When I record I generally try to keep the levels at -6 /-8 db and make absoulutely sure there is no clipping.  When I start mixing I pull up voxengo span and set the k metering to k-14  and reference it frequently. When mastering I do the same till I am done.  Then I play the whole track start to finish, if I peak at say -.9 I move the master fader up by .8 so that my highest peak is at -.1 This gives me a little higher RMS and percieved loudness without limiting the piss out of it.   Just my take and what I have learned to do so far with pretty good results.

You can of course do a 'poor mans' k-system setup and get most of the advantages. Basically just do something like this:

1. Set your analog volume controller knob at some reasonable level, like 3/4s up, and mark that position.
2. Select a number of commercial tracks that represent a wide range of compression. So something like maybe Dark Side of the Moon on the low end and something like Green Day on the high end.
3. Import these songs onto tracks. Leave the track faders at 0
4. Pull the master fader down until those tracks are peaking at whatever you want your reference level to be, say -12dBFS on the master bus.
5. Now start playing these tracks. Adjust your speaker sensitivity levels so that the most compressed ones are uncomfortably loud, and the least compressed ones are a little bit quiet. That will mean that reasonably compressed stuff will be somewhere in the middle. Go back and forthe between the tracks to find a happy medium, but be sure that the heavily compressed stuff is just too damn loud to listen to comfortably.


So now what you have is a point on the analog volume control that you can go to and know that it's at a known point, and you have a master volume peak level (-12dBFS in this case) that you can then allow your own stuff to peak at (with the master fader back up to 0dBFS again of course.) So now, since you've set a known master bus peak, the only way you can get the music louder is to compress it more (bring up the average level.) And you have already determined that the actual SPL in the room is set such that overly compressed music is too loud.

So basically you have a situation now where you will tend to naturally gravitate towards a reasonable and appropriate level of compression for the song you are working on. Too little and it starts getting a bit too quiet. Too much and it starts getting uncomfortably loud.

This is not a formal setup, and doesn't use an SPL meter which is convenient to have, but it actually works very well and if you later do a real one you'll probably find out that you end up very close to the same thing. The basically point of it is to just force you towards appropriate levels of compression naturally, and of course it also provides you with a standard monitoring level so that you can have a consistent Fletcher/Munson response for gauging the overall mids vs. lows/highs balance of the mix (which will change a lot as the volume in the room goes up and down.)

If you have an SPL meter, you can measure the commercial tracks' SPL in the room, then put the master bus fader back up to 0dBFS and turn down the analog volume knob until you get back to the same SPL. Mark this one and you now have a reference position for listening to commercial music that should be at the same SPL as your mix levels are, which is nice for comparison purposes.

People are only referring to peak levels when they mention any recording levels, and that does not take into account what the rms level of the signal is underneath it.  

This is incorrect.


please note my last post:

so, back on topic: 

i personally record almost all of my tracks with no more than -10db peaks..... 

with the exception of drums, which i allow to -8db peaks, but i process them at mixdown, and knock the peaks off with compressors and limiters.... 


i find the closer i get to -8db peak as a MAXIMUM allowable peak.... 
the more 'crunchy' the sound is. 

that actually works for me, with drums, but not with anything else. 


if i record something all the way down to -14db peak (which means an average RMS value of about -23), that seems to be the "SWEET SPOT" for my system, signal chain, method of recording, etc. 

Most people are only seeing and therefore working with peak levels only and that is the reasoning behind my statement as such. Few are measuring rms levels consistently and accurately.

In batsbrew case above he is implying that his rms level is down at -23 and any transient associated with that sound is up at -14 meaning the transient is 9db above the rms value. (this would be for a transient type of sound with some degree of attack) K System wise this is close to the -20 situation. I would adjust the rms signal to be around there putting the peak at -11 above that. That is the good thing about the K-20 ref level. There is 20 db of headroom above it so that peak still has 11 db to go before it clips anything.

I don't agree that digital recording suddenly has a sweet spot. I feel it sounds very similar over a wide variety of levels. Especially at 24 bit recording where this is much less of an issue.

