Goal:
I’m using FFT Spectral Analysis on imported audio clips that are associated with a Submix. The goal is to print the name of the correct music notes as the audio is played.
Current Set-Up:
I’m using Add Spectral Analysis Delegate (Target is Sound Submix) and the Make Musical Spectral Analysis Band Settings node that’s provided. Decibels are being measured to determine what the dominate music note is.
I’ve used a tutorial to get this far, but unfortunately the tutorial contained the somewhat rare Quarter-Tone Scale, which means the musical note recognition will be quite wrong for normal audio files. That means my In Num Semitones (int) is set to 24, which makes the bands. And, if it’s relevant, my starting octave is 4 and the starting note is E.
Problem:
The problem I’m having is that if I change the musical scale (In Num Semitones (int)) to something else in order to get the Chromatic Scale, Major, or Minor Scale, the note detection is no longer giving me an accurate reading and is flickering at a really fast pace. I’m a beginner so I apologise if it’s a rudimentary problem.
I’ve tried changing the values in both FFT Spectral Analysis and Make Musical Spectral Analysis Band Settings, but I’m still having a difficult time landing proper note detection due to the insane musical note flicker and wild readings.
Is there anything you can think of that I should try, and how would you go about analysing music from multiple different scales using FFT?