EagleLogger is an app I made to help log checkins to the EagleNet (I'm the NCS on Tuesdays). It's been an ongoing project with a few simple features with a stack that's usually a reflection of my professional environment. This version, the third, is the most streamlined and stable yet, being the easiest to maintaine while also costing the least in overhead.
Birder.ai is an app my wife invented (originally as an animal sound identifier) that uses the mic on your phone to listen to a few seconds of bird audio. That audio is transformed into a spectrogram and then send into an ML model that we trained to recgonize over 1,000 different species.
Processed over 300,000 audio files for over 4,000 species
Scraped Wikipedia for information on each animal to separate bird/non-bird
Used Sound Event Detection - training multiple models to isolate significant audio segments
Trained model and processed audio in python, converted to JS with TensorFlowJS
The app uses Web Audio API for cross-platform microphone access