Sidechain multiband limiting, MSD, what's the best way to do this in the MeldaVerese?

[Annabanna]

There is a core task in the mix/master realm that I always assumed some magic plugin was about to solve. For years, I've tried everything that came along like Izotope and others with a system of backchannel communication and sidechaining. Even the much more complex kludges involving dummy busses never seem to get it right.

The following is just one example of a task that requires this type of solution, but there are others.

TASK:
Instrumental track is mixed and mastered, and you need to add vocal.

CONSIDERATIONS:
1) The instrumental track is fully mastered, loudness leveled, etc, so the peak level cannot change at all once the vocal is added. Essentially, a fully mastered instrumental and separately mastered vocal go in, and a fully mastered combined track comes out having the same peak levels as the instrumental.

2) The vocal may or may not be present in any given section, so the instrumental track should not be turned down to create headroom or otherwise distorted when the vocal is not present.

3) The vocal should be easy to understand, and while compromises will need to be made, the groove should stay as up-front as possible. Anything along the lines of a TrackSpacer style frequency specific ducking is probably a good thing if it helps to reduce frequency masking and allow the vocal to be heard more clearly... so long as it does NOT take the life out of the underlying groove. In other words, brute force traditional broadcast style ducking is bad since the music keeps ducking in and out audibly.

I've tried this extensively with various setups involving MSD to no avail. Like most solutions I've tried, I can't get the combined peaks part to work properly. I've tried to use various incarnations as well of multiband limiters with sidechaining, etc.

Ultimately, I always end up settling for some solution that only kind of works followed by the combined track running through a broadband limiter for the final peak leveling. I'm not entirely convinced this CAN be done properly with any of the tools I've seen so far, but would like to see if there are any potential solutions in the MeldaVerse I may have overlooked.

[dmbaer]

You might want to take a look at MSpectralDynamics. I just scanned what you wrote rather quickly, but it definitely sounds like a job for that processor. You can find a very in-depth write-up on it here:

https://soundbytesmag.net/spectraldynamicsdynamiceq/

This was written six or seven years ago, but it all still applies, I believe.

[sirmonkey]

When I started reading your post, Spectral Dynamics was the first thing that came to mind.
It's like a bazillion band compressor, and can attenuate the mix, but only in the bands that
coincide with the frequencies in the singers voice.
If you have it in your toolbox, definitely give it a go. You'll have to turn the mix down at least a little.
Put an instance on the mix, and have it sidechained to be triggered by the singers track.

If you don't have it, now would be a good time to utilize the 15 day free demo.

Good luck

[sirmonkey]

[adecadence]

I see that you've tried MSD and the end result doesn't match your goal
"fully mastered combined track comes out having the same peak levels as the instrumental"

Currently I don't think any single offering Melda or not, is able to do it. I doubt their design goals match your need, it's usually much broader and don't adhere to side-chained peak level.
I guess you could MXXX it out (MSD, MLimiter, etc.), and use that as a template for future work?

[Annabanna]

I see that you've tried MSD and the end result doesn't match your goal
"fully mastered combined track comes out having the same peak levels as the instrumental"

Currently I don't think any single offering Melda or not, is able to do it. I doubt their design goals match your need, it's usually much broader and don't adhere to side-chained peak level.
I guess you could MXXX it out (MSD, MLimiter, etc.), and use that as a template for future work?

Yeah, I've spent several entire weekends over the past few years trying to coax MSD to do this by itself or in combination with other tools.

MSD is very good at certain things, but this sort of ceiling critical combined limiting isn't one of them. I'd put Trackspacer in that category too. Both have good tricks up their sleeve for spectral ducking in general, but neither tends to perform well when combined peaks are critical, and using either for anything along these lines at the very least requires follow up with a separate limiter for the combined signal.

I've also done some blind listening tests between MSD and other setups and had better results using NewFangled Elevate in it's max bands (32?) mode in a similar fashion since it seems to have a rather well developed algorithm for outputting a fixed ceiling across the calculation of many bands. Still, though, I end up needing to augment with other tools.

I need to do something along these lines regularly, so I'm always looking for anything that might get me closer, so thought I'd see if Melda power users have any suggestions since I'm fairly well familiar with the toolset now.

[sirmonkey]

Maybe this is bollocks, but...
Maybe sidechain the mix to MSD from the vocal, and also sidechain the vocal to MSD that's on the mix. Maybe set your threshold curves so that you don't gets tons of compression.

Another idea, also probably bullocks, but here goes-
Possibly in combination with the first part, use MCompressor:
Lower your entire mix so that the peaks are where you need them to be. In MCompressor,
use a custom curve. Draw a curve such that it never compresses, or just barely compresses here and there.
Draw the curve below the threshold so that it amplifies the signal.
* So instead of lowering peaks, you are raising stuff below them.
Worth a try?

[MeldaProduction]

Spoiler, that's part of the mixing revolution thing . Yeah yeah I know, "so make that happen finally"... Yeah, I'd like, but people always want features and features