For automatic event detection, the pro version of Raven is needed, I see. Which has a steep price of $400,-