How does one apply the concepts of peak and RMS?

[joelpatterson]

It would seem-- someone correct me if I'm wrong-- that in today's world, there is a distinct variety of playback situations. A mix that sounds breath-takingly deep and gorgeous coming through a spiffy hi-fi set-up will sound meek and wishy-washy on a cellphone.

The cellphone audience is going to need kind of a "standardized" treatment to all the trebley material: a certain degree of constancy and insistence. This is not a "good or great" development to the progress of Man, but fine, find another planet. Also, the lowest level better not be that low.

The best of all possible scenarios would be if you could label eight distinct mixes, idealized to every playback situation, but that's dreamy. So. Short and sweet: your limiting should be chopping off nice, bite-sized chunks (7dB? 10 or 11?) in a completely invisible and non-noticeable way, on the loudest hits and clangiest bursts and peals. Your RMS should hover in the -11 to -7 zone, higher than that and you're in blaring territory, and blaring is not good, no matter what they tell you.

But these numbers only tell a small fraction of the story, the true challenge is to situate yourself in your monitoring environment and keep adjusting levels and things until you are overwhelmingly ya-ya'd AND these numbers are somewhat in these ranges.

[edva]

There are no rules, and it all comes down to ears, but in general, you want your mix peaking at 0 (or a small fraction lower), and your RMS averaging at an appropriate level for the genre, for example, -17 for "easy" up to - 9 or so for "hard".
There is a third measurement, crest factor, which is a function of the ratio between the two. The greater the difference, the more dynamics in the mix; the smaller, the less dynamic range. Most "modern" music has high RMS and low crest factor.
IMHO, and YMMV.

[AwwDeOhh]

If you're talking about "on the way in", most digital converters are calibrated somewhere around -18dB RMS

RMS is basically a way to say 'average over time'.
Peak is talking about the tippy-top of a transient.. or the highest level triggered on your meter

Typically you want to record at a RMS level around where your converters are calibrated to, to avoid adding unnecessary digital distortion and get the best signal to noise ratio.

-24 to -18 (ish) on a digital system is equivilant to 0dbVU on an analog board, i don't know about you, but i never heard of anyone running at the +24dbVU on their boards... so it makes sense not to do it digitally.

If you're peaking at or near -0, you get digital clipping (which is not desirable in most cases)

The somehow popular phrase "record hot" isn't accurate, and even less so since modern 24bit converters have so much headroom, that it behooves one to take advantage of that room.

ONCE the sound is 'captured', you can raise or lower the level to your hearts' content, (keeping the levels under digital -0)

But if you're sending out your music for mastering, i'd advise to leave headroom in the mix, as the mastering engineer has not only the tools but the listening environment to get the most out of the squeeze. If you're mixing at the top of the range, you're not leaving the M.E. much to work with.

[Trakworx]

Originally Posted by drpeacock

I only have an extremely rudimentary understanding of average RMS vs peak and how it applies to a mix. My understanding frankly boils down to, "don't compress/limit things so much" and your record will sound better.

Can anyone give me a short edumacation on correctly applying these concepts?

Thanks.

The RMS values stated in posts 2 and 3 apply to finished masters IMO.

I think you are asking about a mix, not a master or individual recorded track, right? If so, then I'd say a mix, before mastering, should peak below zero and the RMS will vary widely depending on the arrangement of the song (or section of the song), the genre, etc.

If you compress a lot in your mix it will likely have a lower crest factor, which is the peak to average ratio (peak/RMS).

Too low a crest factor and you can lose punch and/or openness, which will decrease even further once a mastering limiter is applied. So it's often a good idea to leave enough crest factor in your mix so that when the ME reduces the dynamic range for loudness there will still be some perceived dynamics and punch in the final outcome.

Too high a crest factor and you can lose "glue" and some elements in the mix may be inaudible at times.

So there's a trade-off and a balancing act, which is part of the essence of mixing really.

Just scratching the surface here...

