K-System - Do I have this right?

Would some of you folks in the know please critique my understanding?

To use the K-14 system of metering, I would:

Playback pink noise and get -14 reading on my "Mains" meter.  I suppose at this point it doesn't matter if we're talking peak or RMS because it's pink noise, I would think peak = RMS.

Now, I adjust the gain on my monitors until I read 83dB SPL on my sound level meter at my listening position in the room, and leave the monitors there.

Now, when I mix, just using my ears, not relying too much on meters, I would tend to naturally gravitate toward getting my RMS values somewhere around -14 dBFS (83 dB SPL), because above that would start to get uncomfortable, therefore leaving me 14 dB of headroom for peaks, therefore I wouldn't need to use compression to reduce the peaks to avoid clipping, all for the desirable goal of maintaining the dynamics in my recording.

I listened to Bob Katz's CD Honor Roll and I really do like the sound of those recordings, which I understand to be significantly due to maintaining dynamics, not to mention the performances, and other recording/mixing techniques.

Thanks for your thoughts, opinions, and guidance,
Sam

Thanks Middleman. Amen on that 83 dB. I just wanted to present it "by the book". I think I'll actually set it up lower too. 

I saw the additional comments - I think I was there a split second after you finished.

Hey there jrfrogers A few other things to think about.  I work at mainly K -14 dB FS as well. Pink noise I have found is about -1db compared to the sine wave. A good test signal is a stereo interleaved wave with a 400 Hz or 1 Kz sine wave recorded so the very peaks of the sine wave are at - 14 db FS. I have uploaded a -14dB tone and pink noise on my Soundcloud and made it downloadable.

http://soundcloud.com/jeff-evans


You now need a VU meter to show you a 0dB VU indication while this tone is playing back at -14 dB FS. Good examples of VST VU's are here:

http://www.klanghelm.com/VUMT.html

The normal DAW meter is not great because it is hovering around some point well below unity. (If you are going to use the DAW make sure they are set to rms and peak) The VU can be used on tracks, busses and mains now so it should also be just reaching 0 dB VU on average music levels. These extra VU's are showing you the rms component of the signal and they are showing it high at 0 dB VU. The peaks won't effect the VU meter much so you don't have to worry about them. Let your DAW do peak metering as normal so you can use both and keep an eye on both.

After the sinewave calibration you will find the pink noise to be around -1 dB on the VU meter. This is more correct.

When making dB SPL measurements in your room make sure on a C Weighting. This is always left out. 83 dB or 85 for both monitors on is not loud. (Unless I am bit deaf) It is a very nice level. Monitoring for extended periods below 85 dB is dangerous. (only from a mix point of view, not SPL) You will be adding too much low and high end to make it sound right.

K System is about three things, choosing a dB FS ref level and putting your sine wave level there, 2nd, have a proper VU meter (or the BlueCat) as well that must have something to show you a full scale 0dB deflection on a scale while the digital level is at -14 db FS, 3rd keeping monitoring levels in your control room consistent.

If you change your K system ref level to either -12 or -20 dB FS then you have to adjust your room level to suit. (and your VU meters! You can set the K ref level in the Klanghelm meters) You don't have to stay at 85 dB SPL either. It is just a good ref level, more to do with health and safety issues. (It is also aligned to the dialogue level in a movie theatre) You can listen to your -14 dB K level ref at any volume you like. But 85 is very good and great for extended periods as well. Listening at higher volume exposes bass and reverb issues as well as other things.

You may wonder about the other K ref levels. K-12 is obviously 2 db louder than K-14 and for 16 bit only has been recommended. Also for broadcast as that is what the -12 level is called, for broadcast. But you need a little more care especially with tracking levels to ensure any peaks dont go over 12 dB. Analog limiter comes in handy here. But when mastering to louder rms levels, K-12 has only a little way to go. K-20 dB FS has that amazing transparent and most transient sound. Good for album tracks and pristine mixes. You can also master from -20 way up to loud mastering levels too but you have got further to go to get there. K-14 is a great overall ref level to work at. My Yamaha digital mixer puts out +4dBu (analog outs of course) at -14dB ref level with the sinewave peaks hitting - 14dB FS.

