Supersaw!

[aciddose]

robin; it should init them to something like phase = 0.5+0.5*sin(5.31539 * i), but i'm not sure.

ah, here:

Code: Select all

	long n = saws;
	float z = 0.36f;
	while (n--)
	{
		phasez[n] = sinf(z);
		z += z*n; 
	}

so, it should be "random" on the first note, but there may be bugs.

oh no, i know what it is; the phases are normally free-running but they're init to zero for the main voice. the "random" phase init only applies to the "supersaw" oscillator unison, but not to the voice main phases.

yeah, so it inits to zero for global unison. nobody ever used that feature much since it's usually not extremely useful and such a cpu hog that version unfortunately sucked. i typed in -32db for the volume rather than -12db as well. what an awesome typo - little did i know i wouldn't update xhip again for years.

igno; samples you make belong to you actually. if you post them here and somebody else puts them in a sample cd, you can sue them but access _cant_. who owns the "hypersaw" samples you found online?

[longbongsilver]

I know it's gone past this already but...

The very distinct shape of the JP8000's Supersaw can easily be identified in a scope. It not only sounds different, it also looks different from all other Supersaw implementations I came across.

That wont be the case for too long:
a) is original supersaw
b) is mine
c) is layered for comparison and I shifted them a bit otherwise they will cover each other.

If that's what a cycle of the Supersaw waveform looks like then I'm shocked. There's a wave that looks like that in Zeta (even labeled as "multisaw" in the name field), but it sounds nothing like a supersaw at all whether you play it in multi mode or not.

[aciddose]

that appears to be a setting with the lowest amount of unison - assuming it's a volume fade. i don't believe anybody posted a graph of the full unison - but it doesn't really look very useful. it's more useful to mimic the shape of the single waveform and then apply a similar detune to the unison.

[Music Engineer]

for fraction 0 to 1 in n steps
result[n] = pow(12throot2, detune depth * pow(fraction, distribution))

wait - this is actually not a power-of-power but an exponential-of-power. the outer pow() has a fixed basis whereas the inner pow() has a fixed (adjustable) exponent. i think, with distribution == 1, this law would be the same as my exponential law, putting equal distances between the different saws on a logarithmic (pitch) axis.

[antto]

i still don't get that
i must be very dumb.. what's "12throot2" and "distribution"
also, what is this supposed to be used for, exactly? the "result[n]"

i always detune oscs while in the 1V/Oct domain where each 1.0 is an octave so my detuning coefficients are similar to this:
pitch_A = pitch + 0.0833333 * detune * coeff_A
pitch_B = pitch + 0.0833333 * detune * coeff_B
then, i convert these into frequency and then to omega (Fc/Fs)

[aciddose]

i think so yes. the original code went like this:

Code: Select all

... factor = 0:
for (int i = 0; i < voices; i++)
{
 factor += (1 / voices);
 delta = pow(factor, distribution) - delta;
 result[i] = pow(12throot2, 2*detune * delta - 0.5*detune);
 avg += result[i];
}

avg /= voices;
avg -= 1.0;
for (int i = 0; i < voices; i++)
{
 result[i] -= avg;
}

i think.

i'm not sure, but that should be the same as the other code? again, i don't remember when i did it but at some point switched to the other layout. the delta thing was a relic from an older method. since the power law is equal no matter which portion of the curve you zoom into, the delta values will line up on the original curve anyway making them redundant. the only difference would be a different scaling factor for "detune depth". this business with 2* and 0.5* i'm not sure about. i have no idea what that was supposed to do.

antto, yes and the problem with that is you're using plain old linear spacing (pitch, expo frequency domain) which is likely to have high correlation between ratios leading to flanging. (in fact, with that method of just space*n, the ratios are all equal!)

[antto]

aciddose: equal? what do you mean?
for a 7 oscs where 1 is always in the center (not detuned) the other 6 would have the following coefficients (erm, this is an example only)
0.31 -0.31
0.57 -0.57
1.00 -1.00
that's for the "coeff" values i mean
the exact values could be later tweaked, and i'm thinking about making this app to run this thing and find the coefficients which make the sawtooths cross eachother not too often

[WOK]

OK, I can't resist to throw in something into this discussion:
http://www.service-1.de/music/supersawtest1.mp3
(just a quick test without any effects)

[sonicpowa]

Here's my take:

Example 1
Example 2

And here's mine:

Super?

Anywhere near the ballpark/carpark/Yellowstone National Park?

No more comments?

I think my attempt sounds pretty damn close to Urs' Example 2 (the real JP8K).

That said I'm no supersaw expert, so I was hoping that the supersaw connoisseurs here might be willing to venture an opinion.

I'm not supersaw expert either, but it surely is damn close. What synth did you use? I think you did some precise eq tweaking also. The original has just a bit more clear sound, maybe due to more high freqs.

[djsubject]

Here's my take:

Example 1
Example 2

And here's mine:

Super?

Anywhere near the ballpark/carpark/Yellowstone National Park?

No more comments?

I think my attempt sounds pretty damn close to Urs' Example 2 (the real JP8K).

That said I'm no supersaw expert, so I was hoping that the supersaw connoisseurs here might be willing to venture an opinion.

I'm not supersaw expert either, but it surely is damn close. What synth did you use? I think you did some precise eq tweaking also. The original has just a bit more clear sound, maybe due to more high freqs.

+1 that is very close!!

Subz

[faab]

+1 that's very close. Impressed

[Ingonator]

I played a bit with my HyperOSC synth for Tassman 4 again. I used only 7 Oscillators, added a HPF to every Osc and switched the phase in Wavelab Elements 7. The resulting sample looks like this:

http://rapidshare.com/files/428306636/H ... tion_1.jpg

sounds a bit more "aggressive" than the other sounds/samples i made with Tassman. And also less "faz" then the others. I'll try with less use of the HPF.

[jens]

Eighteen pages about lonely-housewive techno?


Dang - it got even worse that I feared...

[Urs]

Eighteen pages about lonely-housewive techno?


Dang - it got even worse that I feared...

Indeed. Thankfully the OP got what he wanted a couple of pages ago

[aciddose]

looks like you have the hpf freq too high. you should use 12db, set resonance about half way up, or a third depending on the implementation. (would be like, 75% on some dumber synthesizers with linear control of res)

you'd want to use keytracking = 90% or so and set the cutoff slightly higher, like thirteen/fourteen semitones higher than the oscillator pitch.