How much head room should I leave on tracks ? Volume problems and cliping problems

You should mix your track so that your master bus peaks between -6dB to -3dB. That will leave enough headroom for the mastering process.
There is no headroom for mixing. You mix it until you achieve the proper levels for each track, buses and master bus. Its all relative, and each track will have different levels. But, to get it ready for mastering, your master bus should be between -6db to -3dB

You should also have no EQ, compressors and limiters on your master bus.
Cj

I like to try to keep all my busses - including the master bus - at 0 throughout the project, controlling the final level by using the individual track faders.

Sometimes this isn't always possible without having to go back and adjust 50+ faders, but this is an uncommon scenario for me.

I do try to keep my master and the other busses up to 0db in the whole process. I have been known to throw an envelope on a buss from time to time rather then adjust the individual tracks...especially if they are balanced but just a touch to low or high.

Since I am not sending my tracks out for mastering..... I do it all in  house... I set my levels as needed on my exports. Highest peaks to 0db. Everything else is below that level...and I really don't want a bunch of the peaks at that level...... from heavy compression.

It's very possible to have a sufficiently loud final mix and not compress the crap out of it.

The word is "Dynamic Range".... uhhhh that's two words, but you know what I mean.

mike_mccue


I was in Chicago last week driving under the "el"... It was night time... we had just left Pizza Due. I decided to throw in one my my home made blues jam CDs.... a few moments later the El ran over our head... it was noisy... I thought to my self... "not enough compression on this track". :-)

But Mike...you obviously had plenty of head room because you didn't mention bumping your head on the "el" when you went under it.

I do a lot of recording sources that are only good in most cases for one run through a reel to reel or one pass on under a transcribe record needle, etc....

So I HAVE to ensure I have enough headroom so that it doesn't distort as I have one shot at it... so in this case I usually sit around -18 db for nominal (Meaning that if someone is speaking the meter sits around -18 and can peak up to -4 as volume and intensity of the recording increase.

If you can ensure you have a few db headroom you can always use processing to lift the volume up.

If you can't get enough Drums out of your synth try bouncing the synth to an audio track and then you can process it with volume or compression and automate the track level, etc....

this means no output gain on the compressor and setting your threshold to just push down the peaks-looking at it visually you can see the peaks and determine what db to set the threshold- then put the ratio at around 15:1 or higher

I forgot to mention the attack time should be at 0 and release should be around 400ms to create a semi limiting compressor otherwise you will have higher peaks at the beginning of each sample that hits the threshold as it will wait a certain time before "acting" on the sound.

jcatena I don't agree with your K system views that K -14 and K-12 are useless and not necessary. There are very good reasons why these other K system levels have been chosen and used. I am someone who uses K system levels a lot and I use all three levels for different projects.

Let us say that a certain project needs to end up sounding loud. Say an average of K -6db and that is loud. (after mastering) If you begin a project like this at K -20 then you will end up with a final mix which is also at K -20. Now you have got some serious level increasing to do in fact a whole 14 db of it in fact to get your final level up to our K -6 level. And that is a lot and all the mastering processors will be working hard to do it.

But if I know a project has to end up loud I work at K -12 from the start. Then I have only got to get another 6 db up on my final K-12 mix to achive the desired loudeness. Now the mastering processors are not working nearly anywhere as hard and hence they will all sound much better.

That is the reason for the three K system levels.  I use K -14  for a lot of projcets too. It just seems to sit well and you end up with a pretty loud mix at the end of the day and the transients are still healthy at 14 db or so.

But I do agree that K -20 is a beautiful reference level to work at and the mixes just sound the best. I master this not so hard as well and maybe take it up 6 db max.

What sort of metering are you using with your K system operations? I use real VU meters and also the BlueCat meter works great. Remember you need a rms device to show you 0 db rms when you are working at the desired K level. Sonar does not do this by itself. You have to switch your level (meter) calibrations and monitor gain calibration everytime you alter your K level but I am sure you know that. Glad to see someone though who is into the K system thing. I wish more people would get onto it. It would just save so many issues with clipping etc. I never see one clip light at any of the three K levels on tracks or busses.