What weighting are you guys using? (A Weighting is incorrect, it basically ignores the bottom end and will give a reading that is too low for music) Also don't listen to the sinewave or the pink noise at 85 db SPL. (even with a C weighting) That is quite loud. Listen to music that is just reaching your chosen K system ref level (on your meter) and you should find it is quite nice. But I do agree if you still feel the music test is loud then I may be a little insensitive. To me it sounds just nice and a nice level.

When calibrating in a small room (ie. not a film mixing room) you should not use full bandwidth noise and C-Weighting. Use band-limited noise and A-Weighting, since small room modes will completely screw up a full-range C-Weighting measurement.

Thanks Steve - very helpful. When you say 

"you should not use full bandwidth noise and C-Weighting. Use band-limited noise and A-Weighting",


do you mean:


1) A-weighting on the sound level meter
2) I used the pink noise files from Bob Katz's web site, it's 500-2K -20 dBFS RMS


I did do one speaker at a time.


Sam

Corey  is completely wrong regarding the A weighting. This is where the forum is full of mis information. I was confused re which weighting to use. I did the research and it is actually on Bob Katz website where he says you must use a C weighting (it is not obvious and it took me a while to find it, also it is in his book on mastering believe me) with all your measurements, small or large size rooms. That is why you guys keep talking about 76 db SPL etc. You are obviously on the wrong weighting. If you switch it over to C weighting you will hear a very similar SPL will be 83-85 dB setting. A weighting virtually ignores the bass end. Bob says if the bass is involved then you need to use it, simple as that. Because the bass does add to our perception. 83-85 dB SPL C weighting with music is NOT LOUD! If it is so loud people how come it is the acceptable SPL limit for health and safely over an 8 hour period?

Now Sam. There is a version of the pink noise that has been band limited. You have to be careful with this one. Firstly I would use the full range pink noise first and do your measurements. (That is up on my Soundcloud if you need it)

The reason for the band limited pink noise is to avoid problems that may be caused in your room due to low end issues so it avoids them. But first as a lot of the spectrum has been removed, this signal plays back much lower so you have to compensate by pushing it up level wise so it also just shows 0 dB on your VU meter. (around 6 db from memory) Hence the importance of the VU meters showing 0 dB VU when you are at your ref level. There is not a lot of difference between the A weighting and the C weighting then with the band limited pink noise test. However I found I got identical results with either the band limited noise (adjusted for level) or full range pink noise. (at normal level. You must adjust for levels between these!!!, Another reason to start with the full range pink noise because it is already at the right level. Also the full range pink noise on the Katz website is at -20 dB FS and you have to add 6 dB to this if you are working at K -14)

It is a good idea to use the band limited pink noise but follow it with full range pink noise and music as a backup, because you don't listen to music that is band limited do you! That is why you should also do a final music test (full range material) set at your K level and your VU's should be around 0dB and your meter on C weighting and your SPL level around 83 to 85 dB.

The Sonar meters are true rms and hence they are the reason why they show -17db rms with a -14 dB rms signal level. But as I have said and no one is getting this, you need a separate VU meter (not peak otherwise it will move wildly to peaks and you don't want that) that is showing right up to 0dB VU while you are at the K level. If you don't have this you are sort of wasting your time.

On the Bob Katz website it says nothing about setting your SPL monitoring ref level to say 77 dB SPL A weighting. It only talks about 83-85 dB SPL as the ref monitoring level so therefore he is referring to the C weighting. You can do one speaker at a time but I find both at 85 is very similar to one at a time at 83.

This can all be a bit confusing and I can understand if you feel that way. The band limited pink noise signal is even more confusing because it is not even at the right level. It is simply full range pink noise that started out at -20 dB FS and has had everything below 500Hz and above 2 KHz removed from it but the level has not been adjusted. Bob is assuming you have a proper VU meter that is normally showing 0 dB VU when you are at your ref level so it easy to compensate for the lower level of the band limited test signal. Another reason to START with the sine wave signal I have put up first. Because then you can set your VU meter ref level easily. It is a bit harder with noise as it fluctuates a bit.

The A weighting and C weighting on an SPL meter are different things. The A weighting is good for things where there is not a lot of content in the sound you are measuring under 500Hz. It is also good because it avoids false readings due to any lower frequency sounds that may be around at the time of your measurements. But with MUSIC folks we have bass present so get your SPL meters off the bloody A weighting and start measuring things with the corrcet C weighting instead. You will see once you do that the bass has a very big influence on the reading hence it makes you keep the music (around 8 dB) much lower SPL level for the correct 83-85 dB reading on your SPL meter. Our work involves the full spectrum and that is the reason why we use C weighting in our tests. This is the result of some fairly in depth research.

