Peak vs. RMS detecting Comp/Lims -- which to use when?

[Seamonster]

In May/June 2003 issue?s ?Behind the Gear? interview of Geoff Daking, he states: ?[T]he great compressors all have peak detectors...the shitty compressors have RMS detectors.? As examples he lists the 1176 and LA2A as peak detectors, dbx 165 as RMS (he cites a few others too). In making that statement, he implicitly lumps compressors and limiters into the same category (technically correct), though my sense is that whether a device does peak or RMS detection would be relevant to the specific task at hand. Clearly, it?s simplistic to say ?peak = great, RMS = shitty.? When and for what, we should ask.

I?m curious what people think of this distinction, and how you apply your thinking in practice. E.g., are you more likely to use an RMS detecting device on a single track and a peak detector in a final-limiting context? Or vice versa? I?m not looking for an answer to that specific question, but rather *any* thoughts you might have about the Peak/RMS distinction when choosing a device (outboard or plug-in) for a specific task. I?m hoping to deepen my thinking about how my ?go to? choices vary as between individual tracks, mixes, and mastering applications. Any rules of thumb out there, or does this have to be just another of those ?use your ears, if it sounds right, it is right? things?

Hoagie Hill
Seamonster Sounds

[theBaldfather]

I'll definitely be watching this thread. I've been tempted to post something along these lines myself.

[Babaluma]

what's the difference between the detection of peak and rms? i mean i know the difference in theory between what peak and rms do, but what's actually the difference in the way the circuit detects it? does the rms detector just look at more of the waveform at any one time? how does it work? i am interested to know about this for my modular compressor patch (see other thread).

[@?,*???&?]

I've found PPM to be useful when checking relative levels of individual songs across a whole project, but when I don't have all the songs in front of me, but rather just individual sessions. PPM gets into what I would deem 'apparent loudness' so it becomes helpful when trying to get all your mixes to the same level. Mastering engineers rely on this at times too.

[Colin F.]

You can really say which one is better, IMO. A peak simply reacts to the loudest spikes in the transient, while an RMS is looking at an average (root mean square). An RMS compressor can be excellent where you need smooth, transparent compression such as on vocals or bass guitar. Peak detection works great on drums.

Of course, its really down to the design and character of the compressor, so these rules quickly fly out the window...

[I'm Painting Again]

In May/June 2003 issue?s ?Behind the Gear? interview of Geoff Daking, he states: ?[T]he great compressors all have peak detectors...the shitty compressors have RMS detectors.? As examples he lists the 1176 and LA2A as peak detectors, dbx 165 as RMS (he cites a few others too). In making that statement, he implicitly lumps compressors and limiters into the same category (technically correct), though my sense is that whether a device does peak or RMS detection would be relevant to the specific task at hand. Clearly, it?s simplistic to say ?peak = great, RMS = shitty.? When and for what, we should ask.

I?m curious what people think of this distinction, and how you apply your thinking in practice. E.g., are you more likely to use an RMS detecting device on a single track and a peak detector in a final-limiting context? Or vice versa? I?m not looking for an answer to that specific question, but rather *any* thoughts you might have about the Peak/RMS distinction when choosing a device (outboard or plug-in) for a specific task. I?m hoping to deepen my thinking about how my ?go to? choices vary as between individual tracks, mixes, and mastering applications. Any rules of thumb out there, or does this have to be just another of those ?use your ears, if it sounds right, it is right? things?

Hoagie Hill
Seamonster Sounds

I think like the vast majority of folks we use what we can afford to buy for our studios or use what other studios have hanging around!

How many times in the history of recording do you think someone was like "should I use a peak sensing limiter on this bagpipe track or an RMS sensing one"?

We use what we think sounds best or what we think serves the song or artistic goal the best at that moment..

or what is quickest to set-up!

heh..

Daking is obviously being a right twat in that interview spewing that sort of rubbish..

[cgarges]

I'm calling bullshit on that one. Since when is a 165 a shitty compressor? What about the Urei LA-22? Is it only a shitty compressor when the detector knob is turned to the RMS position. When it's halfway between peak and RMS is it only a half-way shitty compressor? what about the compressor in the Alesis Masterlink? Shitty in RMS mode, not shitty in peak mode?

I totally don't think about whether a compressor is operating in peak mode or RMS mode when I'm using it, unless I need to make an adjustment of a parameter that is already there. I rarely think, "Man, this GML would be so much better without the RMS detector is it." I am much more inclined to think about the sound of the particular unit and what sort of compression it does than whether or not it actually has an optical unit or a FET or whatever.

