discussion / Acoustics  / 9 March 2026

Trying to learn a pathway between Song Meter recordings and iNaturalist observations

Hello!

I recently began using Song Meter products for biacoustic monitoring of birds. I must say that turning the SM's audio files into usable audio clips that can be exported into iNaturalist observations (without just scrubbing through all ~40 hours of recordings in Audacity) hasn't been very intuitive for me, and I'd love some help in figuring out the best plan of attack here.

Previously, I used Birdweather PUCs for this purpose and loved how they would automatically ID birds with short audio files. I was not as impressed with their battery life or durability in the field as stash-and-retrieve bioacoustics monitors. A ~$300 device is not something I want to have to replace, seeing as I do all of my conservation work out of my own pocket.

Enter the Song Meter Micro 2: $150/unit, substantial battery life, rugged-er construction, and the ability to use your own SD cards were all very exciting factors. I set one out at a local WMA cable-locked to a tree and retrieved it 2 days later. I now have 7GB of audio files I need to parse through, a process I'm mostly in the dark about.

My main goal is to create iNaturalist observations of the audio in < 30s clips, something I'd rather not have to do manually for more almost 2 days of data. I've looked into Birdnet Analyzer and Raven Pro, Arbimon, and Kaleidoscope, but nothing is screaming "use this one" to me. From my research, BNA and Raven appear to be my best bet, but I have yet to figure out how to use it to complete my goal.

If anyone has any tips or input on what I should or how I should do it, I would greatly appreciate it! I'm going to keep researching these applications until I figure out something I can use in the meantime!

 

Thank you,

Chase




Update: Chirpity isn't great! It's prone to crashing and isn't always very compliant with the location filter.

 

If anyone has any other suggestions, they would be greatly appreciated!

Hello!  I am a specialist for Kaleidoscope software.  I have some knowledge of the other applications you mention.  Let's see if I can help.

Launch Kaleidoscope.  You don't need the Pro version for this first test.  Once you get past the various welcome messages, you'll be looking at a gray window called the Kaleidoscope Control Panel.  Go to the Kaleidoscope Control Panel File menu and choose Set Defaults. You'll get a window to choose bat analysis mode or non-bat analysis mode.  Choose non-bat analysis mode.

Now back in the Kaleidoscope Control Panel, tap the Browse button on the left portion of the window (under Inputs).  Navigate to a folder which contains no more than 5GB of your Song Meter recordings.  This folder must be on your internal drive or directly connected drive. No network drives, please!

If you go into the folder it will look empty.  You are not there to select any files.  Back out and select the folder and click ok.  You have defined your input batch.

Click on the Browse button for the Output directory.  Navigate to your desktop and make a new folder.  Select that folder and you have now defined an output directory.

Click on the Cluster Analysis tab.  Under that tab, click on the Disabled menu and choose Scan recordings and extract detections (no clustering).  Press the Process Files button in the bottom right-hand corner.

After a few moments, two new windows will open.  The Results window looks like a spreadsheet.  The Viewer has a spectrogram display.

The Results window will contain many more rows than were actual files in the inputs.  The rows represent "detections".  Detections are fragments of larger files.  If you click on a row, that detection will be displayed in the Viewer.

In the Viewer, use the zoom controls so you have a nice big picture of the detection.  The Results window and Viewer are linked.  Look for the left and right arrow buttons under the bottom right of the Viewer spectrogram.  Those buttons move the view up and down the list.

In the Viewer, choose Reload from the File menu to see the entire file from which the detection came.  Go back to any other detection to go back to looking at just detections.

When a detection that you like is visible in the Viewer, go to the Viewer File menu and choose Save detection as Wav.  This is the key step to make little bits of audio which you can then hand off to programs like Chirpity or Birdnet.  You can also select one or more detections in the Results window and go to the Results window file menu and choose Save selected detections as WAV.  You'll get a separate wav file for each detection.

The next step will be sorting.  You will most likely be looking at thousands of detections.  Some of them will be animal sounds.  Some will not.  Some detections may have a single sound event.  Some detections may have multiple sounds at the same time.  The results window will appear totally scrambled at first glance.  Not to worry!  What you need to know to get started is that all this can be tuned and refined. What I have described so far is a raw first step.

I teach classes on how to do this work, including integration with third party bird ID apps.  I hope it's ok to mention that anybody can go to the Wildlife Acoustics web site and sign up for our no-charge classes.  In the mean time, I am a new member here and will do my best to provide helpful input.

Cheers!

Dave Roberts - Wildlife Acoustics