You DO need some sort of VU meter to show you a 0dB VU full scale rms level while you are operating at a much lower digital level. It is vital! (BlueCat meter is OK and very good but a VU is easier to read and very simple) You cannot see things accurately down at -17db or -23 on Sonar's meters. As I have said, unless you have some sort of rms meter to show you  a 0dB VU FSD while at the ref level then you are not using the K system correctly. Studio One can put its (buss and master) meters into K system mode and then they all read right up to 0db VU with 14 or 20 dB of headroom above that. I wonder why they do it! It is because you NEED to see a 0dB VU meter while you are right down at -14 or -20 etc. Otherwise you will never get it accurate. Any advice contrary to this is incorrect. (the Bob Katz website talks about the meter big time!)

Example: If you are adjusting your track level or buss level how are you going to do it. By watching your DAW meter hovering around -17 db rms or watch a nice VU just peaking right up to 0 db VU. See what I am saying. You tell me which is easier. You know when your VU meter is right up to 0dB VU you have got either 14 dB or 20 dB of headroom above that.

My room gives an identical reading with the full range pink noise and the band limited pink noise (adjusted) so explain that one. It means my bottom end is very accurate here.

If you do find that 83-85 dB C weighting levels with music are too loud for you then fine adjust to suit. My speakers and room are set up so that the 83-85 dB level sound pretty nice.
 
Have any of you bothered to switch your SPL meters into C weighting and listened to music? I wonder. I get the impression most of you are stubborn and are stuck on the A weighting instead. The moment you switch over to C weighting the bass really moves the meter big time so you tend to pull the music way down to get the 85 db level reading. It still does not sound loud to me.

It does not matter how big your room is either because the 85 dB SPL signal is right where you are. It does not matter if your speakers are 5 meters away or 1 meter away, you are simply adjusting your monitor gain to get a reading of 85 dB SPL where YOU are. In a larger room however with the speakers further away they will obviously be putting out more sound in order for the correct level to arrive where you are. Very close monitors don't have to work as hard to get there.

You may not get all your answers here Sam due to confusion. To avoid that  I suggest you go the Katz website and carefully read the information that relates to all this.  K system is a great thing and it will standardise all your track, buss and master levels. All your mixes will come out at a very similar level. You won't clip your DAW anywhere to be seen and mastering is also a breeze and very consistent.

Go here are read articles 11 and 12 (Level Practices)

http://www.digido.com/articles-demos.html

The K system is about consistent SPL levels in your monitoring environment for sure but it also about keeping rms levels consistent right through the tracking, mixing, bussing and final master levels also inside your DAW. And to do that we work at a digital level that is lower so we can have headroom. But we need some form of meter to show us a 0dB VU FSD indication while we are down at those lower digital ref levels. You cannot use standard DAW metering when in K system mode, period. Of course we also use our peak metering to keep track of those very transient types of signals that won't move a VU very much but for the most part the VU works very well. Even if you cannot have real ones the Klanghelm meters are a very good option.

Everyone have a good Easter too from the K system guru!

The only signal where peak = RMS is a square wave

Or if you're a Guitar Center salesman.


Kind of refreshing to be debating weighting curves and such, rather than whether SONAR is only suitable for Motown.


Count me among the 76db crowd. What sometimes gets lost in discussions of K-metering is the main reason for employing any standard, which is to encourage consistency. The absolute level that you monitor at is far less important than simply sticking to one level. You ears will train themselves regardless, even if you're in the less-flat region of the Fletcher-Munsen curves.

I would add that compression is the sound of modern music. Learn it use it. Everything you hear in commercial music is compressed extensively. A lot of times it's not the fact that compression is required but that it provides competitive sonics. Whether you use transparent sounding compression, gritty sounding compression or slammed sounding compression. Learn the sonic characteristics of the following and when and where to use them.

CL1B
LA2A
1176 - various versions
API 2500
SSL both channel and buss versions (E&G EQ curves)
Neve 33609
Distressor

And also in the compression vein:

L1, L2 limiters

These all have their own sonic characteristic and there are others which you should understand. For this a UAD2 card and associated plugins are an invaluable learning tool. In summary, once you understand when to use these and the characteristic they impart on the sound. You are moving closer to getting commercial results in modern music. For classical and jazz less so.