And then of course, whether or not it's doing what I want it to.

I understand SO MUCH LESS than Geoff Daking when it comes to compressor design, but I'm just not buying that as an absolute.

Chris Garges
Charlotte, NC

[fossiltooth]

I'm calling bullshit on that one.

Agreed.

People will say all sorts of things in order to sell a product.

[@?,*???&?]

I'm calling bullshit on that one. Since when is a 165 a shitty compressor?

You're right, the actual shitty compressor is the dbx 162- do those things ever work right without screwing up the stereo image?!?!?!

[The Real MC]

I agree, I call BS on that article.

While the detector circuit is the acid test of a good compressor, there are good ones with both RMS and peak detectors. The UREI/JBL 7110 and LA-10/12/22 have excellent detectors, they sound way more natural than my Alesis 3630 did and those also had peak/RMS detectors. Once I heard a 7110, my 3630s were destined for the auction block.

[Brett Siler]

Thats like saying, solid state is for shitty amps, tubes are for good amps. I don't most people would consider the Roland Jazz Chorus a shitty amp it just reacts differently that say a Fender Twin.

Good and bad are arbitrary statements.

[Seamonster]

Okay, so we have some consensus that Mr. D?s comment about peak detectors being better than RMS detectors was offbase, for whatever reasons (I too suspect that he was basically pushing his own peak-detecting compressor). Let?s move on.

How ?bout more on their pros/cons in practice? Of course you work with the tools you have and make it sound as good as you can. Personally, I also like to have some analytical basis for the choices I make; starting points or fall-back points, if you will. Does no one ever think to themselves, ?ah, for this task, an RMS-detector would be just the ticket?? Or whatever. No one?

I have a hunch that if I had a better understanding of this stuff I could better address one of the issues I often find myself asking myself: In the situation at hand, do I want to use a shallow threshold with a greater ratio or a deeper threshold with a more subtle ratio? I go with my intuition, based on experience, but sometimes I wish I could be more analytical about it.

So here I?ll toss out some hypothetical blather for y?all to dissect. On one hand, peak detection causes the comp/lim to work on the ?tops? of waveforms (depending on how deep you go with the threshold). That should leave the lower-level body of the sound relatively untouched, allowing for a clean sound overall. (Unless you end up with more intermodulation distortion right around the threshhold level!) On the other hand, an RMS detector, looking at the density of the waveform deeper down from full-scale, is going to push down more of the wave-top, more equally from above the threshold, as compared to the same ratio/threshhold settings on the peak-detecting comp/lim. Instinctually, I sort of equate peak-detecting with a harder knee, RMS with a softer knee -- but I know there?s more to it (or different) than that, I?m just not sure what exactly.

All else being equal, which detector yields the punchiest mix? Or in the mastering stage, which one does best at sanding down the surface without cutting too far into the grain? The way I just worded this, it seems like the peak-detector would win in both cases. But in my experience, either type of detector can get the right results; it depends more on finding the right settings for the particular device...yada yada, back to the right brain....

More practically: Out of the gear I have (outboard or plug-ins), Izotope Ozone has an an option for either type of detection. So I?m thinking that sometime in the next few weeks I might set up an A/B experiment. I?m not sure what source material I?d use; it might be fun to post a link to a few audio files, but all the stuff I?m working on now is for clients and I?m not authorized to be sharing it with the aether. Maybe if someone had a simple groove that they could send to me in WAV or AIFF format (via yousendit.com), I could run it through Ozone a few times and then we could all listen to the difference. Any takers? C?mon, if you?ve read this far, you must have the guinea pig file sitting on your hard drive right now.

Hoagie

[wedge]

guinea pig file

[Seamonster]

Do you prefer to use peak or RMS detection when compressing guinea pigs? But seriously....

[cgarges]

OkDoes no one ever think to themselves, ?ah, for this task, an RMS-detector would be just the ticket?? Or whatever. No one?

II totally don't think about whether a compressor is operating in peak mode or RMS mode when I'm using it, unless I need to make an adjustment of a parameter that is already there. I rarely think, "Man, this GML would be so much better without the RMS detector is it." I am much more inclined to think about the sound of the particular unit and what sort of compression it does than whether or not it actually has an optical unit or a FET or whatever.

And then of course, whether or not it's doing what I want it to.

CG