1) Harmonic sieve or autocorrelation.
2) Onset Synchrony.
This is how you group and isolate sounds.
Note from earlier today sitting in my post. Must reread source to understand.
—-
Arnold tongue is my favorite of those. Phase-locking. Each of those black thick alleys are areas of phase locking, the inbetween is the colorful chaos. Each connection to “ground” at the bottom? Fractions. 1/2, 1/3, 1/4, 2/3, etc.
—
What I’m reading is fascinating to me as it’s about pitch grouping – and the guy is going through all of the ways that “almost” but don’t work, or that are sufficient but not necessary and therefore not ‘it’.
===
Ahh! if you can’t depend on phase locking, there’s other ways. Oh, there’s one classic way that can handle stretched and compressed signal, out of phase, disrupted, etc… Manhatten something?
it’s asynchronous — hm.. gotta find the name.
===
Clock is mathematically inserted. Manchester encoding is VERY fault tolerant but it has speed limitations.
===
That was quite good.
But yes, forecasting vs forensics.
===
I’ve seen a breakdown into three as well: explain, predict or describe.
I prefer describing. It does not try to explain or predict. Say what you see. It makes a good foundation or can stand alone.
===
It gave me food for thought. Thinking: category (F0) with members (harmonics) orthogonal to the category — is the grouping and the onset time is the discrimination.
That’s why in a few clever classic songs, it’s impossible to tell two distinct instruments apart if they’re written to be played simultaneously and professional skilled musicians perform the piece.