Mastering; the do's, don'ts, why's and why not's

[lance]

I'm starting this thread in hopes of continuing the excellent discussion started in the following thread:

http://www.chrismilne.com/uadforums/viewtopic.php?t=7691&postdays=0&postorder=asc&start=15

And I hope to move into more specific techniques and practices involved in mastering.

First I must clarify that I agree with the conventional wisdom that most mastering projects should be sent out to a professional mastering facility. And, with the project I am currently starting, I plan to do just that. I do, however, believe that in some cases mastering your own project is the only feasable or desirable option. I think that with a good understanding of the limitations and pitfalls and a problem solving approach to these obsticles that success is possible.

But, where does one start?

Lance

 

[Chi-Squared Mastering]

But, where does one start?

With being able to hear what you're working with. The most important pieces of a mastering setup are the accuracy of the engineer's ears, speakers, and room. If you're not 100% sure about all three of those items it's time to measure 'em and fix 'em.

Reading material commonly suggested is Bob Katz' book Mastering Audio: The Art and the Science. It doesn't exactly contain a step-by-step walkthrough of how to master, but it does cover the reasons behind it all. What and why are a big part of that book, and it's definitely worth reading no matter what role you play in music production.

After that point, engineering is engineering. Once you know your monitoring chain and the ins and outs of audio, it's a matter of practice makes perfect.

You mentioned the limitations and pitfalls of mastering your own material. I'm sure others will chime in with more, but I think the hardest part about doing that is how intimately you already know the material. By the time you practice, play, record, do retakes, mix, maybe remix..... you know every last thing about that song. The big problem with that is you won't be able to identify what's wrong with it anymore. Maybe the hats are too loud, but you've heard it so many times you're used to it. Maybe the backing vocals are way too quiet, but you've heard them so many times you know they're there. Objectivity becomes hard to find when you're that familiar with the track.

If you don't get "fresh ears" by sending your material out, you might be able to overcome the above by shelving the song for a week or so before attempting to master it. Don't listen to it, don't play it, don't sing it. Work on something else for a week. Then, when you reopen the song, suddenly you wonder where those background singers went, and why it sounds like someone's smacking you in the eardrum with a hihat.

 

[Dan Duskin]

Yup... the room, the monitors, the room-treatment, the D/A conversion, and the cables. i.e., unless your room is close to perfection, you cannot do a good mastering job. In fact, this is one of the main reasons we send stuff out to mastering houses... because they can hear it better than we can!

As for do's and do not's... all I can think of right now is:
- Use only as little EQ, Compression, a.k.a 'processing' as needed.
- Always compare every change you make to the before (A/B).
- Their is no \"one set of processes\" that works (every mix is different).
- Always dither LAST.

 

[Richard Hunter]

one question about dithering last:

if, for home brew mastering, do i dither before applying fades? adding noise to a track after fading seems counterintuitive. why is it done this way? wouldn't we want the dither noise to fade out gracefully along with the rest of the track?


second, i agree about the fresh ears thing. if you plan on distributing your music, get an ME. At the very least invite another engineer who hasn't heard the material to come to your studio and critique your mix.

i work with a band, the lead singer has great voice, but the style of music is such that his voice is often a little buried, and he slurs words together (like so many rock acts). The melody is instantly recognizable but the words are hard to understand, at least im told. since ive heard the songs SO many times, i know all the lyrics by heart, and it would never occur to me that the lyrics might be difficult to pick out. (+1 for liner notes w/lyrics) these sorts of things, as far as the rest of the mix goes, is what an ME can help you with.

 

[imdrecordings]

\"Richard Hunter\"one question about dithering last:

if, for home brew mastering, do i dither before applying fades? adding noise to a track after fading seems counterintuitive. why is it done this way? wouldn't we want the dither noise to fade out gracefully along with the rest of the track?

If you are working with a 32bit or 64bit float application and are bouncing to a 24bit track. Then yes, you would have to apply a 24bit dither.
The way to avoid that is by simply bouncing the files as a 32bit or 64bit file, depending on the application. No dither is needed.
Any time I bounce in C4 or SAM V9, I bounce the track as a 32bit file. No need for dithering the cross fades.
Double dithering is not a bad thing and in fact is necessary for communictaing or bouncing between lower bit rate devices. But it's not a good thing either.

second, i agree about the fresh ears thing. if you plan on distributing your music, get an ME. At the very least invite another engineer who hasn't heard the material to come to your studio and critique your mix.

All this talk about mastering and people second guessing whether or not they should have their material professionally mastered. Seems to be coming from people who have never had their material Mastered by an ME. I was skeptical at first, back in the day, but took a chance. My skeptisicm made me pay a little extra to sit in on the mastering session and I earned a huge appriciatation and understanding to what it is they do. I urge anyone who has never done so, to ask if they could sit quielty in on the mastering session and observe. Even if it's not their own. Just call and ask, explain why.