If I were recording bats  signal now at a K-14 db level then the rms part would be at -14 and the peak would extend up to -5db which is still fine but with only 5 db of headroom above it. The K-14 ref levels is going to 6db louder than the k-20 ref level but with less headroom for emergencies etc..(Note with other types signals you can have an rms part still at K-14 and the peaks only 2 or 3 db avove that)

There is a simpler way to calibrate DAW's and monitor systems as well. You do need some measuring devices you cannot get around it. Load up a stereo pink noise test signal at the desired K syetem ref level. (Bob Katz webiste has a perfect pink noise test signal at K-20 if you cannot make it yourself, just add 6 and 8 db to it to make the K-14 and K-12 test signals respectively) Your VU's should be showing very close to 0db VU. (They have to be set for the same ref level obviously, Klang meters can bet set at any ref level)

You simply adjust your monitor gain for either 83 db out of each seaker alone or 85 db for both. (Pink noise is a must otherwise standing waves will severly effect tones by only moving the meter slightly)

If you are comparing commercially mastered music to your own work at your DAW ref level etc it will obviously be much louder and pin the VU's. So you simply park that on a track and drop it down so the VU's show 0db now.  eg will be -6 to -7db on average.

When you master now you simply put the commercial track back to unity gain on the faders and adjust your VU meters to be calibrated at say -6 or -7db rms instead. Then the commercial trrack will appear normal and peak 0VU and now your DAW output will be shy by the same amount. You perform your mastering processing and slowly get it up to match. The monitor gain in your room has to be lowered by the same amount as well to maintain the same 85 db in the room.

83dB can be too loud for a small room. It definitely was in mine. The ultimate purpose of it isn't to reach a particular SPL, but to have overly compressed material be too loud in your room, reasonably compressed material be just right, and very uncompressed material will probably be a bit overly quiet in the quiet parts. In a studio sized room, 83dB is probably there. In a small apartment room, I just felt it was too loud. I went more for the mid-70s. Going back and doing it by ear like I did above (as an experiement) got me to almost exactly the same place, which isn't suprising.

Obviously if you have the SPL meter then use it. But it's not a requirement that you use 83dB or 85dB. If that's too loud, then it's too loud and adjust downwards.

I think I may have mentioned this once before. 85 dbA is loud. 85 dbC weighting is not very loud. It is a beautiful volume. Listen to it all day long, no fatigue. C Weighting is the correct weighting to use.

PS Dean is right though. Even at C weighting 85 db of pink noise is still quite loud but 85 db of music is not. And we don't listen to pink noise all day do we? So that is why the pink noise signal can seem a little loud.

All I know is that computers are 100% not over, maybe squeezing 103% at most. Get to 105% and you what I call a A-hole, where the recording peaks over the limit a computer can handle and drops to the other side of the center line putting a cracking noise in the recording (I done this not on purpose with old band analog tapes.) Analog recording could go over the limit but distortion may be a factor. Better studio conditions of digital may handle more, but with home recording equipment and computers that is the way that it works. The VU meters in Sonar can be adjusted to respond faster up to a point if your computer can handle the load. Right now they probably are 40ms delay where you can set them to 10ms delay I think (I am correct) but it will take more CPU power. Whatever happens it is a cumulative effect with each track adding to the total overall volume in the end. Of course some tracks will be softer than recorded in the end while others will have to be lowered to keep from peaking over 100% on a computer. No leeway (unless and probably bigger studios have better equipment that allows them to do a little more, not so with home equipment). The more you pay the more you get, but there is a point where increasing the total production becomes cost prohibitive in the end and may not be needed.

Just an an aside, Gearslutz.com is now coming up as a porn site as well, anyone else seeing that?

For my happier mediums and crowded mixes:

Lead vox track: -10 dcbs as a starting point for peaks

Kick: - 4 dcbs (peaks)

Snare: - 6 dcbs

Backing vocs - 20 dcbs

Ambient strings: - 12 dcbs

Guitars: if Haas'd: -15 dcbs

Chuga guitars: -6 dcbs +/-

During mixing, the ears take over and violate a lot of my conventions.

Panning requires unpredictable ear-decisions. 

Extreme panning sometimes requires lower volumes (rock) or higher volumes (dance).

Per Bat, drums dominate my dancier mixes, and are louder.

Hi Phillip Just a few observations. 

What is a dcbs? Is this your slang for decibels. Others like my self may be wondering what unit you are referring to. Decibels are best abbreviated to db. That is the convention.

Are you referring to peak or rms levels. Big difference!

Also you cannot recommend levels for a mix because there are way too many variables. The first is how loud the track is recorded to start with. And the other is what you think is a nice level for say backing vocals may be too loud or too soft for me or others. I think with mix levels people just have to put things where they think they sound best and what suits the material the best.