[theblue1]

Aeverage levels of -7 dB RMS are, to my thinking -- hideously crushed.

15 or 20 years ago, a 'reasonable' RMS average for the intense parts of a song was considered to be in the -18 to -16 dB RMS range.

Obviously, fashions of approach have changed, record labels, convinced that having the most crushed tracks around means a competitive advantage since, in their minds, consumers will simply prefer what is loudest, send out the word that good sounding masters are "not loud enough"... etc.

But, nonetheless, even by today's terms, anything hotter than -10 RMS can be considered squashed, seems to me.


Trackworx has an important point. If you are sending your work to be mastered or doing it yourself in a separate process, you don't want to go anywhere near crazy in compressing your mix. Create the best mix you can without worrying about "competitive loudness" and let the mastering engineer work his "magic."

(Admittedly, a lot of these so called mastering engineers are complete tin-ears who wouldn't know a decent track if it was shoved up their alimentary canal and the proof is in some of the utterly unlistenable mastering one hears applied to both new works and -- someplace where their incompetence and malfeasance can be readily observed -- in 'remasters' of classic releases. But that rant's for some other thread.)

[drpeacock]

Thanks everyone! And yes, I'm talking abut the mix, not the master.

[joelpatterson]

I thought I was talking about the mix too. I could be wrong...

I want the ceiling (the limit beyond which no sounds goes) at around -.5 dB's. This is definitely a peculiar and quirky approach for which I accept full responsibility-- but it always seems like if you shy away FROM TOTAL MAXED-OUT ABSOLUTE ZERO it's just easier on the ears, all around. So this is a scenario where the peaks are all -.5, but when you hear the kick hit, you'll see your limiter advise you that 7 or 10 or 11 dBs of potential noise have been swallowed up/avoided/however you want to put it.

This is the whole purpose of a limiter, I think-- in some magical way known only to itself, it shaves off excessive blasts of sound, but it still leaves you with the impression that a kick has sounded, and sounded fully and nicely.

This isn't the kind of thing to leave to the mastering engineer, this is something you need to do here and now. So as you construct your mix, the levels should be sort of more-and-less bouncing up against the -.5, and dropping down low enough you leave you with a total overall average RMS of ballpark -10, give or take. You're not bouncing up against -.5 just barely-- you're "bursting through" it each time. [Every other time? Every third time?] The critical thing is that you mustn't ever notice any "hollowness" or "thunkiness" or anything odd at all-- these overs should be like someone said, invisible and un-noticeable.

Someone also made another good point: a slavish attention to these readings too keenly will be injurious to the sonic health of your creation. Always and forever, it must sound amazing first and foremost.

[Trakworx]

Originally Posted by joelpatterson

I thought I was talking about the mix too. I could be wrong...

Levels in the -11 to -7dB RMS range are definitely in the "mastered for loudness" zone.

Originally Posted by joelpatterson

I want the ceiling (the limit beyond which no sounds goes) at around -.5 dB's. This is definitely a peculiar and quirky approach for which I accept full responsibility-- but it always seems like if you shy away FROM TOTAL MAXED-OUT ABSOLUTE ZERO it's just easier on the ears, all around.

That's not at all peculiar to me. Leaving .5dB peak headroom after final limiting is my standard procedure when mastering. Many MEs leave between .3 and .5dB. It prevents intersample peaks from clipping during playback, SRC, and when files are converted to MP3, AAC, etc.

Originally Posted by joelpatterson

So this is a scenario where the peaks are all -.5, but when you hear the kick hit, you'll see your limiter advise you that 7 or 10 or 11 dBs of potential noise have been swallowed up/avoided/however you want to put it.

FWIW, 7 to 11dB of gain reduction from a limiter on a mix is considered extreme by pretty much all mastering engineers.

Originally Posted by joelpatterson

This is the whole purpose of a limiter, I think-- in some magical way known only to itself, it shaves off excessive blasts of sound, but it still leaves you with the impression that a kick has sounded, and sounded fully and nicely.

