Skip to main content

Ok so it's back to recording 101 here. I had read a previous thread on setting the rms to levels of -12 to -14 and concerning less with 0db.
Now my concerns are the definition and measuring means of rms. I am guilty of using 0 db as the "General" peak referencing point. I am not a fan of squashed sounds. Most of my work is acoustic strings and vocals. The central unit in my set up is Pro Tools 001 LE. Can I find some schooling on this topic here and/or reference material. Many thanks...
terry

Topic Tags

Comments

anonymous Thu, 10/30/2003 - 11:02

Meters are a complicated topic, I´ll give an example: In the pro-world you will see three different types, VU´s, Peak and Digital.
Vú = (Volume Units) corresponds to how your ear will hear loudness, it has very slow responcetime, I think it is in the region of 30 ms (correct me if I´m wrong) to rech its reading. This means that about 16.000 samples will pass before you see it on the meter.
Peakmeter is faster, about 10 ms risetime. First introduced in broadcast and mastering where overloading your transmitter och disc-cutter would cause problems. But still, for 10 ms almost 4.400 will pass before you see it.
Digital, or dBfs, reads (in theory) each sample and shows it on the meter, so it´s very fast.

If you line up a masteringsystem for: 0Vu = 0 Peak = -14 dBfs. A good mix will produce appr: +2Vu, +9-12 peak and -1 to 0 dBfs. The difference is called crest factor, difference between fast peaks and program. In digital, overload doesn´t exist, it´s a definition. If you have more than 3 continues samples of full reading, it will be seen as an overload, and a flag will be added to the word (Sony system, this could be altered in the machine by jumpers). If it was 0,1 dB or 20 dB doesn´t matter, they are all treated the same.

Many studios keep a lower setting on their digital multitrack: like -16 to -24 dBfs (as close miking will give more peaks) while in mastering a -12 to 18 dBfs will be used (as mixed program has a better balance).

So most important is: what type of meters are used and exactly what do they show. An older engineer could probably record to digital only using Vu´s, but the sad thing is...no single meter will give you the whole picture.

I belkive you can find som tech info at manufactors like RTW, NTP and DK-Audio.

Regards
Olle

anonymous Thu, 10/30/2003 - 15:34

Hi terryc,
I'm a garage DIY kinda guy and here's what I pay attention to in full mixes - this is the mastering thread, right ?

First I get a couple of tools together for my budget - free works great:
http://elementalaudio.com/products/inspector/index.html

After spending a little money I also watch Ozone2 spectrum and just got a behringer RTA DEQ2496 I'm learning.

With those tools I have VU, Peak, RMS, Overs, Averaging, 1/6 octave Spectrum, full resolution spectrum. Anyway enough tools.

About rms, technically it's some type of value derived after averaging a bunch of roots/squared for the duration of some time window. Ha Ha not too technical !

What I use rms for is a quick judgement concerning the relative density (and potential over-squishiness) of the program material. If I have a program of a 5 piece rock band and the rms is -15dB then it usually sounds pretty good to me assuming the dynamics of the program (rms to peak) includes headroom up to -0.3dB peak or something safe so there are no digital overs (technically 0dBfs but I usually go a little under).

If the same program is pushed into a mastering limiter then the rms increases, the rms-to-peak headroom decreases and unless you have a really good mastering limiter you might get a couple of overs which are a non-no. After a point either the program will sound too squished or distortion will occur. If I push a program like the one I mentioned up to -9dB then it usually sounds squished to me.

Most of my favorite rock band CDs sit around -16dB to -12dB rms. I've got a commercial Cher CD that I'm sure the rms is at -6dB or so and it sounds squished and unnatural.

Anyway that's what I use it for right now. Giving me a feel for the dynamics and life of the program. I usually watch it a lot if I'm pushing into a mastering limiter just so I can remember what value equates to a particular dynamic density.

Unless you're gonna write your own fir filter I'd just use an RTA like Inspector (VST software) or DEQ2496 (Digital hardware). I think if you're on Mac then Waves has a Paz RTA.

kylen

anonymous Thu, 10/30/2003 - 20:00

Terry,
I don't run ProTools but I run Wavelab and Sonar. Wavelab is my mastering software and it has a meter to reference the RMS levels, which I stand by with all my soul! Does PT have such a meter or an "analysis" you can run? I will always make a song reach an average RMS of -14 to -13. I never concern myself with 0db. We have discussed this before and I agree with some of the guys comments. It IS possible to get 0db, while retaining a RMS of -13/-14. But if you find this is not possible with your material, then don't sweat it! Just get a good, average level for the entire cd.
0db means nothing to me, a good, even level throughout the recording means everything!

My only advise is....Make everything even. Thats it. And please don't ruin a good recording by mastering it above -11db. That is just record rep crap, they know nothing about this, except..."make it the loudest cd ever!" "It will sell more copies"...........WRONG!!!!! it won't.

