QDelay

Hi Tilr,
First off - thank you for this gem. This could be the best 2 tap delay ever, if you'd just iron out one mistake (that sadly a lot of developers actually do):
For the ping pong mode to actually work correctly, you mandatorily need to sum up the two input channels to mono, before you pan either left or right and feed the wet portion (the delay unit itself).
Meaning, the "width" knob in the mix section must do panning (not balance, which it does right now). because when doing balance, you actually discard the other input channel entirely.
So f.e. if you have a synth that feeds the delay in ping pong mode and you set the width slider in the mix section to one side, and the synth maybe pan modulated by a slow LFO, the delay does not get input on the opposite side - hence when the signal from the synth is fully on the opposite side the width knob is set to, there's simply no delay as the delay discards the opposite channel. So in ping pong mode essentially what you have to do is, that when the width knob is centered, the signal that feeds the wet portion of the delay, is stereo. However - the more one puts the width knob to one side, the opposite side must be summed up to the side the knob is set to (that's correct stereo "panning"). So essentially when the width knob is fully left, the right input channel must be summed up to the left side, so that both, the left and the right channel are summed up to the left, and vice versa. That way one does not loose the other input channel - which is absolutely crucial for a ping pong delay to work correctly.
The reason i am being so strict here is, that if this is fixed, this delay can simply do everything a 2 tap delay could provide. It would be my end-all delay when it comes to 2 tap delays.
So yeah... I hope i was clear. It's actually an easy task, but the explanation might be not and my english might not be good enough to explain this issue.

Great tip, indeed it loses sound if I use a single channel of audio, thanks never occurred to me, will patch the following days.

Hi Tilr,
wow - that was fast. :)
Yes please - that would be nice.

... add on:
I just realized that there's no way to route the saturation (tape/tan-h) into the feedback path. Which is a shame - as this would allow for pretty realistic tape delay emulation. Maybe you find a way?

Check the github pages, I explain that adding saturation in the feedback path increases energy each pass and makes the system unstable. I don't think its doable, not from my experiments anyway.

...well - that's actually what it's all about. And yes, it needs a bit of tweaking but the idea is, that the saturation actually makes the feedback go over 100% (which is what the eq also is doing when pushing it. The saturation here should actually be the limiting factor here - literally. If the feedback explodes due to the eg or the saturation, the user must turn down the feedback, just like in an analog tape delay. That's actually the fun about these tape delays - to play with the increasing saturation, the delay time and the feedback,

Thanks, I'll have a go at it again then, meanwhile I fixed the Ping-Pong in latest patch, it now balances stereo and mono input according to Width, hope it works well and the balancing is done right, note its difficult to obtain a stable gain with changing width but I think it works well enough.

Great - i can confirm the mono summing works as it should now. Thanks so much. :).

Nice, next patch however its improved, the balancing in last version mixes stereo with mono according to Width, next patch is pure mono and preserves the RMS amplitude.

Great! off to check. Thanks again for your effort. :).

Hi Again,

Just to let you know that I added saturation to feedback path (optional) in latest patch, thanks for both suggestions, one fixes the ping-pong mode the other sounds much better for tape like saturations - requires tweaking the feedback or Trim volume but the result makes a big difference.

Oh man - gorgeous! Will check asap - thanks again! :).

I can confirm that it works flawlessly. :).

I have one further gripe tough with the ping pong mode:
I don't know the internal feedback routing - but in a traditional ping pong delay, when you put feedback to 0, normally there's always both delay taps once (both at the exact same level). This ensures that you do not have a hangover to one side in terms of the delay level. Which is not only great for short slap back delays, but also important if you use the ping pong delay on pads - because that way you don't have that overhang to one side. Basically in a ping pong delay the feedback is after the first two delay taps. Delay tap 1, delay tap 2, then feedback.

So you're saying the feedback multiplier is only applied every two taps? Easy to do but I must be sure it is correct.

If I do so, the taps will only decrease in volume every two taps, I can confirm this is the behavior in ReplikaXT so I'll probably adapt as well after some testing.

Thanks again for the suggestion making this delay better.

Got it to work, feedback is applied on a single channel only.

