less then 10 samples delay in code??

[nix]

awesome work to be shuffling asm like that!

[Tepeix]

Thanks ! This one is very simple, not using array or rndint.
I updated the file.
I forgot to mention that the mod input must be in -1 to 1 range or it will break feedback or raise the volume .(
For more depth an other code are needed or we have to use more unit in series.

[nix]

here's the delay I wrote a long time ago, in quasi dsp==>
https://bit.ly/ECHOecho
yah- I can't write in assembly, just looking there to see what it(Assembly) can give me as modular devices
...sorta getting this vastly logical paradigm more
-and u have wiped the board with me mate heh
keep shufflin'!

[Tepeix]

Very cool delay !
I could not open it in flowstone because i'm on 3.6.
But i try it with LMMS and like it a lot !)

[Tepeix]

I make another one that is more minimalist.
Hard for me to measure which one is faster for the same depht.

This one have only 2 samples delay and the mod mix only those 2 sample.
To get the same depth more of this one must be used in series.
But there's no more conditional line inside so maybe faster ?
Well, it's more a vibrato than a delay

Code: Select all

streamin in; streamout out;
streamin mod; float a1=0;
float a=0;
movaps xmm0,in;
movaps xmm1,a;
movaps a,xmm0;
movaps xmm2,a1;
movaps a1,xmm1;
//mix mod//
subps xmm1,xmm2;
mulps xmm1,mod;
addps xmm2,xmm1;
movaps out,xmm2;

I first get some problem with feedback at more than 0.5.
Then including a denormals in the loop it's gone... ?
The more the delay is short the more the feedback must be down ?
In the example we have 16 delay, so a total of 16-32 sample delay.
Also the quality is maybe better than the other that was skipping some sample when mixing the mod.
This time the mod input must be 0 to 1. Negative value give very bad and loud sound !

[nix]

would you like the 4 Alpha?
you're welcome to it if u have purchased

nice work

[Tepeix]

Thanks !
I don't know where to get the alpha but i just get recently Flowstone so i wait a little.
I was using Synthmaker before.

Now i try to see if my little delay make some sense to use.
I didn't find a way to precisely check the cpu.
I think for a very little vibrato it could be faster. But not when we need more depht.

There's something different from a normal vibrato, an interpolated delay
will only mix 2 sample but here we have a lot of sample mixed in series.
So when the Osc is in middle position (or in +0.5 -0.5 for the first one) the signal lost a lot of treble !
It's a sort of wha/vibrato

I try to find a way to mix more sample but maybe it's better with an array.