Mike I can understand your curiosity and I am happy to explain. Firstly I think digital recording needs a system. With analog we had VU's telling us levels going in and out of equipment and it was easier to get some sort of consistency. But we do know that 0db VU did represent different magnetic flux levels at times on tape but even with that there was still a sense of calibration going on. If you knew what flux level represented 0 db Vu then it was easy to set things up again to read 0 db VU going into and out of a mixer etc.

Many people who are recording digitally are using no system of reference of any kind. It is not about preventing clipping. We can all do that with care. But if we were to measure some sort of average level of a mix from 10 people here you would get 10 different levels.
This where K system helps to reference things a bit. I abuse the word rms too but I am really referring to average level of the music or the average level of the main body of the music (without the transient peaks involved)

So when I say K -14 db I am referring to the peaks of a sine wave reaching -14 db. That is the standard that Bob Katz uses. The actual rms value of that wave is 3 db down from that and that is why the Sonar meter shows -17 db when I am looking at a K -14 db level. So we need to get clear about that.

OK now I use a digital mixer and this helps a lot. You could not do this easily with an analog desk because it is hard to get it exact all the time. My VU meters are 600 ohm (correctly terminated) and they are fed from a special VU meter amplifier and I can adjust the gain of that amplifer very accurately. The VU's are across the stereo output of the digital desk. My soundcard is an EMU 1212 and it is connected digitally to the digital mixer. If I send a K -14db signal from the soundcard it goes right through the digital mixer to the outputs and comes out at exactly the same level. (unity gain right through the mixer) Now I set the gain of the meter amplifer to show 0 db on the VU's. It is not important what the actual voltage is that is driving the meter movement but FYI it is 0.775 V. The whole purpose of the K system is you see 0 db VU when the K level is chosen. The mixer also sends very accurate callibrated levels from its own built in oscillator so the first thing I do is tell the mixer to send K -14 then adjust the VU's to read 0db. This corresponds to an rms voltage that is 17 db down from 0.775V (0.109v or 109mv)

If I generate a sinewave say in Adobe Audition at -14 db (and import that into Sonar) then the soundcard sends exactly the same level as the mixer does. When I measure it, (from its analog outs which I dont use BTW) it is also 0.109v so I know something is right. (I have a true rms digital meter) Now I turn off the alignment tone from the mixer and play the -14 db signal from Sonar into the mixer and 0.109v comes out from the soundcard and mixer and the VU's show 0 db.

If I decide to work at K -20 then the first thing I do is reset the alignment level from the mixer (down to -20) and it is now 6 db quieter and the meters show it of course. But because I have got a calibrated meter amp driving the meters I can switch them up 6db so they are now showing 0 db again. But what this means is that now everything I do in Sonar is 6 db down from before. Hence the extra headroom. Because the meters are more sensitive now, any input signal can be 6 db lower from before and the meters will show 0 db.

I use the BlueCat digital K system meters too and they all agree with the VU's. Both on tracks and busses. If I want to work at K -12 then I do everything as before except now everything in Sonar is working 2 db higher than K -14.

Bob Katz has got a reference level of K -20 (full bandwidth pink noise) on his website that you can download. If I setup my system for K -20 and plonk that on any track it shows exactly 0 db on the VU's so I know I am doing everything right.

As far as VU's go I just love the ballistics of them while the music plays. I know they are not showing true rms but some sort of average level, but it is the ballistics that I get the most information from. One can learn heaps from watching how VU's behave with very well mixed and mastered material. Then when you are mixing they show up all sorts of issues in your mix. (If you are wondering, the frequency response of the VU's is flat from 20Hz to over 20 Khz!, they dont even budge over the whole freuqency range so they are capable of showing a lot!)

When I am mastering I can reset the system now for higher levels. When I import a commercial CD onto a track and I have got the system setup for say K-14 the meters will pin over of course. But it is simply a matter of either lowering that track level or adjusting the VU meter amp to bring it back to 0 again and it is by how much I can tell how much louder a commercial CD is above K -14. (about 6 to 8 db on average) So when I am mastering I simply recalibrate the mixer and the meters to say K -8 and then my K -14 mix will fall short of course but I then know how much I have to extract out of say the EQ, the Multiband and the limiter in order to bring it back up again to 0db.