I work with a band, the lead singer has great voice, but the style of music is such that his voice is often a little buried, and he slurs words together (like so many rock acts). The melody is instantly recognizable but the words are hard to understand, at least im told. since ive heard the songs SO many times, i know all the lyrics by heart, and it would never occur to me that the lyrics might be difficult to pick out. (+1 for liner notes w/lyrics) these sorts of things, as far as the rest of the mix goes, is what an ME can help you with.

I don't know if an ME can help you with that. But I guess buy cleaning up the bass and lower mid, they are adding more deffinition to the vocal... :?
And dude! It's rock.... your not supposed to understand the words. It's the feeling they give off. The words are meant to be interpreted and learned later.

 

[lance]

All this talk about mastering and people second guessing whether or not they should have their material professionally mastered. Seems to be coming from people who have never had their material Mastered by an ME. I was skeptical at first, back in the day, but took a chance. My skeptisicm made me pay a little extra to sit in on the mastering session and I earned a huge appriciatation and understanding to what it is they do. I urge anyone who has never done so, to ask if they could sit quielty in on the mastering session and observe. Even if it's not their own. Just call and ask, explain why.

I agree with you 90%. Myself, I have a great respect for Mastering Engineers and the mastering process. I had my first record mastered and this next one will be as well. Personally I believe that's what makes a demo an album. What I have an issue with is the all too common discouragement of the learning process. What's wrong with someone wanting to learn how to master? For all we know, the guy asking what plugins will give him that \"mastered\" sound, might jump into it with both feet, learn the right way, set up a great room with excellent gear and become the next generation's \"Bob Katz.\" With the right guidance of course. Or he may go on to crush the life out of every mix he touches but who cares we don't have to listen to it.

 

[Dan Duskin]

Correction...
At any point in which you convert from a higher bit-rate to a lower bit-rate, dither last (just before converting).

In regards to fade outs and fade ins. You should always dither AFTER fades, and never before! Dithering before fades will cause the dither to disappear and quantization+stepping artifacts will become audible. So once again, always dither last (just before conversion).

 

[boody]

But, where does one start?

Lance

That's easy; you go out of the studio where you mixed your project and into the real world of different audio systems to check your mix. In the car. In the living room. On the I-pod. On all your friends installations. On crappy systems. On audiophile systems. The more the better. Look for common problems, like bass response, brightness, depth, focus, widthness, boxiness, dynamics, air, separation, space. And always listen to a great sounding record that's comparable and that you know very well on the same systems; this is your reference. Write down any problem you find. Compare your findings and look for patterns.

Then when mastering: you will want to change as little as possible. Only do what you need to do. If you need heavy corrections, better redo your mix. If the problem lies in one or two instruments, redo your mix. If there is anything you can't solve with compressing a little transients, eqing a little for balance, roll off the sub a bit, maybe give it some air (16khz+); redo your mix. You have that option, so better attack the problems there where they should be defeated; in your mix.

Then, when the master is done; go out again and repeat the realworld test. This is the reason you want mastering: universal audiosystem compatability.

my 2cts of course
Budy

 

[Richard Hunter]

Correction...

In regards to fade outs and fade ins. You should always dither AFTER fades, and never before! Dithering before fades will cause the dither to disappear and quantization+stepping artifacts will become audible. So once again, always dither last (just before conversion).

sure, i know this and this makes sense, but won't the dither noise now be audible as a track fades out? i guess i'm asking if it is, and if not, why not, normal for me's to fade a track out (at the end), dither, and then reapply a fade @ the end in order to have the dither noise fade out with the rest of the track.

The way to avoid that is by simply bouncing the files as a 32bit or 64bit file, depending on the application. No dither is needed.
Any time I bounce in C4 or SAM V9, I bounce the track as a 32bit file. No need for dithering the cross fades.
Double dithering is not a bad thing and in fact is necessary for communictaing or bouncing between lower bit rate devices. But it's not a good thing either.

I work in DP, so no 32 bit bounces for me. DP automatically dithers all audio going from the 32 bit floating point domain to the 24 bit fixed domain. I was asking specifically about fade outs at the end of tracks, 24-bit to 16 bit.

And dude! It's rock.... your not supposed to understand the words. It's the feeling they give off. The words are meant to be interpreted and learned later.

you don't have to tell me that, i say bury bury bury that vocal most of the time, for rock anyway. my point was just to show how differently a fresh set of ears will hear a track ive heard a hundred times.

 

[Plec]

sure, i know this and this makes sense, but won't the dither noise now be audible as a track fades out? i guess i'm asking if it is, and if not, why not, normal for me's to fade a track out (at the end), dither, and then reapply a fade @ the end in order to have the dither noise fade out with the rest of the track.

I don't really follow here... is the dither noise too loud for you so it disturbs the sound? The point is that it should help the sound...

Fading out again after you've dithered down to 16bits really takes away the whole point of dithering a fade. You should be able to hear a not so small difference between the first fade you did and the second one which you truncated.

I can do it for you if you like... I have the best fades in the biz, only 500$!

 

[Richard Hunter]

[

I can do it for you if you like... I have the best fades in the biz, only 500$!

yea ok, and maybe ill let you digitize my vinyl collection for my ipod sometime. jeez

man you are really busting my balls lately.

