I know a lot of you do mastering or have paid for mastering...

  • Thread starter Thread starter foreverdown
  • Start date Start date
Ok , I have one more question.I heard it's easier to get the volume up on an overall track, by running it back to an analog board , and back.Something like Digital's limit is 0 DB and nothing more, but like with Analog it's not really a limit.So in mastering is everything usually kept in digital, or does it go to analog?
 
It depends on (A) who you send it to and (B) the track. I tend to work in analog whenever I can. It just sounds better (IMHO). Still, it's a track-by-track thing.

Analog has "headroom" while digital doesn't. Running through analog might give a louder signal while it's still analog, but it's still going to peg the converters on the way back to digital. Obviously, top-quality digital conversion can handle this in more "style" than lesser converters. But I digress...

And please, PLEASE realize that a song's "optimum maximum level" is decided BEFORE the "RECORD" button is ever pressed... It has a lot more to do with the raw sounds recorded and the arrangement of the piece, the room, mic selection, engineer's talent, etc.

A monkey with an L2 can make a mix "loud" - Getting it to that "optimum maximum level" with it actually still sounding GOOD is why you send it out for mastering in the first place.

Whatever you do, DON'T concentrate on volume during a mix. Don't limit the mix, don't compress the mix for the sake of volume, don't bring the mix's PEAKS above around -3 or -6dBfs. Give the M.E. some room to work. I can normally make a nice, dynamic mix much louder than one that has been "pre-smashed" during mixdown.
 
Okay, another question.Do most mastering engineers charge extra say if you not only want the song mastered but the instrumental also(for release on a single) ?
 
Volume

I have a whole cd mixed, but I am wondering how I can get the volume up without any clipping?
 
You asked about pricing. In general the high-priced spread is about $100 per song. Very reasonable is about $50 per song. Any offer for much less, and I doubt the mastering engineer is the real thing. If you are a pop superstar, they'll quadruple the price, because they know you're loaded.-Richie
 
Massive Master said:
don't bring the mix's PEAKS above around -3 or -6dBfs. Give the M.E. some room to work. I can normally make a nice, dynamic mix much louder than one that has been "pre-smashed" during mixdown.
Click to expand...

John -

I keep seeing you post this. I've been getting some mixes from the folks here on the board that seem to be taking your advice, but honestly I don't see the reason for not getting the mix to peak at around -1, -6dBFS is low in my opinion. One of the first things that I need to do to with mixes like this is raise the level, which causes unnecesssary processing rather than if the mix was at a level which used the full range of digital. I would like to know your reasoning behind this or where you've read about his being the case. -1 dBFS should be enough insurance with reasonable digital gear to ensure that there were no overs due to calibration error. That is not to say that you "have to" reach -1, just that there's no harm in doing so. Aiming for -6 as a peak seems way too conservative and self-defeating.

The real issue IMHO is that mixes don't remove much of the dynamic range of the material in order to work. This is different than the peak using full scale. I like to see the RMS level for mixes coming in at no greater than -18 dBFS.
 
Last edited:
masteringhouse said:
John -

I keep seeing you post this. I've been getting some mixes from the folks here on the board that seem to be taking your advice, but honestly I don't see the reason for not getting the mix to peak at around -1, -6dBFS is low in my opinion.
Click to expand...
Why? The analog noise floor will dictate your dynamic range, not the digital, so it will be the same whether you peak at -1dBfs or -20dBfs. It buys you nothing but certainly runs a higher risk of clipping somewhere...

20 years ago it mattered as we had few converters capable of a full 16 bit dynamic range, but now it's just silly to try to get it close to FS.

If your digital gain degrades the signal, then your software/hardware is broken.
 
bblackwood said:
If your digital gain degrades the signal, then your software/hardware is broken.
Click to expand...

Brad -

I respectfully have to disagree here. ANY processing in digital degrades the signal to some degree (even just level) since additional processing and calculations are done to the audio. This was/is BK's big beef with normalization, otherwise it would be perfectly acceptable to normalize the audio at mix time before runing it through a DAW for mastering (if the signal were not being degraded).

Every calculation adds a little more in the way quantization distortion.
 