I find that when mastering the only way to get all the tracks to sound the same loudness is with the VU meters. When I get all the tracks of an album just hitting 0 db VU it is only then that they all sound at exactly the correct volume going from track to track. And what is interesting is when you look at the tracks in Adobe Audition they are not all necessarily at the same height either. It depends so much on the program material.

So once again if I set up my system to say K -14 and played 10 different mixes from anyone here I would get 10 different levels on my VU's. Not only that, even if one person here sent me 5 mixes they would probably all be at different levels as well. But by using a system of reference you can get all your mixes nice and consistent.

Of course I recalibrate the monitor level in the room everytime I change a K level so no matter what K level I am working at I am hearing the same level in the room. I do that with band limited pink noise. (500Hz to 2 KHz) Bob has also got a K -20 pink noise (full bandwidth) download and that also plays back at 0 db when I have my system setup at K -20 (actually it is about -0.5 db but I think that is just the nature of the noise)

I dont know how anyone can do anything without VU meters. The Sonar meters are good for peak of course and that is important but the rms levels are so far down they are meaningless. Every decent mastering suite in the world has VU meters in there and all the best engineers use them. The Durrough is a great meter too. It was just cheaper the way I did it. They can reveal even more information. VU's great for setting record levels too, you just keep adjusting the input signal to get 0 db on the meter and that is it! If you are working at say K -14 then you know you have got 14 db of headroom above and that is why you never get clipping when you use a K system approach.

I hope that helps. Sorry this was so long. I have been accused of writing a novel at times on these forums. I wish I could be as succinct as Bit! I do teach this stuff too and it is just the teacher in me wanting to get across the concept as best I can.

Sorry Dean but I get the impression you dont fully understand it. Firstly it is about the digital world because we know that hitting 0db FS is not good so it is about staying well under that. In the analog world we often had 10 to 20 db of headroom above 0db VU so we did not have to worry about transients but in digital we do.

But you have made an incorrect statement here:

The point is that you set up your system so that you know a given level in the digital world corresponds to a given SPL level at the listening position. That's the key to the whole thing. So, whichever K level you are doing, you set up so that the K-System 0dBFS corresponds to a level in the room that is known.

The correct statement here is that: So whichever K level you are doing you set up the K-system 0db (RMS) corresponds to a level in that room that is known.

And as I have stated above, when I use the term rms from now on I mean that rms signal is a continuous tone except that the K level represents the tops of the sinewave peaks so it is 3 db higher than true rms.

There are three K reference levels don't forget. K-12, K-14 and K -20. They all show 0 db VU (rms) and yes I fully agree that they all correspond to the same SPL level in the room. Except the they all have different amounts of headroom available being 12, 14 and 20 db respectively. Bob based his K system design on the movie audio standard which is 85 db and that the digital rms level to achieve it is at -20 on the film. Which allows for not only constant levels of dialgue etc in the cinemas but a huge 20 db of headroom above if needed.

The K testing signals are all continuous tones or noise, its not about peak signals anywhere but it is about about how much headroom above you have. Bob Katz suggests that 12 db is about the least amount of headroom we should have in order to preserve transients but 20 db is better of course.

But yes on your point about maintaining a constant monitoring SPL in your room, that is very true and useful. It is about that but it is also about the fact we can do it at three continuous rms digital levels that have different amounts of headroom above.

Everyone forgets there are two aspects to a signal, its rms value and peak value. Being older and coming from the analog days we were concerned about rms only. But now in the digital world all the VU's and rms indicators have gone and the younger generation is obsessed with peak levels and that is important too because of the 0 db FS ceiling. It is the reason so many have problems with recording and mix levels because they are only looking at one aspect (peak) instead of both.

I wonder how many here actually have the rms and peak engaged in their metering. Bob is just trying to bring back the rms aspect as well as the peak aspect. But even if you do have rms and peak engaged, it is not the same as seeing that rms signal on a standard VU meter at 0 db VU. We really need both now. But as the rms values are so far down they are hard to see on any standard metering (and if you are only looking at peak you are not seeing it at all!) and that was the thought behind making a K system meter that shows a full scale indicator of 0 db (rms) while the signal is actually sitting at the K level well below.

A normal VU meter wont show the transients because it is too slow but the Bluecat meter is excellent because it shows both at the same time.