Now I had to leave this as optional, off by default, the reason is that it does not work as good with ping-pong Width and for example it kindof breaks the Ping-Pong-Mid preset/config, from what I understand this is how classic Ping-Pong used to work and nowadays is less common, in either case I think it does sound better.

Will be pushing this patch tomorrow, its an option in the settings menu.

Uploaded as a delay mode instead of setting, select it on the menu where you select Tap, Normal, Ping-Pong.

Hi Tilr,
i can confirm it works great.
Here are some things you might want to consider - though these are not urgent or mandatory:
- in classic ping pong mode the delay itself is louder than the dry input signal at 50% wet (0 feedback, no eq, no drive, no modulation, etc - just plain delay). You might want to compensate for that so that at 50% input equals output level (without feedback).
- you could consider to make every "section" (diffusion, mod, drive, tape and pitch shift) on/off switchable. Not always one needs all these options, yet they use cpu, even if unused. You could do that either by placing a clickable "dot" below each label, which is grey when switched off and the same pink you use in the plugin, when on. Or you could simply make the labels clickable - again, grey label means off, pink means on (initially i instinctively wanted to click on these anyway). :)
- this goes also for the individual eq bands - on/off switches per band would be nice.
- you could consider to make the eq switchable between input and output with a drop down menu on the "input" label (just like you did with the eq/saturation pages dropdown). The reason i am saying this is, that the saturation can cause a lot of mud below the frequency of the actual input content - with an eq at the output one could deal with this very effectively (imho even more useful than an eq at the input).
- as it can be a tape delay it would be nice if the delay would feature a bipolar knob in % (where fully left is time 0 and fully right is multiplied n - centered the knob does nothing), so one can smoothly in/decrease the delay times regardless on wether they're synced or not. That is the actual fun of tape delays - to play/automate the delay time.
- the last thing you might consider is, to actually allow for another page in the right area where additional delay based functions could be placed. To start with you could, instead of summing only the ping pong delay modes to mono at their wet input, you could allow that for every delay mode - and instead of using a switch, you could use a knob (so one can freely dose it). Left is summed mono, right is fully stereo. Because not only the ping pong section would benefit from this, but also the other modes. f.e. to have the "normal" mode summed up no mono and then pan it to one side. This is what Jarre did all the time on his older records - he used the original sound panned to one side, and the delay panned to the opposite. Do this with two instances oppositely panned - not only you get a very nice stereo image on your mixes, but also it instantly sounds typical for that time area. That in conjunction with the saturator and the tape - even more instant vintage feeling.
- on that page you could then place other routings in future updates, just as you see fit when you see fit.

By the way - thanks for allowing delay times of 0. this makes it possible to use the eq and the saturator as just such - without the delay. And might i say this not only sounds great, but also enhances the usability of the plugin - it is now not only a delay, but a flexible saturator with eq (especially if one can route the input eq to the output instead) and a pitch shifter. :)

These are my findings. Again - these are only suggestions and not mandatory necessities, but who knows, maybe you see it the same way. :)
However - thanks for your absolutely awesome work and for free. (consider a big "heart" emoticon here).

Forgot two things - you could consider to:

- allow for phase inversion per tap with a switch
- allow for the feedback slider to be bipolar

The reason i'm saying this is, that one can do actually nice flangers with Qdelay - which could be greatly enhanced by these two options.
But it's good now - i've already stolen enough of your time. :).

Thanks you understand more about delays than I do.

Will take some time to process this, all I can say for now is that EQ bands can be bypassed by right clicking them (maybe I can include an additional bypass button), and most everything unless is not processed when turned off, including shelf/peak EQ bands when at zero gain, unless I forgot something.

Will let you know when I make a new patch based of this, big cheers emoji.

Hey, I'm trying to go through your points, a few things I am not sure I'll be able to implement:

- PingPong classic loudness: I tried with an oscilloscope, the volume is the same for dry and first PingPong taps, maybe you're setting the delay time too low and they overlap?
- EQ Band bypass: Added Bypass switch to the band button instead of just right click.
- EQ Output: Added
- Toggleable sections: Not sure, maybe I'll add this later, like I said, things at zero are not processed, my problem with this is that not all sections have a label, eg: Pitch and tape share the same label.

