Exponential Envelope

MichaelBenjamin · 14 Jul 2009

what do you mean with working, how does it not sound good with short notes? do you mean the step between decay and release?

MyCo · 14 Jul 2009

yes... it clicks, when then value on note off is higher or lower than the sustain level. It happens a lot when you have long attack or decay times

MegaHurtz · 16 Jul 2009

multyply it by itself if it is a 0-1 range.

MegaHurtz · 16 Jul 2009

and sqrt makes for log... if you want log it`s cheaper to do F1*F1=F2 (F1-F2)+F1="Log" sounds tight

MyCo · 16 Jul 2009

exponentiating a linear envelope doesn't make a right "semi" exponential envelope, because the attack phase is inverted. It'll have a slow start and increases fast... a right exponential envelope starts fast and decreases its rising. Had the same idea long time ago (when I started with S|M), but if you use it you'll notice, that there is nearly no peak between start and sustain.

I've found another way to make a fast "nearly" exponential envelope... it's a little bit heavier, but it should work

aliasant · 16 Jul 2009

MyCo wrote: exponentiating a linear envelope doesn't make a right "semi" exponential envelope, because the attack phase is inverted. It'll have a slow start and increases fast... a right exponential envelope starts fast and decreases its rising. Had the same idea long time ago (when I started with S|M), but if you use it you'll notice, that there is nearly no peak between start and sustain.

I've found another way to make a fast "nearly" exponential envelope... it's a little bit heavier, but it should work

If you feel like you absolutely must share that little "Scaler thing" I think we can forgive you. If not all is still ok 🙂

Just love to test new little things like that.

MegaHurtz · 16 Jul 2009

what do you mean with semi exponential.. 2^F/2.. I dont get it.

MegaHurtz · 16 Jul 2009

what i ment was this? exponential and logarithmic. (only on a o-1 scale.)

MyCo · 16 Jul 2009

Exponential is eg. 2^x, 3^x....

what you do is a parabola: x^2, x^3....

with "semi" I mean, that it isn't a real exponential, just a fake... but x^2 isn't a fake/real, it's not even near an exponential, because of the inverted attack.

MegaHurtz · 16 Jul 2009

2^(12/F).. Is what I do to calculate pitches. It`s absolutely bang on exponential. levels I do parabolic in your words. Only in my words parabolic is when you do the same trick on a sine. So you want to calculate the square root.. but not end up with a similar result. Im sorry im completely lost then.

MegaHurtz · 16 Jul 2009

I did notice some scale difference in the math, hence the innacuracy`s of pitching the F*F way. If it`s that you mean i completely agree it`s off by miles. But then again, thats why discrete versions of these functions are mathematical cpu killers. For levels, in akoustics they always refer to log10 functions, never heard of others.

Eska · 16 Jul 2009

y=1-((1-x)*(1-x))

MyCo · 16 Jul 2009

@Eska: you mean for the attack phase? yes that would do it, but you can't use it on the linear envelope, you have to write your own envelope code completely...

Eska · 16 Jul 2009

Sch... I made a linear envelope with stage dependent post conversion, but of course its in the loop, I forgot that.

MichaelBenjamin · 16 Jul 2009

yes, i also did the stage based conversion after the barebone linear envelope, left the attack linear, and exponentialled the rest (this sounded alright and seemed efficient enough). you can use the env primitives with access to the stage(adsr/adsr4), or your own adsr, or infuzions 1.7 adsr retrig but i think you already knew that so where is the problem?

MyCo · 16 Jul 2009

I thought, someone did it already, and would share it, but doesn't look like...

MegaHurtz · 16 Jul 2009

well.. you can determine if the eg is at peak level, after a key is pressed and take it from there.

MyCo · 17 Jul 2009

So... I've started a project for this: Selfmade Exponential Envelope