I put this in the Coffee House in response to Ben but it may also go un noticed there and others might find some these tips useful.

Reference CD's

I start by always getting the client to supply me with reference CD's. They will always have CD's that they are into and in the style of the music you are about to master. I find them very valuable. On the last album I mastered, I had 3 or 4 ref CD's which was great. I start by listening to them just on my monitors to get a feel for what they after. Not all tracks from a ref CD will fit but there will always be one or two that are right on the money. Same instrumentation, groove, feel etc everything. They will always say I want it to sound like so and so.

Also you can get a measurment of the average rms level they are after as well. Because they are definitely going to say they want it as loud as so and so as well!

During the mastering sessions I rip the most relevent tracks into waves and drag them into the session. I organise it so I can switch them on at any time. While I am getting EQ and the compressor right (not the final limiter) I make sure the ref tracks are playing at exactly the same level as what I am doing in the mastering session.

Signal Chain

I use the LP64 EQ as my main EQ and feed that out to a SMART C2 compressor and then back in for the PSP Xenon for final limiting. I start be EQing the bottom end. I have found that the bottom end on a lot of great mastered tracks is similar. I used Span to find out what it is. It is basically everything down to about 60Hz and a pretty quick roll off after that. I found this out in my car. I have got a killer system in there and the bass is over the top. But even in my car I found on many commercial CD's the bass just kicks nicely and does not rattle your teeth. But on a lot of mixed unmastered tracks the bass rattled my teeth so I decided to dig in and find out why. You have to clean up the low end and after some experimentation I found this steep curve produced the same result in the car. I do low end first. It effects everything that comes after that!

This is where the ref tracks come in now. I start to kick them in and compare the bass and the highs and try to match. Mids obviously need some attention too. I find smooth curves over a wide frequency range work best but with tiny amounts of boost or cut like +1db or so.  The idea is to match the mids to the ref track. It is amazing how much the mids will jump out with only +1db (or less) of boost but over a wide range of frequencies eg from 500Hz to say 3Khz or so. (Big sharp spikes in the mids eg high Q settings you need to be wary of. But saying that Danny did that to one of my tracks and it sounded killer. But the track was less than perfect to begin with and Danny just fixed the offending area really well. So in some cases I say yes to that but overall I would be avoiding it)

A lot of mixes seem to build up energy around 200 to 300Hz. I find a little dip there can make a masive difference. Maybe -3db or so and not over a very wide range either.

Getting things louder

In my previous post I mentioned measuring rms levels of a commercial track. If I am working at say K -14 and obviously a commercial track is going to blast it. So I pull it down by the required amount until the meters are just reading normally again. I might have to pull the track down 7 db which means the ref track is now at K-7 which is pretty loud! I am finding quite a few clients are expecting to get to this level. (unfortunate!)

Now I know that the PSP limiter can get me whopping 7 db in level upwards without stress but I still like to try and steal some gain form elsewhere as well. Like the EQ. If I can lift that up after doing the EQ even by 1 or 2 db then it means less work for the limiter. Sometimes the unmastered track is below K -14 and some gain has to be made up here. If I am sending a signal out to the compressor it is a good place to add some gain (makeup gain) in the analog world. But if you all ITB then there are still places you can do it. The track the un mastered track is on for a start can be lifted in level. The INPUT settings going into and OUTPUT settings coming out of the LP 64 is another spot. If you are using a ITB compressor makeup gain there of course as well as the limiter can add gain to the signal going in.

The compressor I don't drive hard so much. I like to go for about 3 db of gain reduction. The attack settings are important here. Slow it down so it does not sound like it is jumping all over the music. eg 10 ms or so. (even 30 ms on the C2) The compressor should just create this lovely evening out of the mix type effect. I use the makeup gain and maybe get another db of gain out of it. That means the PSP Xenon only has to do maybe +4 or +5 db of loudness upward shift. There are quite a few controls on a PSP Xenon and that is why I like it so much. No other limiter gives you as much control over how the limiter is actually sounding like this.

I do all this at 24 bit as well. The PSP also does the dithering down right at the last minute to 16 bit. I just print the track now (even as a high speed bounce) and do final things in Adobe Audition. Studio One has got a great mastering page where you can set up all these things over all the tracks. I have to print the C2 tracks though because they are outboard obviously. If you were all ITB you would not even have to do that.

I keep switching the ref tracks in and out all the time to make sure. Mastering is also about getting all the tracks to work together as a group and you may need to fine tune the odd track here and there. When I am using the limiter I re calibrate my K system levels to K-7 so the VU's are now showing the final levels for the mastered tracks and the ref tracks. That way I can compare easily and see if I am still falling short or getting too loud.

Quieter Tracks

Tracks that only have a vocal and guitar or vocal and piano will sound louder than the big kickass tracks. (even though both are reading the same on the VU) I find dropping them by 2 to 3 db usually brings them in line with the others.  Don't try to make the kickass tracks as loud as the quiet ones, I go the other way around. I lower the solo tracks to match the kickass tracks.