My main reasoning is the clipping distortion that occurs in the D-A reconstruction (the swing of the sine wave, blah, blah...) during an analog pass. I had my suspicion about it and then I found the Aldrich papers which put it all into perspective for me - although I barely understand the physics behind it. :o

When I started running my own "unscientific" tests, it was pretty clear (no pun intended) - Even just on a DA/AD loop with no additional processing, the exact same signal with a few dB of headroom came out more clearly - Better highs, more focused mids, the whole 9, if they were brought to the same apparent volume after.

On "smashed" stuff, the waveform during DA reconstruction can exceed 0dBfs by 6dB or more (So, I just use a -6dBfs rule of thumb as it's an easy target and a nice safety net).

On one hand - If something is staying ITB start-to-finish for SURE, this is a little (only a little) less important. In practice, I'd love to see all digital recordings with a peak ceiling around -3dBfs or so (it'll never happen :mad: ). The problem is that ANY hot signal, during mixing if outboard is being used, during an analog pass in mastering, and of course, the final listening by the consumer (on crappy converters, normally) will have reconstruction distortion. The only questions are "how much" and "how will it be handled."

On the other hand - Some converters handle the reconstruction with much more finesse than others - Lavry being my favorite so far. This was the first thing I tried on them. They're my favorite for several reasons, but the full-scale loop test blew away my Apogee stuff... And everything else I had.

On either hand - I see this as a win-win situation for the most part. First, the DA construction is much cleaner. If that isn't enough, I've found that once most people stop worrying about hoarding bits, they concentrate on better sound - whether they realize it or not, they're making better recordings. That's good for everyone.

[EDIT] I see that Brad's been here during my typing - He's probably much more qualified to explain the previous than I am. I also dug up a link to Nika Aldrich's paper explaining it for the most part (I suppose I should've just looked for this before and saved us all a lot of time :D )

http://www.cadenzarecording.com/papers/Digitaldistortion.pdf [/EDIT]
 
Massive Master said:
I've got several on my site... Not "great" ones... Normally I put up files that had strange problems, and how they came out after.

It'd be easy to put files up that already sound good... :D

Well, there's a couple of those also... :rolleyes:
Click to expand...

For your samples on your site to have any basis in reality IMHO you need to level match them. Of course when I hear this loud thing vs this low thing I'll assume the louder thing is better.
 
Massive Master said:
My main reasoning is the clipping distortion that occurs in the D-A reconstruction (the swing of the sine wave, blah, blah...) during an analog pass. I had my suspicion about it and then I found the Aldrich papers which put it all into perspective for me - although I barely understand the physics behind it. :o

http://www.cadenzarecording.com/papers/Digitaldistortion.pdf [/EDIT]
Click to expand...

John -

Thanks for the info. I will digest it further when I have a bit more time.

At first glance the paper seems to be more of an argument against mastering levels going too close to 0 dBFS for consumer CD players. For me this is a different issue than mixing/mastering using equipment with good D/A conversion.
 
masteringhouse said:
Every calculation adds a little more in the way quantization distortion.
Click to expand...
And where does quantization error occur? The LSB...

In a 24 bit word, the real world dictates you won't hear it as that's -140dB or so below FS. Heck, BK even argues that dithering a 24 bit word isn't necessary but he does it to be 'safe'.

Granted, I should have said 'audibly degrades' - but there's a point where advice based on math doesn't matter anymore. Whether -6dBfs or -18dBfs is that point. You can prove it to yourself by taking a mix, dropping the gain by 12 dB and bouncing, then returning it. If you hear a diff, your software is broken, if you don't....
 
masteringhouse said:
At first glance the paper seems to be more of an argument against mastering levels going too close to 0 dBFS for consumer CD players. For me this is a different issue than mixing/mastering using equipment with good D/A conversion.
Click to expand...
Exactly - it's about very high RMS levels causing reconstruction data to go over, but only test tones can generate these high overs. Again, easy to prove - crush semothing beyond belief with your limiter at 96K then SRC it to 44.1k - you may tickle 0dBfs but you certainly won't see +3 (much less the theoretical possibilty of +6 in overs)...
 
bblackwood said:
And where does quantization error occur? The LSB...
Click to expand...

Except when the accumulator carries this bit over to the next bit and it potentially propagates to all of them.

bblackwood said:
In a 24 bit word, the real world dictates you won't hear it as that's -140dB or so below FS. Heck, BK even argues that dithering a 24 bit word isn't necessary but he does it to be 'safe'.

