On the topic of recognition though, I think it would be much easier to implement speech recognition using the core algorithm used here, rather than from FFT for

On the topic of recognition though, I think it would be much easier to implement speech recognition using the core algorithm used here, rather than from FFT for example. I'm only at the beginning of trying to use the frequency detection algo for something else apart from visualisation... The difficulty is that the dat it generates is much more detailed for higher frequencies, so it's hard to adapt modern ML algorithm inputs to it, I need to think more...