Final testing

Then I go for a long drive in the car and check the whole thing out! Or have a dinner party and play the CD in the background. Any little thing will usually stand out a mile for some reason even when you are paying no attention to it. There will usually be some changes to some tracks. I also give the CD to the client and let them listen to it for about a week and they often come back with just minor changes but important ones none the less.

I will do more as I think of it. I don't try to change the sound of the mix very much at all but rather enhance it. It is a big mistake to completely change the sound of a mix. Unless it is terrible and needs some serious operations but in those cases you should send them back and say remix them or don't take the job on if it sounds terrible. You will never get it right no matter what you do.

Timely thread. I was fighting a couple of versions of a mix tonight and some experimental mastering approaches. The 300Hz dip was required to keep distortion out of the final. I still could not get the RMS to peak values of commercial tracks but as you point out above, an external compressor can help close in on the gap.

Nice post Jeff
Thanks

Jeff: Great read and info for those curious about some of the things that go on with mastering. There are a few other things I'd like to share with you also that you may want to try.

You mentioned the 60 hz thing. It may appear that way, but it's not really the case at all. If you listen to the new rock today or even hip hop or rap, you can definitely hear frequencies that can go down to 40 hz on some things. You may not see them register on your scope, but they come by way of sub harmonics that may have a tight Q so that the material has a tighter low end. A good way to control this stuff is with a multiband compressor. The UAD precision comp is unreal at this. As a matter of fact, it's so transparent I have since retired my API 2500 from my mastering chain in certain situations. The cool thing about the UAD MB is, it works like a cop on patrol. It only kicks in when it needs to and doesn't "work all the time".

For example, I may want a little 50 Hz rumble to come through on a kick drum. I'll carve it up with a tight Q just to get a bit of that sub harmonic going on and then tighten it up so it's not flabby or rumbly by setting my MB to control it. It will only kick in when it gets past a certain threshold and you won't even hear it. This is also good on bass guitar notes that may be due to a 5 string or 6 string bass. Those low notes will jump out at different times to where you either have to automate an eq on that section or just run the MB instead and it will patrol the border for us. :) Of course sometimes the API is used along with the MB to where I can use less of the MB...but it depends on the material. But I just about always like to have the MB on patrol because it just keeps things tighter and allows you to let certain frequencies you'd never normally allow, to come through in moderation with full control. :)

Middleman: the reason for this is due to peaks. If peaks aren't controlled, that's where your mix is going to cap level wise. The one thing with making things loud is to control the mix as well as peaks and NEVER over-work your compressor or limiter to do this for you. This is where everyone fails when they try to get commercial levels. The key to being loud is to have a mix that's tight without peaks all over the place.

For example, we can use a compressor to control peaks and a limiter to finally handle everything. That's still going to give you a "cap" before you notice artifacts. If you're going for commercial levels, you have to manaully level the audio. I learned this trick from Bob Katz but took it in a slightly different and more "anal" direction.

Once you control those peaks in full, the mix not only sounds better, but it LOOKS better too and believe it or not, there are numbers we use that describe a mix. The better the mix sounds and looks, the better these numbers look as well. Quick story for you...

I did a master for a studio a few weeks ago. The studio and myself were both happy with how everything sounded. Somehow, something got screwed up on the studio's end and the client was unhappy and went over the studio's head and came to me. They wanted to know if I actually mastered the last file they had received. When I told them "yes" I was asked "when"?

The mastered file *I* did was done in early March...they had just received this other file a few days ago and they felt it definitely wasn't my work because I had done work for them before through this studio. When I got the file that was supposedly mastered by me, sure enough that wasn't the one I did. It didn't sound right or look right. And, the "numbers" I'm talking about were not MY numbers. I have my own thing with min/max samples, RMS, DC offsets, audible noise in beginnings and ends...all my files look just about identical all across the board. The numbers thing (though sometimes meaningless if something sounds terrible) is sort of like my trademark or fingerprint so to speak on my work. You'll never see a DC offset and my min/max sample numbers will always be exactly the same on both sides. These numbers of course change due to how loud a client may want a master to be. Like at -0dB we'll get somewhere along the lines of a 32764/65 on min and max samples (32768 means you've clipped somewhere and it's hard to get 32767 on all ends unless you're Bob freakin' Ludwig) and at -0.1 we'll hit the area of about 32400 min/max samples at 16/44.
 
But those numbers can tell a story when they don't match and you can tell when someone relied too heavily on compression or limiting instead of manually taking care of the audio. Making those numbers all line up to where the audio looks good and sounds good is no easy task. It's taken me years to get them all close like that. But like I say, they usually speak for themselves because when you have a good looking wave file that sounds good, just about always the numbers behind the scenes look good too. :) There's an entire process I do to get things like that. The editing alone per song can take me an hour, but it's worth it because the end result makes an incredible difference and in my opinion, it's the right way to do it.