One of the main goals of the K system is to be very accurate with your metering and that is why standard DAW meters (except Studio One of course) will not give you the accuracy down low on their scales compared to reading something that is approaching 0 dB VU on a VU meter scale.

It is absurd to think the meter becomes less of an issue and you can just feel the right levels in the room. That is rubbish. You could be out by 1 or 2 db and not hear it that much overall level wise and you certainly won't see it on a DAW meter that is down at -17 dB. (or -23 db) Also if your room level is out by a dB or two the error will compound. BTW I have a permanent SPL meter in my room all the time. Very revealing actually.

Katz talks about the meter very seriously in his info. He used a very expensive Dorrough meter to do the level calibration and also to keep check on exact levels. Metering is vital and very important otherwise your K levels get out of whack. It is the single most important aspect of it.

I am not referring to mastered levels or anything like that either. K system refers to unmastered levels at all times. However when mastering I shift the K ref from the ref I was working at to commercial ref levels instead eg K -11, K -8 or even K-7. I noticed a Steely Dan CD was mastered the other day to K-11. Good combination of volume and transients. Many albums are much louder than that. K system also allows you to measure and accurately track commercial mastered levels too. (something many here keep saying they want to do) But you also need the meter to do it as well. Even more reason why the meter is vital, important and critical. Without it you are actually working in the dark.

Many people do not have any concept of the VU meter because they were not producing so much music in a time when that was all there was. The modern DAW has simply dropped it and gone over to mainly peak reading devices. You really need BOTH! Many around here have not even seen or used a real VU meter either. But once you do you will never go back. It also suits the K system of metering very well. It is interesting that it is still present in all major recording and mastering facilities. The single main reason why people screw up their track and buss levels is because they don't know how to meter correctly and they don't have a VU meter to help them.

Also any advice about only trusting your ears and not trusting a meter is also rubbish. In fact a quality VU meter can tell you heaps about a mix. The ballistics of the meter can actually inform you amazing amounts of info about your mix. It is intersting to watch very well mixed material compared to your own on a VU meter. They move quite differently in fact. (if your mix is poor that is) Even if the levels are right. I have noticed as your mix gets better the VU starts to move in very similar way to a great mix. It does take some practice to learn to read and interpret a VU meter response. Something many here won't bother to do. (This is where real VU's come into their own as well) Also if a VU is swinging wildly but you cannot hear why, it makes you look into the tracks and find the offending track. And often it is a track with subsonic rubbish on it that is not useful anyway. Unless your monitors go right down there you wont ever hear it either and it will get through.

With lower volume monitoring levels I go all the way and use a single mono speaker with L+R being fed into it and I monitor that at a very low level (65 dB or less) That is a real eye opener in terms of your mix and vocal level setting. So for me the 85 dB SPL C weighting sounds Ok but it may have been all those years I played in very loud bands that makes it so. LOL! Do remember though that the ears best response to music is NOT at 77dB or where most of you are monitoring, it is actually in fact around 83-85 dB so there is good reason to work at that level. It won't kill you! I have said this before, your mixes and mastering will sound its best if you do work at that level. I think if you are going to monitor lower than that then you need to come up there for checking your mixes at least. I go much higher for loud mix checking. eg 95 dB sounds pretty cool and so does 105 dB but obviously only for short periods. You really need to check your mixes at many levels. You are doing yourself  a disservice if you don't.

Download this pdf, from the ATSC (Advanced Television Systems Committee), entitled "Recommended Practice: Techniques for Establishing and Maintaining Audio Loudness for Digital Television". 

Chapter 10 describes setting up and calibrating different types of mixing rooms. It notes that loudness is dependent on the size of the room: identical SPLs sound louder in a small room than a large room. 

Their recommended calibration procedure specifies 76db as the target for rooms under 1500 cubic feet (this would describe most bedroom studios), 78db for rooms under 5000 cubic feet, 80db under 10,000 cu. ft. and 85db only for rooms over 20,000 cu. ft. in size.

When I settled on 76db it wasn't because of any scientific justification or expert recommendations. It simply felt right. Loud enough to sound full but not tiring.

Thanks for the document bitflipper. I have a small room and 76 is good for me too.