I see. Well, I'm sure you'll find a solution.tilr8 wrote: Tue Feb 10, 2026 7:17 pm I've spent the afternoon around this and its not a trivial.. I haven't given up on this diffusion problem but may take a while as I have a bit less time now.
The problem is that to do pre-smear or reverse smear, future audio samples are needed, in Deelay the smear size and delay size seem to be linked, also depending on the settings the smear is not symmetric around the center tap, these are just clues on how this may be achieved because I have no idea so far.
Like I say this may take some time, I'll open github issues for this and Sinkmusic suggestions so I don't forget.
QDelay - new free Dual-Delay
- KVRian
- 739 posts since 20 Jul, 2009
-
- KVRian
- 1119 posts since 4 Jan, 2007
I've done this on the delay I coded on MixMaxtrix. Put some Schroeder allpasses on the feedback loop. Play with the delay line sizes and coefficients, but make the coeffients do nothing when disabled (so no mix of the delay signal). If you have two taps the coefficients may vary.tilr8 wrote: Tue Feb 10, 2026 7:17 pm I've spent the afternoon around this and its not a trivial.. I haven't given up on this diffusion problem but may take a while as I have a bit less time now.
The problem is that to do pre-smear or reverse smear, future audio samples are needed, in Deelay the smear size and delay size seem to be linked, also depending on the settings the smear is not symmetric around the center tap, these are just clues on how this may be achieved because I have no idea so far.
Like I say this may take some time, I'll open github issues for this and Sinkmusic suggestions so I don't forget.
Basically the same thing you would do on an allpass loop reverb, but keeping the diffusion low.
Let me have a look at the code...
-
- KVRist
- Topic Starter
- 171 posts since 17 Sep, 2023
Let me know if you find something, what you describe is just diffusion in the feedback loop? If thats the case it does not produce the reverse swell that you can find on Deelay.rafa1981 wrote: Tue Feb 10, 2026 8:21 pm I've done this on the delay I coded on MixMaxtrix. Put some Schroeder allpasses on the feedback loop. Play with the delay line sizes and coefficients, but make the coeffients do nothing when disabled (so no mix of the delay signal). If you have two taps the coefficients may vary.
Basically the same thing you would do on an allpass loop reverb, but keeping the diffusion low.
Let me have a look at the code...
Thanks!
-
- KVRian
- 1119 posts since 4 Jan, 2007
Here. https://codeberg.org/RafaGago/artv-audi ... _delay.hpp
It seems I was using modulated allpasses and also a rotation matrix.
BTW, that delay also does L-C-R. L-R-C, etc if you want to get an idea how it's done.
It seems I was using modulated allpasses and also a rotation matrix.
BTW, that delay also does L-C-R. L-R-C, etc if you want to get an idea how it's done.
-
- KVRist
- Topic Starter
- 171 posts since 17 Sep, 2023
Hey, its still a bit early but the diffusion seems to be symmetric in your case.
I am not sure how this is done but maybe its a clue and will help solving this problem.
If nothing else there is DSP code here to review for days.
Thanks a lot.
I am not sure how this is done but maybe its a clue and will help solving this problem.
If nothing else there is DSP code here to review for days.
Thanks a lot.
-
- KVRian
- 1119 posts since 4 Jan, 2007
Note that I was talking about the diffusion.
For the pseudo-reverse I have never done it, but isn't this just having a buffer with equally spaced taps (points at where you take an output) and applying an ascending volume curve from head to tail?
Then of course this buffer would have no feedback itself, but it would be fed into the delay afterwards from the taps.
I think the midiverb had one of those reverse delays and on a neighboring thread there are the decompiled formulas for all the algorithms. At some point I might add/tweak some to TurboPaco, but I have quit coding plugins; it's a very deep rabbit hole
For the pseudo-reverse I have never done it, but isn't this just having a buffer with equally spaced taps (points at where you take an output) and applying an ascending volume curve from head to tail?
Then of course this buffer would have no feedback itself, but it would be fed into the delay afterwards from the taps.
I think the midiverb had one of those reverse delays and on a neighboring thread there are the decompiled formulas for all the algorithms. At some point I might add/tweak some to TurboPaco, but I have quit coding plugins; it's a very deep rabbit hole
-
- KVRist
- Topic Starter
- 171 posts since 17 Sep, 2023
Haha got it, I thought something similar with something like I do with the reverse delay, its more complicated though, I'll try to figure how this Artv DelayVerb produces symmetric diffusion taps first. I hope I am testing the right one, couldn't find the "Diffuse Delay" delay mode you suggested, using this instead:


-
- KVRian
- 1119 posts since 4 Jan, 2007
Yes, it's that one. It's been a while since I do some music.tilr8 wrote: Tue Feb 10, 2026 8:51 pm Haha got it, I thought something similar with something like I do with the reverse delay, its more complicated though, I'll try to figure how this Artv DelayVerb produces symmetric diffusion taps first. I hope I am testing the right one, couldn't find the "Diffuse Delay" delay mode you suggested, using this instead:
EDIT: I have been reading at the code unfortunately I was very stingy with the code comments but it seems to be as I remember it; some diffusion as one would do in an allpass-cascade reverb but keeping pretty short and "dry".
There is also some rotation matrix there, as it would be used on a FDN reverb, but I think that one serves a different purpose.
Last edited by rafa1981 on Tue Feb 10, 2026 9:03 pm, edited 1 time in total.
-
- KVRist
- Topic Starter
- 171 posts since 17 Sep, 2023
Got it, I'll be checking the source code to see if it does what I'm looking for, the spectrum looks nice, symmetric diffused taps, but I am not sure if its centered on beat for example, will take some time, thanks again for all these references, funny I never saw this and I search for open source implementations a lot.
