THe adaptive cancellation approach does sound like a much more effective, and probably much simpler to implement approach. Of course, it would require using two identical microphones, but that should not be very difficult. Back in the 1960's I did use an manually controlled cancellation system to reduce background noise a whole lot, probably about 40dB,m although I could not measure it at that time.
Why not take out all stadium noise with an adaptive noise canceller and then add back zuzu-free generic stadium roar when the real roar peaks or whenever you think it's appropriate, like a laugh track on a sitcom.
Soccer fans won't mind. They're probably mostly that same sitcom demographic
Excellent article Kendall! I'm always delighted to see practical DSP problems and solutions presented in an easy to understand manner.
You described a real-world problem involving an imperfectly known signal, followed by a methodical approach to solving it beginning with the simplest possible solution and then progressing to more complex solutions, subject to real-world implementation limitations of your chosen hardware platform.
Since the signal of interest was audio, it was great that you discussed psychoacoustics and how your ultimate solution to the problem took advantage of that.
This is the kind of article that can get young engineers interested in DSP and in a very non-intimidating manner. Thanks again for writing this.