I'm in the early beta testing stages of my app that flags TV ad breaks to trigger an on-phone action. I have fine tuned it as far as I can take it and would love for someone to review my code to see if I'm missing anything or if additional tools could be applied to improve detection accuracy and reduce false positives. I'm open to whichever technical route you feel is most suitable: CNNs, transformer-based audio models, classical DSP, or a hybrid. The current system performs well with some programming (e.g, sitcoms) but struggles with higher-dynamic range content such as action shows/movies where explosions, music, and other audio events can resemble commercial breaks. I'm looking for someone with experience in software engineering, audio processing, signal processing, machine learning/AI, or related areas.

Rate- $30/hr, 5 - 10 hours initially needed 
Contact- Thomas Orton, 803-429-8517, tforton12@gmail.com