I recently hired another engineer that was really good. He's been at it nearly as long as me and has mastered and engineered for some pretty credible artists. He's been sitting in on my sessions recently and during the process he just keeps on saying "dude, you're a lunatic with this stuff...I can't believe the attention to detail you give this stuff as well as how precise you are with every little thing!"

But these are the differences between a dude that likes to polish audio and call himself an ME and a guy that really IS an ME. Anyone can try to do what my friend Tony calls "little m mastering". You're not going to compete with the commercial music out there. Don't even waste your time. Ozone, all those plug suites etc...not to bash on them, they work for people that are just messing around...but trust me, it's not the right way to do it. There are so many things to consider when mastering that need to be done BEFORE you even physically master the material.

I go through 3 phases here. I'm currently working on the new Wallow Sound CD from our good friend ChuckC on this forum. Once he sends me the final mixes, they need to be edited from the ground up. I have to listen to each song in headphones to listen for pops, crackles, noises, oscillations, clicks, and any other nuance that may degrade the audio. While I'm doing this, I'm manually leveling the audio by hand or with automation and adjusting peaks and writing down every change I make in note pad as well as the times in which these changes were made. Every little noise, click, pop, peak or "thing" that gets adjusted, gets logged to the numbers as well as what tool and setting was used to remedy the problem. Every plug used gets saved with name of song and the client name as well. 

This way, if I have to go back and remaster something from the ground up, I always have my little cheat sheet for faster editing and knowing exactly where to look for things and what was used to fix them. When this is done (which takes anywhere from 30 minutes to an hour depending on how much needs to be done) it gets saved as "name of song Pm" which stands for "pre master" and will come out to be a nice -3dB without extreme peaks. From there the file is ready for the mastering process. When I add all my processing and am happy with how it sounds, it gets saved as "name of song master <insert bit/sample rate>"

From there it goes into a "suite" where limiting is applied, dither, sample rates are converted and final DC offsets are removed. From there, dead space in beginnings and ends are removed and levels at start and end are adjusted to infinity so there is no hiss or noise to begin the tracks and there's nothing audible in the ends.

Fades are done at this point so that my limiter wasn't tripped off. Always do those last after everything is done and converted. The last thing you want is a compressor or a limiter trying to adjust your audio on a fade in or fade out. When all this is done, there is another listening session in headphones to make sure all pops and clicks etc have been taken care of as well as through 4 different monitor systems. If they pass these tests, all files are saved as "name of song_DZL_Master16_44". DZL stands for DanziLand and is my final file name that tells me everything has been done. That name doesn't go on a tune unless it's been totally worked and tested.

So this is along the lines of how I do things here. I feel I can compete every bit with commercial recordings level wise and can get even better sound with a slightly lower volume. People have to remember...when you master something the right way, it can be a little lower because once you turn it up, it goes up and up and up until you max it out. With a super loud master, it caps off at about 5 on your volume knob and doesn't get any louder as you go up. It gets more distorted and sounds terrible. By the time you hit 7 or 8 on a good master, you can't breathe because it's so loud and clean and it will walk right on by the loud master. :)

-Danny

Danny, how are you adjusting the DC offset for alignment?

OK, well, I can already tell this thread is going to get bookmarked under my "favorites" browser tab.

Good points Danny about the low end. I must ensure that what hapens below 60 Hz as well can take on different filter slopes etc. It is more about still hearing 40 Hz all right but just down at a more suitable level. I think you can do a lot by just shaping the low end a bit to remove a lot of over deep rumble etc. Even before you hit the multiband comnpressor. Then I agree that a multiband can be great for controlling low end.

I am a bit into getting the overall frequency response right before going into a compressor as opposed to hitting a multiband compressor harder with bottom end and expecting it to take care of it. Why not eliminate it before it even gets there. Then I think in some ways the compressor is reacting to other things in the music and it will have a different effect.

One of your uses for the LP64 is for bottom end duties and I think that is wise as well. I find the more you get your response right before the compressor the less I feel the need to use a multiband and the full range compressor can work really well such as the C2.

I love this forum.

I'm in the same camp as you, Jeff. I don't turn on a compressor or a multi-band until I get the eq curve where I want to be. But once I get it there, I can add a few additional sub harms and then just control them with the MB which really works well on bass drops and other low end stuff that we don't really want to come in as "rumbly". :)

Middleman: I use either Wave Lab and Adobe Audition as both of them remove DC's completely without messing with anything else. I use Adobe most of the time for editing and will remove the DC's there first. It also allows for DC adjustment for individual sides. I'll tell ya man, people laugh at me for using that program, but honest when I tell you, it is one powerful editor that is loaded with options I can't seem to find anywhere else. The only thing close to it is Rx 2 Advanced...which seems to be modeled from Adobe's architecture in my opinion.

Then, once all the processing goes on and I use my limiter, I'll pick up maybe like .03% on both sides (which still isn't bad) and WaveLab completely eliminates it from there.

-Danny

Thanks. Good tip.