AY UGen freqtotone method is incorrect.

[telephon]

Yes, it seems to be the freqtotone method of the AY ugen is not correct.

*freqtotone { |freq|
		// Approximate empirical...
		//^(109300 / (freq - 3.70727))
		^(110300 / (freq - 0.5))
	}

Here is a test:

Here is the datasheet: http://atariage.com/forums/index.php?app=core&module=attach&section=attach&attach_id=397711
On page 18ff. there is a formula that needs proper implementation.

```supercollider
// starting point for tests.

(
Ndef(\x, { 
	var mapping = { |x| (110300 / (x - 0.5)) };
	var freq = MouseX.kr(200, 700);
	var f = mapping.(freq);
	AY.ar(f, f, f) + Saw.ar(freq) * 0.1 
	
}).play;
)

[jamshark70]

There is also perhaps some relationship to the sample rate. I used AY successfully at 44.1 kHz and it was by and large in tune with other synths. But just now (perhaps for the first time) I ran it at 48 kHz. It sounded a bit less than a minor third too low.

This is puzzling because 1/ if the AY sample/hold behavior works by counting samples, a higher sample rate would produce a higher pitch, but instead it's lower, and 2/ the ratio 48 / 44.1 in 12ET is just shy of a semitone and a half, not even close to 2.5 or 3 semitones.

Baffled...

PS The formula at this stackexchange thread is even worse -- 440 gets about a P4 difference at 48 kHz, and it varies with frequency.

a = {
	var freq = 440;
	var toneFactor = 2000000 / 16;
	var tonemap = { |freq| (toneFactor / freq).round };
	var tone = tonemap.(freq); // AY.freqtotone(freq).postln;
	[AY.ar(tone, tone, tone), LFTri.ar(freq)] * 0.1
}.play;

a.free;

[ahihi]

the emulator uses a default clock frequency of 1773400 rather than 2MHz. accordingly, i tried

(
a.free;
a = {
	var freq = 440;
	var toneFactor = 1773400 / 16;
	var tonemap = { |freq| (toneFactor / freq).floor };
	var tone = tonemap.(freq); // AY.freqtotone(freq).postln;
	[AY.ar(tone, tone, tone), LFTri.ar(freq)] * 0.1
}.play;
)

(in addition to the toneFactor, i also changed the round to floor)

at 44.1kHz, this sounds to me even closer than the original freqtotone, but i didnt measure yet. still not sure what is happening at 48kHz though..