DSP Robotics Support

Again and again a strange delay.

Very thanks to so many asm guru in this forum !
I never could do this without your examples and tutorials !
And so happy and surprise when it finally works

I don't know if this one is optimizing something ?
I wanted to try this idea whatever.

This time it's a mono delay.
But it use the 4 SSE channels to skip most of the process at 1/4 rate.
4 continuous sample are packed in the 4 SSE channels of 1 sample.
The delay process those sample then shuffle the 4 sample.

The process of 4 to 1 sample could feed multiple delay.
But then the feedback must be separated from the output.
(feedback don't need to be shuffled)

So a filter or effect in this feedback loop will only have access to
multiple of 4 sample delay. (Or some misaligned channels ?)

The delay could only do time multiple of 4.
I think about an update to include on optional 3 sample delay.
But how could this one affect output and feedback at the same time ??)

A little crazy idea;)

How to transform a simple delay into a modulated one ?

It does not work with the stock delay.
But it does with the excellent simple delay of Trogz and Martin.
(the buffer must cycle all the memory of the mem)

This solution cut some of the noise but not all.
Instead of having 2 delay that are interpolated,
this interpolate the last sample with the current one.

An update of the 4 to 1 Sample Delay.

I find a more convenient way to do in 3 step:
1) Inputs Convertors. pack 4 continuous sample in 1.
They could feed multiple delays.

2) Delays engines. We could place delays in parallel or in serial.
(Despite their look they are all mono)
Output and feedback send could be the same now.
(or maybe separated to mix the volume of each delay independently)
I work on a way to shift the delays by 1,2 or 3 sample.
Could make successfully one of the 3.
There's also another way, but filtering a little.
(in this step every code works in 1/4 rate.)

3) outputs convertors. They could be feed by multiple delay to reduce the process needed.

Bypassing the step 1 and 3 will not work fine

Finally the lfo is a little less noisy with some strange technics.
Will try to tune it more but i don't know if it could be better..


Edit : The lfo seams not noisy now.
Maybe just some very little deformation ?

Anyway, i plan to make a normal interpolated delay for this particular setup.

The Float Delay

I made one in the synthmaker time.
This one is a lot better.)



Using Feedback and Rate it's possible to loop and change the speed of the loop.
Changing time reset the delay, it's more a max time, while rate change also the delay.


I try to find a way to make it more usable as a control modifier.
The problem is anything i try to overclock the ticker does not affect the stream.
(I suppose ruby will work but it take more cpu.)
It's like every ticks that are faster than 100hz will never touch the stream but will be in a same sample time..
Did you know a way to fast tick that affect stream ?
(there's a little overclock stream test module inside)