Yes, at moderate settings a good limiter does this. At extreme settings "fully and nicely" goes out the window.

Originally Posted by joelpatterson

This isn't the kind of thing to leave to the mastering engineer, this is something you need to do here and now.

Actually this is exactly the kind of thing to leave to the ME. If you're planning on doing any processing to the stereo track after mixdown, then it's better to leave the limiter off. Every ME will tell you that. EQing and compressing already-limited material is a recipe for distortion, harshness and lack of punch. The brickwall limiter should be the last processor after everything else but dither and SRC to the final delivery format.

Originally Posted by joelpatterson

So as you construct your mix, the levels should be sort of more-and-less bouncing up against the -.5, and dropping down low enough you leave you with a total overall average RMS of ballpark -10, give or take. You're not bouncing up against -.5 just barely-- you're "bursting through" it each time. [Every other time? Every third time?] The critical thing is that you mustn't ever notice any "hollowness" or "thunkiness" or anything odd at all-- these overs should be like someone said, invisible and un-noticeable.

It can be helpful to put a brickwall limiter on your 2 bus while mixing in order to hear an approximation of what might happen to the dynamics in mastering, but it's best to remove it before printing the mix, and make sure there is no clipping. If you're not going to have it mastered then do whatever you want, but if you're sending to a ME then please leave the limiting to him/her so it can be applied after all other processing is done. Most MEs, myself included, will ask for a remix if you send them a limited mix.

Best,

.

[rocksure]

Originally Posted by joelpatterson

This is the whole purpose of a limiter, I think-- in some magical way known only to itself, it shaves off excessive blasts of sound, but it still leaves you with the impression that a kick has sounded, and sounded fully and nicely.

This isn't the kind of thing to leave to the mastering engineer, this is something you need to do here and now. So as you construct your mix, the levels should be sort of more-and-less bouncing up against the -.5, and dropping down low enough you leave you with a total overall average RMS of ballpark -10, give or take.

So you are a limiter on the 2-buss kind of guy then it would seem.
I ain't. I never ever put a limiter across my stereo mix ( unless it's a TV show mix). I do allow the odd very quick transient peak to get up to -O.5 or so. I do have a light bit of hardware compression going on the mix buss ( 1db maybe), but no limiting. I just don't like the sound of a limiter on there, and can't honestly see the point in having one there.
All the severe squashing is best left to mastering in my mind anyway.

[drpeacock]

Thanks everyone, very instructive so far.

[Zep Dude]

Joel,

Good info. But send me a mix and I'll show you why it's better to leave the limiting to the mastering engineer. Glad to do it gratis just for the sake of the discussion.

[joelpatterson]

PM enthusiastically sent.

[theblue1]

Trackworx is on the money with regard to the wisdom of leaving some headroom in order (as per Joel) to avoid intersample peaks that could cause some lesser playback equipment to distort their output circuits.

Unfortunately, with some signals, leaving .5 dB of headroom may not be enough to prevent intersample peaks from going above 0 dB FS on reconstruction, so it may be best to use an intersample peak aware meter.

Happily, the nice folks at SSL make a free plug in meter for just that use. (It does takes some resources so you may want to hold off inserting it in the master buss until the mastering phase.)

[Trakworx]

Originally Posted by theblue1

Trackworx is on the money with regard to the wisdom of leaving some headroom in order (as per Joel) to avoid intersample peaks that could cause some lesser playback equipment to distort their output circuits.

Unfortunately, with some signals, leaving .5 dB of headroom may not be enough to prevent intersample peaks from going above 0 dB FS on reconstruction, so it may be best to use an intersample peak aware meter.

Happily, the nice folks at SSL make a free plug in meter for just that use. (It does takes some resources so you may want to hold off inserting it in the master buss until the mastering phase.)

Yes, and some limiters have this feature built in, like Fabfilter Pro-L.