Like a lot of you, I watched colsto’s video about locating a recording using aircraft sounds and thought it was damn cool. My next thought was how to turn that into something I could actually use on an investigation.
So I started working with ChatGPT to build it.
The result is Doppler Lab, an open source toolkit for experimenting with the technique. It pulls aircraft tracks, downloads terrain elevation data, helps you examine recordings, and compares possible recording locations against the sounds those aircraft should have produced.
There are precompiled downloads for Windows, Linux and Mac, plus a quickstart and generated practice recording so you can try the workflow without collecting your own data first. There are also classroom materials for anyone who wants to teach the process.
Here’s where we actually stand: it still needs a lot of testing. We ran my own Tucson recording through it. The data collection and analysis pipeline worked, but the result did not support a reliable location. A control with the sound patterns reversed scored just as well. That’s a finding worth publishing, and it’s documented in the repo.
The goal is to make this technique practical enough to use consistently and quickly during an investigation. We have more work to do before we can claim that.
I’m sharing it now because other investigators, audio people and developers will bring ideas and test cases I won’t think of. Try the practice, look through the methodology, and open an issue with what works, what breaks or what needs fixing. Let’s see how useful we can make this.