Plug-in Delay Compensation broken with external side-chain

[NateVanDeusen]

It would be nice to have the issue resolved but unfortunately we just have to stay with Logic 10.6.1 and can't update logic 😒

[steinweg]

Thats really a bummer...

But did anyone have any luck working around this horrible PDC by using the IAC Driver to MIDI Trigger something like Cableguys Volumeshaper, instead of using a compressor with external SC?

I just tested it, it seems to may actually be a solution (at least in terms of a work around on consumers end), but im still working on getting it to react and interpret the MIDI Trigger more reliably, which is sometimes not really picked up by the plugin (ShaperBox 2 in my case).

If anyone wants to join my experiments:

https://www.cableguys.com/midi-trigger-in-logic.html

[Flatliners]

Hey all, it seems like in logic 10.7.8 they fixed the issue (the sidechain signal was not properly PDC‘ed when feeding the sidechain of a compressor that is on an Aux/Bus in a sessions with latency-inducing plugins).

I‘ve tested it with Trackspacer, ProC2, ProMB, Logic onboard compressor. All kinds of scenarios: Empty session, Full session, latency-inducing plugins on aux/on tracks. Sidechain-Compression was tight in all scenarios.

Test System is: MacMini M2Pro, MacOS Ventura 13.3.1

 

[Ersin]

On 7/2/2023 at 10:36 PM, Flatliners said:

Hey all, it seems like in logic 10.7.8 they fixed the issue (the sidechain signal was not properly PDC‘ed when feeding the sidechain of a compressor that is on an Aux/Bus in a sessions with latency-inducing plugins).

I‘ve tested it with Trackspacer, ProC2, ProMB, Logic onboard compressor. All kinds of scenarios: Empty session, Full session, latency-inducing plugins on aux/on tracks. Sidechain-Compression was tight in all scenarios.

Test System is: MacMini M2Pro, MacOS Ventura 13.3.1

I’m on Ventura with M2 Max and this is not the case for me 🤷‍♂️ 

[Flatliners]

3 hours ago, Ersin said:

I’m on Ventura with M2 Max and this is not the case for me 🤷‍♂️ 

Can you give an example scenario? I’ll try to reproduce it. 

[sunbrother]

On 7/2/2023 at 5:36 AM, Flatliners said:

Hey all, it seems like in logic 10.7.8 they fixed the issue (the sidechain signal was not properly PDC‘ed when feeding the sidechain of a compressor that is on an Aux/Bus in a sessions with latency-inducing plugins).

I‘ve tested it with Trackspacer, ProC2, ProMB, Logic onboard compressor. All kinds of scenarios: Empty session, Full session, latency-inducing plugins on aux/on tracks. Sidechain-Compression was tight in all scenarios.

Test System is: MacMini M2Pro, MacOS Ventura 13.3.1

Are you sure you didn’t have low latency mode enabled?

[miguel___]

I am not having this issue in Logic 10.7.9. The test that many people confirmed doesn't seem to show any delay in my system.

[sunbrother]

9 minutes ago, miguel___ said:

I am not having this issue in Logic 10.7.9. The test that many people confirmed doesn't seem to show any delay in my system.

I think without some sort of master test kept updated by a thread owner this is bound to happen since they keep fixing pieces of the issue and breaking others. Some of the tests in this thread aren’t problems anymore, but I just had this issue in a recent session so it’s definitely still happening in some situations.

Edited October 6, 2023 by sunbrother

[sunbrother]

On 7/2/2023 at 5:36 AM, Flatliners said:

Hey all, it seems like in logic 10.7.8 they fixed the issue (the sidechain signal was not properly PDC‘ed when feeding the sidechain of a compressor that is on an Aux/Bus in a sessions with latency-inducing plugins).

I‘ve tested it with Trackspacer, ProC2, ProMB, Logic onboard compressor. All kinds of scenarios: Empty session, Full session, latency-inducing plugins on aux/on tracks. Sidechain-Compression was tight in all scenarios.

Test System is: MacMini M2Pro, MacOS Ventura 13.3.1

I'm pretty sure this wasn't fixed. Trackspacer, ProC2, and Logic's compressor are in basically zero latency plugins and ProMB's delay is going to be ~15ms. In fact, I think any workarounds I posted earlier only serve to reduce the delay to the individual plugin receiving the sidechain. If you do the same test with soothe and listen to its sidechain cue it's impossible to get the delay below 42ms, which is a 2048 sample delay at 48khz, and the exact latency of soothe.

Edited October 10, 2023 by sunbrother

[Flatliners]

On 10/10/2023 at 8:35 PM, sunbrother said:

I'm pretty sure this wasn't fixed. Trackspacer, ProC2, and Logic's compressor are in basically zero latency plugins and ProMB's delay is going to be ~15ms. In fact, I think any workarounds I posted earlier only serve to reduce the delay to the individual plugin receiving the sidechain. If you do the same test with soothe and listen to its sidechain cue it's impossible to get the delay below 42ms, which is a 2048 sample delay at 48khz, and the exact latency of soothe.

It is fixed (at least on my system). Macmini M2 pro, Ventura 13.6, Logic 10.7.9. 