Mike

anonymous Fri, 10/31/2003 - 04:17

launchpad i do appreciate your advice very much and i too am a firm believer in letting the ears do the settings. what i have run into and thus the interest in this thread is relating to some of my mixing techniques. i do mostly acoustical and vocal and mix to get that live down front sound onto the cd. this is where some little problems have started to crop up relating to what i believe may be over use of limiters causing some possible sample clips. if i may side step here maybe you will have a better picture.
to give you an idea of what i do let me walk you through a vocal track for instance. first let me say that i track straight to protools and mix completely in protools. my 1642 mackie is used for monitoring/routing only. i record on two mics (at4060 & akg3000)to two mono channels. i duplicate these tracks. on the duped tracks i do a minimal pan l/r, lower the volume maybe 6 db. all four tracks are bused to an aux input which is panned at the maximum width of the four tracks. this aux input is duped and again i lower the volume about 6/10 db and treak the pan. these two aux inputs are bused to the vocal submaster.
my reasoning here is that the dupes with panning and reduced volume are giving the vocal a much fuller sound in the audio field. my concern is as follows.
the duping increases the volume level as it goes down the line, so i use L1/L2 as limiters only (no actual compression) to control the beast. often i have found myself with more limiters on the screen than i should probably have. this direction is working for me except that from time to time i suddenly pick up an anomally/distortion in the mix.
now am i out in left field? how would you gain larger than life sound here? is metering a wise addition?
protools has no rms metering. i hear that plugins do exist but are usually for mac and tdm systems. i too have wavelab that i purchased second hand with no manual but at this point i use it sparingly. guess i need to shorten that learning curve.

anonymous Fri, 10/31/2003 - 06:08

Can you describe the anomily/distortion ?

Is it any one track or bus getting overs or going into the red - does PT show you that ?

Or is it just a crunchy spot in the music that hit the limiter too hard - or is the limiter breathing too hard ? If you remove som limiter combinations does the distortion go away ?

Inquiring minds ! Ha Ha

I use Cool Edit Pro and Sonar3 along with enough peak meters to tell me when any bus goes over. I don't look to an rms meter for distortion usually - unless it says -6db ! I would look there for anomilies and artifacts but you can kind of do the same thing by turning off the dynamics processors and going thru your gain structure again track by track, bus by bus.

kylen

anonymous Wed, 11/05/2003 - 17:24

Guys - just beware of one thing:

Just because none of the samples exceed 0dBfs, doesn't mean there's no clipping. Fact is, it's almost a mathematical certainty that a mix that's peaking intermittently at 0dBfs will produce a waveform that, when reconstructed, will exceed Full Scale by upwards of 6dB (and therefore clip the consumer's deck)

Leave yourselves as much headroom as possible. I like seeing my RMS values down around -14, but only if my peaks are well under -1. I often set my final limter to an out gain of -1 or less, just to ensure minimal out clipping. (You'd have to turn it wayyyyy too far down to completely avoid clips)

anonymous Wed, 11/05/2003 - 20:07

Very well said!
I too run an avg RMS at -14, with peaks at -3(never higher). That gives plenty of headroom and the waveforms look perfect, not smashed.
I refuse to master a recording that sounds anything like the new "squash rock". It's all "sonic crap"!

If you want it loud....Turn it up on your stereo!!

Mike

Michael Fossenkemper Thu, 11/06/2003 - 08:04

It's great that you are able to do that. But I've found that unless your mastering jazz or classical, it's virtually impossible to please the client volume wise without smashing the peaks. In fact I just finished a rock record last week that averaged around -10 rms and the client called and asked if i could squeeze it up another couple of db. I spoke with him about it for awhile and realized that it was going to be me or someone else that does it. So I did it. I squeezed it up to -8rms. Does it sound as good, no, but the client is happy and sent me 2 more projects. Volume is a daily battle I fight as well as the other guys here. It is the single most requested thing I get. "I want it loud, and as good as it can sound at that volume". not the other way around. Radio is not the sole culprit, the CD changer and ipod I think are more reponsible. Radio somewhat corrects the volume differences where as the later two do not (with the expection of the ipod's worst sounding limiter ever built by man to try and even it out). Newer music is loud but it's for a reason too. It's a mobile world, people listen to music all over the place and usually not in a silent environment. Try walking down the streets in NYC with an Ipod listening to music pre 1990 and you get about half of it. Samething goes for the car with a window rolled down. Is it right? I don't know. so now i spend a lot of my time trying to please the clients while retaining some sense of musicality in my work. I do have some clients that prefer sonics to loudness and those are the CD's that go into my IPOD. So how do we correct it? buy what you like. It's not the mastering engineers fault, it's the consumers fault. Music is one of the truest forms of capitalism in the world.

anonymous Thu, 11/06/2003 - 13:54

So what we can agree on is.... most recording- and mastering-engineers know that more headroom makes records sound better.... and our clients dont understand and we have to please them toget a louder/more distorted product.

And who should we blame...the consumer...the record company...the producer... the market...
????????????

Maby we should blame the ourself.....

Olle

anonymous Sat, 11/08/2003 - 10:45

Originally posted by siolle:
So what we can agree on is.... most recording- and mastering-engineers know that more headroom makes records sound better.... and our clients dont understand and we have to please them toget a louder/more distorted product.

And who should we blame...the consumer...the record company...the producer... the market...
????????????

Maby we should blame the ourself.....

Olle

Fact is, if we all refused to do it, people would get used to the idea that mastering engineers respect their craft and the musicality of the work, and are not willing to sacrifice that for the almighty dollar.

Of course, while we're talking about ideology...

anonymous Mon, 11/10/2003 - 15:42

Fact is, it's almost a mathematical certainty that a mix that's peaking intermittently at 0dBfs will produce a waveform that, when reconstructed, will exceed Full Scale by upwards of 6dB (and therefore clip the consumer's deck)

Griffinator - you mean because of some potentially extrapolated vectors ? I mean if the preceeding 'x' number of samples were headed towards the Moon then stopped (because of a clip or something) then the reconstructed signal might overshoot or something ? I hadn't thought of that. Wonder how I measure for that kind of scenerio ?

kylen