Topic: Order of applied effects in the remastering process

[fandelive]

Hello,

I've been trying to remaster my own live concert tapes within the last couple of years. So far I've gently experienced with a bit of EQ, Compressor, Hard-Limiter and De-clicker (hand clappers; can't blame them), Normalizer, ...
I also surgically applied a volume lowering effect (can't remember the name of that effect, please help!) on the woooooooo-ers to limit (wait, is that effect called a limiter?) their negative impact on the music.

I know the quality of the final product also depends on the order in which you apply each of those effects within the post-processing chain... and that's where I'm a bit lost.
So, how would you proceed? What effects do you like to use and in what order in the chain?

Thanks!
-fandelive

[opsopcopolis]

There’s sort of an accepted order of operations, but not necessarily a right or wrong way to do it. My personal preference is EQ first, then dynamic processing (compression/limiting), then normalize if you want to. You can think about it in terms of how each process effects the next one; you don’t want to be compressing based on frequency content that you’ll be removing (hence dynamics after EQ) and you don’t want the normalize algorithm to process based on peaks that will be removed by comp/limiting/de-click (hence normalize at the end.)

All that said, if you’re using a limiter correctly, you shouldn’t need to normalize after. Set the out ceiling to something like -.1db and push the input gain until you’re limiting the stuff you’re hoping to limit.

[nulldogmas]

There’s sort of an accepted order of operations, but not necessarily a right or wrong way to do it. My personal preference is EQ first, then dynamic processing (compression/limiting), then normalize if you want to. You can think about it in terms of how each process effects the next one; you don’t want to be compressing based on frequency content that you’ll be removing (hence dynamics after EQ) and you don’t want the normalize algorithm to process based on peaks that will be removed by comp/limiting/de-click (hence normalize at the end.)

Yep, same here. De-clicking can take place pretty much anywhere in the process, but I prefer to do it first only so that if I end up re-EQing, I don't have to re-de-click, since that part is the most time-consuming.

And I always do EQ while leaving some headroom, so that I don't have to worry about some frequency boost leading to clipping, then normalize after. I usually apply extremely light dynamic compression for just the loudest stray peaks before normalizing — just eyeball the range that 99% of the music is within, then do like a 4:1 ratio for everything over that.

[vibrioidxire]

Always EQ before any type of limiting/compression. If you EQ a frequency that was limited in a compressor, the audio can sound super trapped in a way.

[nassau73]

Personally, I prefer to make a copy of my master recording first.

Then, I go through and track each song as I go - at the same time, working on processing the screams and clapping as I go. In the end, I have a "tracked and cleaned" copy that I can now try various EQ or whatever post processing I want to try.

Thing is - when I come back a day or two later, if I don't like the way my post processing effort sounds, I just go back to the "tracked and cleaned" version to try again.

For me, this saves alot of time since now I have a base copy to work with that's pretty much the original master track with the basic cleanup already taken care of. In the past, I'd get all the processing done, then track and clean and decide I didn't like the result and then go back and do everything all over again.

At this point once I believe I’ve repaired any perceived shortcomings in RX, I’ll switch over to Ozone to try various post processing.

[Gutbucket]

All good advice.

One meta-level up from this, a general principle I find useful similar to what nassau73 mentions, is to sort of conceptually split post-processing into a few separate categories, in this case the corrective stuff done first, and the polishing / mastering stuff second.  The corrective stuff consisting of all the baseline things like appending, trimming, and aligning files, balancing pairs, fixing drop outs, intermittencies, noise problems, attenuating specific unwanted noises, general EQ corrections for each pair or element.. all the stuff that would come before the actual mixing of sources (which is the second category for me editing multichannel stuff, and will apply to SBD matrices, but won't apply to straight 2-channel AUD recordings). Then the polishing, mastering, tracking, labeling stuff performed on the resulting mix as a separate category.   

Same as what the other folks are saying, just a way of ordering the overall processing in terms of logical sub categories.  This also makes for good, obvious places to save copies, so that one can return later an basically know what has been done, and move on with what still need to be done, or redone.

