Sidechaining a gated vocal delay

After experimenting with bypassing and enabling a gainer during silent parts (not very effectively, since it’s hard to get the timing right), I’ve come to conclude that what I want to do is have a vocal with a (long) delay on it that I only want triggered during silent passages. So I was thinking of inserting a gate on the delayed vocal that I want triggered by the dry vocal. How do I go about this?

To elaborate, my current mixer setup is:

A vocal with a low cut, compressor, EQ, slight (short) delay, and reverb. Basically, a “wet” vocal.

I also have a longer delay by sending the vocal to a send track and keeping the source. This send track contains further reverb and a filter. But essentially what I want to do is only have this delay triggered when the “wet” vocal goes silent.

Any help is greatly appreciated.

My spontaneus idea would be an additional send track, put a gate on it (duck mode) then insert the signal follower, which sends to the gainer of a second send track (your long delay). Mute the (post) output of the first. It’s a bit clumsy but should work…
To clear it up a bit…
The Gate is used only to trigger the signal follower if the volume falls below a certain level. Play around with the sensivity of the SF. The only problem that i see here is that the release time of the SF won’t probably be long enough for your purpose… :-/

There’s a small difference which is unclear to me, in what effect you would want, maybe worth to think about: (Signal Follower is key in both cases)
Do you want Delay/Echoes of all words to only be hear when the vocals are silent? Then you should progress like plasmaniac said, or just a signal follower routed inversely (min higher than max) to the sendtrack’s gain (you have to fix a gainer device at the end of that track as you might have figured). Be warned, in my experience this gives a very “busy” effect, rumbling, like even the breathing pauses for the vocals are filled with echoes.
Or do you want the last word, or syllable or two to echo a little into the pauses? In that case you are better off automating the Send’s Amount (or the Delay’s Send if you got a simple Delay). Mind that you have 256x the (timeline) precision in automation tab as versed against the pattern fx commands.
Sometimes it’s even working best to copy the words you want to echo into some new samples and launch these on a separate vocal track, but then you might just wanna do adlibs again yourself.
Tip for the Gate device with vocals is make Hold and Release a little longer.

The thing is that the vocal comes in phrases, and there is a pause in between each, perfect for the delay to come through. In essence it is about echoing the last word, but automating or copying the words I want to echo is difficult to get the timing right in the pattern. I never thought about automating being more precise, but I still think having it all work automatically would be best, especially with the right time-based settings.

Sorry but I’m still confused. So I’ve created a new send track, put the gate and signal follower on it as you said, and routed the main vocal to that instead…but that’s where I get lost. Can’t I just put the signal follower and gate on the delayed send track? All I’m hearing is choppiness; I still don’t really get it… Is that why I should be putting the gate on another send track, so the delay doesn’t get choppy?

I might try this automating trick but if you could try explaining it a bit more detailed that would be appreciated. I would send you the source file if I could too but I’m using a lot of third-party effects, which I guess shouldn’t really matter. Let me know if that would help and I’ll prepare it. In fact I might just do that.

How about copy the whole sample, and silence out the parts you don’t want? Or doesn’t this work for some reason?

Ok, i see… So the vocal is also the “trigger” i.e. that part that gets quiter?
What I meant is you route it to BOTH send tracks. That additional send track is only for triggering or to “invert the behaviour” of the signal follower.

The routing looks like this (can’t provide .xrns at the moment, that would make things easier…):

Vocal track → send to S1 AND S2
S1 → Gate → Signal Follower [set to control gain of the gainer on S2] (-> Post volume of track: mute!)
S2 → your delay FX chain → Gainer

The idea is:
When the signal on S1 falls below a certain threshold, the gate is opened which then -thanks to the signal follower- cranks up the gain of the gainer on S2. Effect: Your delay chain gets louder then the source gets quiter. Again: there is no audio to be heard from S1!
I admit that it is not the ultimate solution and it needs a lot of trial and error tweaking of the various parameters…

But i suggest that you should go for automation. Although it’s more “editing work” to do, it’ll give you much more control over the desired effect and you get rid of complex routings. It’s not “real sidechaining” anymore, but who cares… ;)

If you’re wanting to repeat the last word or phrase or whatnot it can get quite complicated to try and set something up that will do it automatically! Are all the repeated section exactly the same length? One beat? One bar? If not give up now!

If they are you can do it something like this. Example assuming you want a single beat length repeated.

Play you vocals all starting a beat earlier in the timeline that you actually want to hear them. This should basically be the same length as your Delay/Echo repeats.
Use a system similar to that mention by Plasmaniac above. The purpose of the Gate it to give you as clean quite sections as possible, so you have the repeat effect brought in cleanly. The slopes/curves on the Signal Follower device, along with Sensitivity, can also help here.
Back on the Vocal track have a Delay of a single beat with 100% and 0% dry. Only the delayed sound should now play, bringing it back in line with where you originally wanted it.

Then when the vocals go quite you’ll have the first beat’s worth going and filling the Delay’s buffer, which will them be repeated as desired.

But it might be far easier to use a dummy instrument (IE one that doesn’t actually play any sound) and the Velocity/Key Tracker to trigger the Repeater or similar device (I have to admit I’m yet to really try and get the Repeater to do what I want…)

Yeah that’s what I was kind of doing before when I “automated” a gainer bypass via pattern effects, but I’ve since learned it’s better to do this via automation envelopes.

But what I really wanted was to have the whole delayed vocal come in when the main vocal goes silent, though I must admit that with the little luck I had trying I did find it became quite busy as Cas said. So I went with the idea of automating the gainer volume, which, though painstaking, actually worked rather well.

Thanks, I had a go with this new information but I guess I have to tweak it a lot longer to get the hang of it. Still don’t fully get it but I’ve done what you say so I guess it’s a matter of tweaking the threshold and sensitivity. But like I said I went with the automating method and am quite pleased with the result.

Your post makes sense, apart from this part I don’t get, as I don’t see how it will be repeated. I mean I tried it but I don’t see where the delay comes from, let alone hear it.

Anyway, I’ve gone with automating the gainer volume of the delayed vocal but DAMN, I just saved over my project after mucking around with all this and now I’ll have to do my delayed automation again! Good thing I wrote most of the points down. Always prepared for the worst. ;)

Here is the result I just managed to render:

I’d do it manually, automate a simple gainer and play the sound exactly where you want every time. I find that most times i try to find automatic solutions like this i end up spending more time tweaking it, not getting the sound i want, than if i do it manually.

I would be very tempted to simply chop off that last word and put it on a sub-track by itself w/ delay only on that track.

I said it because you were talking about enabling/disabling a gainer device to mute stuff. It works, but only per-line, whereas if you run the vocal track and automate the gain with right-click, you can set different levels 256 times per row. (It would be nice to have visual waveform running transparently, behind the automation, but that’s impossible for now)

In the link below you can find an example song file where both methods are used. No Gate device though. Ideas/Notes;
First track also has lots of devices not necessary for the example but I’m at parents’ PC now with no mic and no samples.
Dito for first send track.
Solo the first track and in the mixer view look at what the signal follower does with the gain on the S01 send track.
Note that the delay devices in the send tracks have their “send” option, aka mute source, checked.
Turn down the post volume on the original source track with the bass line, to hear the ducking taking place.
Load vocals in the instrument 01 when playback is stopped and try it out on the second track - take that one as a simple example.
3798 delayedvocals.xrns