I agree about what the digital levels are for the rms part of the sound not being that important but I can see where they came from. The 85 db spl cinema audio thing with 20 db of headroom above that. The next major level shift upward though would be 6 db and that is where -14 came from. I have been working at -15 for a long time before I discovered K system.

The fact that the digital levels could be anything is not quite true because if you chose a level of say -40 db as a reference, then the digital noise floor which might be down at -90 db is now only 50 db away from our ref level. It may become audible. But -20 means we have still got a very hefty 70 db down before we are going to hear any digital noise. This is where 24 bit recording is a big plus too. Even if the CD ends up at 16 bit, that initial resolution there is well worth it.

You are so right about today the rms and the peak levels of the music today are too close and what Bob Katz is also saying is let us get that rms part back down to a sensible level and just turn up your monitors. If we drop the rms part back down from  -1 or -2 down to -14db then we can turn our monitors up another 12 db and we are right back to where we were. Except the music can sound transient. Imagaine how good a heavy metal mix would sound mastered at -20 but listening at 105 db spl. The drums would knock you over.

I dont think we will ever hear the full potential of digital recording and how truly nice it can sound until we get all mastering houses to conform to a standard and why not make it -20 and then crank up our monitor systems to make up for it. We have got great monitor technology and we can deliver high powered loud music without distortion. But we are using up that last 1 db of digital to make up for all the poor listening systems out there and environments.

I agree the K system is very helpful and it just keeps that constant monitor level in the room and yes I totally agree that it can effect how much compression you use on a final master. Less usually. I still find though the best way to keep tabs on your spl levels is through a mike in the room feeding a rack mounted spl meter. Marking your monitor control is great too, but you need a larger more accurate knob in this situation for it to work.

There is a wonderful CD by Ahmad Jamal that was all recorded at 24 bit 96 k without a compressor in sight right to the final CD. You have to turn it up but it is an experience. Ahmad often goes from a whisper to a roar in an instant and it knocks you off your feet every time it happens. This is rare digital experience though. They are doing it now to with people like Brad Mehldau. Sure they are using compressors but they are backing off a bit and the music is sounding just so much more transient.

I'm in a small apartment, so there's no way I can use 85 dB SPL. But it doesn't matter, and the 85 dB is really assuming a mastering type room with far-field speakers. In a small room with near fields you are supposed to lower it to something more appropriate.

The actual level isn't important, it's just that it's set so that overly compressed material is uncomfortably loud and very highly dynamic material is probably a bit uncomfortably too low in the quieter parts and a bit too loud in the loudest parts. If so, then that's a good level in your particular room to mix on, since you'll naturally gravitate towards an appropriate level of compression.

My room is small enough that mid-70's SPL is quite loud. So I have my K-14 scheme set up a mid-70's SPL with the test tone. Anything slipping up over the 0dB K-14 range is getting quite loud in my small room at that SPL (particularly given that I have relatively large near fields as well, but with a heavily treated room of course else I'd be in big trouble.)

Thanks for the detailed explanation about your VU meters Jeff!

best,
mike

85 dB SPL is incredibly loud in a small room with near fields. The THX standard for home theaters is 85dB and no one would set up their home theater in a small apartment for the full 85dB SPL level, not if they don't want to get kicked out. It would be way too loud. I know because I used to have a THX setup, very well calibrated.

I do turn it up louder occasionally to see what it sounds like. But 85 SPL, there's no way I would do that in this apartment regularly. The mid-70s level is plenty loud enough to hear the details in my small room.

Headroom/clipping -- subtract volume levels from each track while playing somewhat loudly.  IIRC, Bob Katz also suggests this.  This is artistic as well.

Superior Drummer 2 / Add-ons -- I stopped using their ridiculous tiny Fx's and compressors.  cw Sonitus and others are better.

Seasoned mixers prefer to start with 2 much headroom and work up.  But artists think positives and negatives, effects and counter-effects, level changes and compliments.

All of us love to share ideas on this and other production techniques.  God bless!

sandyeric,

 You can test your test tones in a wave editor. You can view test tones on an analyzer in SONAR. You can also create your own test tones using any tone generator you have in your software. The test tones at digi doo are accurately labeled... they are what they say they are.

 Here's a link to a premade VU meter package:

http://www.crookwood.com/~crookwood/vu-meters.html

 They seem to have some pertinent info about VU meters and digital systems.

best regards,
mike