& for the record, no, dither noise is not too bothersome to me, it just seems counterintuitive to add signal to a digital 0. i understand it makes the fade sound smoother, but when the fade ends, you aren't left with digital silence. i was curious as to why this is acceptable. perhaps someone who knows can answer this for me.

 

[Chi-Squared Mastering]

for the record, no, dither noise is not too bothersome to me, it just seems counterintuitive to add signal to a digital 0. i understand it makes the fade sound smoother, but when the fade ends, you aren't left with digital silence. i was curious as to why this is acceptable. perhaps someone who knows can answer this for me.

The goal isn't to get digital silence, the goal is to have more accurate audio.

Pretend the computer is using a "zero to ten" scale for volume. Song comes to an end, you fade out... The volume goes to 5, 4, 3, 2, 1. Now what? There's still supposed to be more to fade, but 0 is pitch black silence. To lower the volume past 1, or even in between any two other numbers, the computer needs decimal places.

Anytime you do ANYthing to digital audio, the computer needs more "decimal places" than it started with. If you fade after dithering, you're introducing additional small numbers that the computer can't count properly. That leads to what's called "quantization distortion", when the computer is forced to truncate or round those small numbers.

Dither helps the computer deal with those very small numbers, which translates to increased fidelity and resolution. Instead of the 96 dB range defined by 16-bit audio, it sounds more like we can hear down to -108 or so because dither noise is helping the computer represent those very small numbers accurately.

 

[Richard Hunter]

[quote="Richard Hunter":3lewfe3j]
for the record, no, dither noise is not too bothersome to me, it just seems counterintuitive to add signal to a digital 0. i understand it makes the fade sound smoother, but when the fade ends, you aren't left with digital silence. i was curious as to why this is acceptable. perhaps someone who knows can answer this for me.

The goal isn't to get digital silence, the goal is to have more accurate audio.

Pretend the computer is using a "zero to ten" scale for volume. Song comes to an end, you fade out... The volume goes to 5, 4, 3, 2, 1. Now what? There's still supposed to be more to fade, but 0 is pitch black silence. To lower the volume past 1, or even in between any two other numbers, the computer needs decimal places.

Anytime you do ANYthing to digital audio, the computer needs more "decimal places" than it started with. If you fade after dithering, you're introducing additional small numbers that the computer can't count properly. That leads to what's called "quantization distortion", when the computer is forced to truncate or round those small numbers.

Dither helps the computer deal with those very small numbers, which translates to increased fidelity and resolution. Instead of the 96 dB range defined by 16-bit audio, it sounds more like we can hear down to -108 or so because dither noise is helping the computer represent those very small numbers accurately.[/quote:3lewfe3j]

thanks for your response chi, i understand what dither is for and how it is used. however, once a fade is applied to a track, pre-dither, you get, using your analogy, values of say, 5,4,3,2,1,0......0

the fade goes to zero. when @ 0, there is no original signal present in the file. then you add dither noise, which makes the 5,4,3,2,1 sound smoother, but you dont end at 0. you can't if you are adding signal. or can you? It would be more like 5,4,3,2,1,0.01...with the harsh steps between the integers mollified by the dither.

i imagine what you end up with, the dither noise at the end of your fade is inconsequential, especially considering the inherent noise floor of most cd players. that's basically my question: do you as mastering engineers care about this?, but none of the proclaimed ME's here seemed to have been able to answer the question, let alone attempt to understand my question. in fact it's been met with more condescension than anything else. maybe i was just not being clear enough.

so nobody cares about the little bit of noise you add to a 100% faded out track?

wait, don't answer that, im done caring.

 

[Dan Duskin]

Their is no need for complete digital silence... that's my two cents. The dither is so low in volume, plus the fact that it's actually a pleasing white noise to listen to.

HOWEVER, you are not the first person to ponder this question! Apogee has a solution to the 'problem'.... it's called \"Auto Black\". It just means that the dither shuts off if complete digital silence is reached. You can find this in the UV-22 dither. I'm not sure if any other dithering tools have this or not.

Personally, I don't use auto black... i like to let that pleasing white noise (like a very soft tape hiss) just keep on rolling.

 

[TheEastGateMS]

Personally, I don't use auto black... i like to let that pleasing white noise (like a very soft tape hiss) just keep on rolling.

VERY soft.

I do the same.

I mean, c'mon, when placed properly, you shouldn't even hear it without turning things up a lot.

 

[Chi-Squared Mastering]

that's basically my question: do you as mastering engineers care about this?, but none of the proclaimed ME's here seemed to have been able to answer the question, let alone attempt to understand my question. in fact it's been met with more condescension than anything else. maybe i was just not being clear enough.

so nobody cares about the little bit of noise you add to a 100% faded out track?

wait, don't answer that, im done caring.

I guess you may not see this if you're done caring, but sorry if anything I posted sounded like condescension. I thought I WAS answering the question, and tried to be clear, not patronizing. If the question is just do we care that we can't reach digital silence, the answer is no. Hope that helps.