Granted, I should have said 'audibly degrades' - but there's a point where advice based on math doesn't matter anymore. Whether -6dBfs or -18dBfs is that point. You can prove it to yourself by taking a mix, dropping the gain by 12 dB and bouncing, then returning it. If you hear a diff, your software is broken, if you don't....
Click to expand...

I won't argue the above, only to say that quantization distortion is cumulative. We're always trying to reduce this where possible. Granted you won't hear it by dropping a mix by 12 and raising it by 12 once, but try this a couple of times along with any other processing you are doing and I bet you will start to hear a "graininess" that I associate with this type of distortion.
 
masteringhouse said:
I won't argue the above, only to say that quantization distortion is cumulative. We're always trying to reduce this where possible. Granted you won't hear it by dropping a mix by 12 and raising it by 12 once, but try this a couple of times along with any other processing you are doing and I bet you will start to hear a "graininess" that I associate with this type of distortion.
Click to expand...
Umm, but that's the point of the discussion, isn't it? Doing it one time is harmless, so why risk running levels hot if you are only going to risk quantization error that one time at the mastering house? Not to mention that if your mastering guy uses mainly analog processing like me, then it's a moot point...

Just seemed like a strange thing to call John out about when in reality, I'll bet 90% or more of the mastering engineers out there would tell you to be more cautious with levels than try to run it at the rails...
 
bblackwood said:
Just seemed like a strange thing to call John out about when in reality, I'll bet 90% or more of the mastering engineers out there would tell you to be more cautious with levels than try to run it at the rails...
Click to expand...

Hey, I don't want to give the impression that I'm "calling John out", I just see him posting this point a lot and want to understand his logic behind this. In fact I noticed a similar statement on your site as well, and I'm not calling you out either. It's a very good topic and discussing both sides of the coin helps folks understand the issues rather than taking some magic number on blind faith.

Yes, I'm a bit (pun almost intended) old school I guess as far as bit conservation having dealt with 16 bit mixing and mastering as I'm sure that you have. Old habits are hard to break. You can take both sides too far.

Again my reason was that I'm getting these really low level mixes lately. Some way below -6 on peak. This can be as annoying as dealing with the hot stuff since there may be artifacts and distortion in the original mix that goes unheard until the level is brought up and the client hears them for the first time (I'm fighting one of these now which is why I'm so sensistive to this issue at the moment). Once brought up and the warts begin to appear the blame goes to guess who.

Since every 6 db roughly corresponds to a bit, recording peaks at -6 is like recording on a 23 bit recorder instead of 24 bit. No big deal.
 
Last edited:
Massive Master said:
Whatever you do, DON'T concentrate on volume during a mix. Don't limit the mix, don't compress the mix for the sake of volume, don't bring the mix's PEAKS above around -3 or -6dBfs. Give the M.E. some room to work. I can normally make a nice, dynamic mix much louder than one that has been "pre-smashed" during mixdown.
Click to expand...


I wish someone would have told this monkey that before I mixed it down and conviently lost the original tracks on my hard drive :(

Know for another question, what does a guy do when hes already done the don'ts to the mix for instance limiting the thing to 0db with the L2 and wants to still get the songs mastered?
 
I get those in a lot. It can still be done, it normally just limits what can be done.

Such as: Steady things are easy to deal with. Hum, hiss, noise... Once they go through buss compression or hard limiting, their levels are all over the place.

Low overtones wreak havoc on buss compression - The whole mix can be pumping to an overtone that isn't even audible. Once that compression is on there, it's basically impossible to remove.

If the tracks are "safety" limited, it's not *as big* of a deal. If they were squashed or pushed into the limiter, that could be another story.
 
Give him your best mixes on cd, dat, or whatever but make sure its good. If you plan to just sell it to friends and family, master it yourself. If you plan to do something with it, find a mastering lab and ask for the 'economy package' unless you can afford more. Or, you could find most studios that have been around a while that do a decent job for about $35-$50 per song (again the economy master). Make sure your mixes don't vary too much from song to song. Make sure they all song like they belong to the same project (for final mix).
 
You must log in or register to reply here.
Back
Top