Now the more tricky ones:

- bi-polar feedback knob - Can you explain how this works? Vital synth does this but I don't understand the difference, have to admit I haven't searched yet, by the time you reply maybe I'll have figured out.

- panning the output - cant this be accomplished with the current pan knobs? how is it different? Note the PingPong is not summed to mono at the end, it is summed to mono before it enters the delay. Also dont understand this "Left is summed mono, right is fully stereo.", right is stereo?? how can one channel be stereo if its one channel? Sorry for the confusion.

- Phase inversion per tap, that's easy to implement, what is the use-case? Just to make sure its implemented right.

- The bi-polar knob to either set a multiple of the delay time, or as a percentage like other delays do would be welcome, my problem is lack of space in the UI.

Just noticed that the panning in QDelay mutes either of the channels so it does not allow full panning to one of the sides, maybe that's the problem.

Ok, i'll try to answer here step by step as well:
Re - classic ping pong level

here's what i did:

• I set the delay times to 500ms.
• No feedback.
• No modulation.
• No reverb.
• No saturation.
• No tape.
• No pitch shift.
• No eq - not at the input, nor in the feedback.
• 50% mix.
• no overlaps - meaning the input is shorter than the delay times.

The result here is, that on my dbfs peak metering shows around 3 dbfs more level on the left and on the right tap (individually) than the level of the input.
However, it's not a biggie to me. But nitpickers might nitpick. :).

Re - toggleable sections to save cpu:
Well, when everything is used, the CPU hit is justified. But when f.e. only the delay ist used - nothing else - then imho it's not. This to me indicates that something is using a significant portion of CPU - even though only the delay is used and everything else is not used.

Re - bipolar feedback knob:
This means, that the feedback is positive (not phase inverted) when the knob goes to the right, and it is negative (then it is phase inverted) when the knob goes to the left. This is great for flangin fx or everything with relatively small delay times that are being modulated by the the LFO.

Re - panning the output:
You answered it yourself with the last individual answer you gave. :) In other words - also here the panning is not panning, but ballance and hence discards the opposite channel in the stereo signal.

Re - phase inversion per tap:
Also here this comes into play with short delay times and modulation. The phase inversion partly cancels the signal and can lead to a different, often wider stereo image and can add a distinctive character, which is some cases is preferred.

RE - bipolar delay time scale-knob:
I feared you'd say this - and it is understandable. On the other hand - that is what is a very strong part of a tape delay. It's an essential part of tape/dub delays and enhances the sonic outcome greatly for quite some music styles. But of course it is your product - you're the boss - so it's your choice on whether you'll find a way to implement it or not and if it is worth the effort that hast to be put in.

Strange, something is odd with your tests, I set the same same settings regardless of selected mode "PingPong Classic PingPong etc" the feedback taps are never louder than the input...

I have a patch to release in a few days that addresses most of the things you pointed out.

Unfortunately for the delay multiplier or offset I haven't found a solution yet due to lack of UI space, other than that lots of your suggestions are covered.

About CPU usage is pretty low here, only 0.06% unless diffusion is used which takes quite a bit more.

Cheers,

Released a new patch with most your suggestions, including sum-to-mono panning (instead of collapse), output EQ, negative feedback and other things.

About the CPU usage, maybe its the waveform drawing, try disabling it, see if it improves CPU usage (Reaper meter does not account for UI drawing).

The delay multiplier will have to wait, I think I have a clue where I can implement this but haven't done so yet.

By the way the negative Feedback already flips the polarity per tap, each time the feedback is fed into the delay its multiplied by a negative and flips.

Regards,

Hi Tilr,
just a heads up - sorry for being so unresponsive, a lot to do here. Will try as soon as possible and let you know - might take some days though. Thanks again for your immense effort. :).

Hi Tilr,
sorry for the late reply. Just had a go with the latest version (1.0.7) - and it works marvelously here. Patiently waiting for the delay multiplier implementation.
What i noticed though is, that if the delay times are short enough that the LFO depth can go lower the 0ms delay, some strange things happen here.
Add on: The drawing of the analyzer is not the factor for higher CPU - at least nothing significantly changes when i switch it off.
Thanks again for your effort. Already Qdelay is my go-to delay in terms of 2-tap delays. Plus the "bonus" to use it as a pure and very flexible saturator (without the delay) - love it. :).