 

you can easily reproduce it. Import an audio drumloop on track 1. copy the loop on track 2 and set the output of track 2 to an aux. when you press play, they are perfectly in synch as expected. 
now put pro C2 on that aux. enable the sidechain audition and „ext" in pro C2 and set the sidechain input to track 1. you are now listening to the sidechain signal on an aux as well as to an audio track (track 1). They are perfectly in synch if you press play. 
 

now comes the news. If I now insert latency-inducing plugins on the aux channel before ProC2 (e.g. logics linear phase eq) on my old logic version (10.5), the two signals aren’t in synch anymore as logics PDC doesn’t compensate the sidechain signal. 

on my new logic version (10.7.9) both signals remain in synch. I didn’t test all possible scenarios, but this was the only scenario on my old system where the PDC failed, and this now works in 10.7.9.

 

 

Edited October 19, 2023 by Flatliners

[des99]

There were improvements in sidechain handling between 10.5 and 10.7, so some of the previous behaviours were fixed indeed. However, there are still outstanding issues with this, it's not been completely fixed yet.

[Flatliners]

3 minutes ago, des99 said:

There were improvements in sidechain handling between 10.5 and 10.7, so some of the previous behaviours were fixed indeed. However, there are still outstanding issues with this, it's not been completely fixed yet.

Ah Ok. Which scenario / issue isn’t fixed yet? Maybe I stumble across something in the future. 

[sunbrother]

24 minutes ago, Flatliners said:

Ah Ok. Which scenario / issue isn’t fixed yet? Maybe I stumble across something in the future. 

Running a sidechain to a high-latency plugin (like soothe, or many vocoders) is the biggest one. You can hear it most clearly by adding a sidechain while playing in cycle mode because it takes a moment for Logic’s incorrect compensation to kick in.
 

Things like Trackspacer and Pro-C are basically zero latency, but for anything else this AU developer put it best:

”Logic does not typically compensate the latency for tracks feeding a side-chain on a latency-inducing plugin. We’re not aware of a reliable work-around, unfortunately, except of course to pre-delay the audio going into the side-chain, using a copy of the source track that is not routed to the mix if needed.

The Logic team is aware of this issue, but it would probably help them prioritize it if you provided feedback to them via the “Logic” menu.”

 

Edited October 19, 2023 by sunbrother

[Flatliners]

16 hours ago, sunbrother said:

Running a sidechain to a high-latency plugin (like soothe, or many vocoders) is the biggest one. You can hear it most clearly by adding a sidechain while playing in cycle mode because it takes a moment for Logic’s incorrect compensation to kick in.
 

Things like Trackspacer and Pro-C are basically zero latency, but for anything else this AU developer put it best:

”Logic does not typically compensate the latency for tracks feeding a side-chain on a latency-inducing plugin. We’re not aware of a reliable work-around, unfortunately, except of course to pre-delay the audio going into the side-chain, using a copy of the source track that is not routed to the mix if needed.

The Logic team is aware of this issue, but it would probably help them prioritize it if you provided feedback to them via the “Logic” menu.”

Yes you’re right, just checked it with soothe and proC2 with lookahead enabled. There is the latency again. Doesn’t even matter if soothe is on an aux or on an audio track..

[Daniel Harvey]

On 11/12/2018 at 6:20 PM, StevenTOAST said:

I have a SORT OF fix that goes like this (most of the time I use the kick for ducking via sidechain, but it works for any other sound sources, I usually make a group of 16 busses just for these chains):

1. Send the Kick (Inst 1) to a bus, let's say Aux 1

2. Aux 1 settings: output=none, add a Sample Delay

3. Send Aux 1 to Aux 2

4. Aux 2 settings: output=none

5. insert ducking compressor onto desired track, Side Chain to Aux 2

6. on compressor, switch the sidechain to "listen" and use the Sample Delay on Aux 1 to sync both clicks (original source kick and the one that passes through the sidechain path)

7. repeat later in the project, the latency will be different as we add plug-ins

8. don't forget to check the sync before rendering the final mix

9. that's it

10. tip: you can adjust the delay to feed the sidechain kick with few ms earlier, so the sound "pre"ducks - to be honest even if Apple will correct this, I will use the same method, I find this tip pretty awesome in getting a nice, bouncy, ducking

 

This is THE WAY. Great tip.

+Will add that I've observed the sidechain trigger latency sample delay value is equal to the sum of all sample latency induced by plugins in the entire signal path up-to the sidechain compressor (but not after). So, you can add up the number of samples each plugin in the chain leading up to the sidechain compression (or LFO-tool, Volumeshaper, duck...), then program the sample delay to have that value, and it will line right up audibly. The visual feedback of meters is still off, though, but at least it sounds accurate.

I find when mixing with large sessions with lots of plugins & multiple sidechaining elements I will need several sample delays in a row to provide enough delay on the ghost for the longest threads of processing...and I will need several ghost kicks to trigger sidechaining of various elements along the way, each ghost with a different amount of sample delay (one with zero, one with ~5k, one with ~10k...honestly up to ~30k for the final one hitting the end of the entire mix routing, it adds up...).

 

[sunbrother]

Wow. Good to know that option is there. My mix-critical sidechains mostly happen on my buses so this wouldn’t be crazy to work into my template. It seems a little error prone for every little bit of ducking I might do, so I have a feeling I’ll continue doing a lot of that manually anyway.

It would be great if they just fixed this already…