First consider where each thing you need/want to do best fits in the overall categorical divisions, then within each category determine what the best order is in which to proceed.

[voltronic]

Lots of great advice here. I generally follow Gutbucket's "category" approach with periodic exports. I do almost all of my processing in RX except for EQ (if needed), which I do later in Reaper. The reason I can get away with that is because the only kind of limiting I do in RX is Declick / Decrackle to knock back periodic or steady applause to levels below the peak in the music. I can do this in a way that the music under the applause is unaffected.

Later on in Reaper, I will have Ozone Maximizer on the master bus at minimal settings to raise the overall level a few more dB while reducing a few of the highest peaks. Note that I wouldn't recommend this to everyone, but the music I record has wide dynamic range to the point I need it narrowed a bit for the finished product.

[Scooter123]

So I apply JEMS protocol in RX, De-Click (for clapping), Azmuth, and De-Phase right out of the gate, then deal with specific issues. 

[Chilly Brioschi]

Save the master, and work a copy.  - ALWAYS !

Log your changes on a change sheet. (be detailed, you'll thank me later)

Great advice here on clean-up, and remedies

Overall balance can benefit from banded compression, but save that trick for a few projects forward
The theory is about resonances and formants in the room and "containing" them
This shouldn't be a forward priority.

Normalize is the final step to media limitations, I think that was covered
Not as much to save the groove or not top the CD, but every D/A has an U.L.

Ask back here if any specific questions, I see a few who have much to share in the thread.
I know very little, but have a resource or two who are retired and quite generous with information.

To the thread....
Peaks at 3db, 6db, or 12db ?    (Minus implied)

[opsopcopolis]

Save the master, and work a copy.  - ALWAYS !

Log your changes on a change sheet. (be detailed, you'll thank me later)

Neither of these are necessary if you’re don’t use a program that does destructive editing (but I guess people here love audacity, so…)

I generally set my limiter out ceiling to -1

[morst]

Neither of these are necessary if you’re don’t use a program that does destructive editing (but I guess people here love audacity, so…)

Audacity old versions allow you to work directly on the original tracks, but it does not destructively modify the track itself.
Perhaps you're thinking of Sony Sound Forge?
Audacity new version doesn't even allow you to work directly on the original.

[live2496]

Limiting should be the last in the order of operations. Other things don't matter so much but every step affects then next thing downstream.

I like to do corrective things first like de-essing.

Gordon

[Chilly Brioschi]

I think some aren't aware that an amplitude signal limiter is just an aggressive compressor.
It "smacks down" the peaks.
The expression "Comp-limiter" isn't used as much as it used to be

Play with your limiter on very dynamic signals which exceed the set limit and give it a good listen.


Watching the power consumption of a Power Amplifier and comparing strongly limited signals against the unprocessed dry ones is surprising.


As an aside, the ability to roll back changes might not save a dumped file during an edit, and most audio editors don't record and log the settings during edit stages.

[boomfizzle]

So I apply JEMS protocol in RX, De-Click (for clapping), Azmuth, and De-Phase right out of the gate, then deal with specific issues.

couple of questions:
Are you using adaptive azimuth or adaptive phase at all?  I use adaptive level inside the RX7 Azimuth module with confidence to resolve channel gain imbalances, but am always concerned about those other two options introducing artifacts to musical recordings, as Izotope says can happen in its help file.

If I wanted to try the de-reverb module, would that be best before or after phase/azimuth?   Does Izotope 'prefer' the phase/azimuth issues are sorted before trying to de-reverb, or would de-reverb best be completed before any manipulation of the file?

TIA!

[detroit lightning]

Recently upgraded to Izotope RX 10, and finally trying to figure it out.

In general, I think I'll use it to:
-reduce hum
-deal with quiet music / loud clapping issues
-wind
-breathing on the mics

I'm sure there are a million other things I can do with it, but trying to keep it simple to start.

For things like de-click, or using the repair assistant, a few questions:
-does it matter what order you use them in?
-do you apply to only isolated areas of the waveform (around the loud clapping, for instance) or just broadly to the entire file?

Any other useful/easy ideas to get started with?