pre-mixing mastering?

[freeztar]

Does anyone master individual tracks before mixing, and then master the song? I am a newbie in recording/mixing, and I usually master each track before marrying. Is this commmon practice??? It seems commonplace to me, but who knows? I usually set every track at max vol. and compress/NR and then mix each into one song that I additionally compress/NR. Is this commonplace?

 

[Chibi Nappa]

I would say no. My advise would be to not do any normalizing etc. to a track before mixing. It doesn't give you any benefits (why make it louder if you are just going to adjust the volume of it later on?) and it could hurt your sound quality due to extra processing. As far as compressing/eq'ing a track before mixing: I would avoid that as well. Compression and EQ settings don't mean much to a track outside of the context of the other tracks. You have to see how they all work together. Save that stuff for mixing, not before.

And if you record everything properly, you shouldn't have to do any noise reduction durring any phase of song production. You can seriously damage your high end with noise reduction.

As for compressing a song after it is mixed: yes, that is commonly done. However, unless you have some seriously good ears and experiance, you may very well make things worse. After you do the final compression, give it a good listen in comparison to the mix of the song before compression. Don't be afraid to go with the orriginal uncompressed mix if it sounds better. If you are sending the song off to someone else for mastering, I wouldn't do any compression at all after mixing.

 

[TexRoadkill]

Good advice from Chibi.

In general never process a track while soloed. It's how it sounds in the context of the mix that matters.

 

[Fishmed]

Does anyone master individual tracks before mixing, and then master the song? I am a newbie in recording/mixing, and I usually master each track before marrying. Is this commmon practice??? It seems commonplace to me, but who knows? I usually set every track at max vol. and compress/NR and then mix each into one song that I additionally compress/NR. Is this commonplace?

It does seem to be an uncommon practice. I do "Clean" tracks before mixing. From experience and my technique, I know what tasks need to be done before I mix. Many of my clients are not consistent musicians, so there are times I have to smooth out their performances. I do normalize, remove any severe peeks, and sometimes mute any stray noises. I have found that smoothing the tracks before mixing, the numbers of severe peaks are greatly reduced on the final mix. I eq and compress during the mixing process.

Since my clients have low budgets, I have to do minor mastering (mainly making the songs louder and consistent with each other.) I would love to hear what a Professional Mastering Expert would have to say about my mixes. Knowing my flaws would allow me to improve my technique.

 

[Chibi Nappa]

Re: Re: pre-mixing mastering?

I do "Clean" tracks before mixing. Many of my clients are not consistent musicians, so there are times I have to smooth out their performances. I do normalize, remove any severe peeks, and sometimes mute any stray noises.

"Cleaning" tracks is one thing.....Moving a severely missed bass guitar pluck, cutting out the "noise" (stick clicks, anyone who talks right before they play, etc) from the beginning and ends of tracks, replacing a screwed up guitar part on the 2nd verse from take 3 with a correctly played part from take 1.....stuff like that. That falls more under "editing". I frequently do that before mixing, but I don't process the sound itself. I just move it around and cut/copy it.

And again with the normalizing before mixing.... Why would you do that? What do you hope to gain from it? It does nothing that moving a fader durring mixing wouldn't do. I say the fewer times you induce digital math errors from signal processing the better, so just skip it.

 

[Fishmed]

Re: Re: Re: pre-mixing mastering?

"And again with the normalizing before mixing.... Why would you do that? What do you hope to gain from it? It does nothing that moving a fader durring mixing wouldn't do. I say the fewer times you induce digital math errors from signal processing the better, so just skip it.

I have never noticed any negative effects from normalizing tracks. By increasing the levels, I have more headroom on the board to work with. What programs have you used that messed up your track? I would try to record louder signals, but the musicians are inconsistent and I have to prevent digital clips. Increasing the volume of the tracks afterwards keeps the line level noise to a minimum during mix down.

 

[Chibi Nappa]

It is not a huge effect. You would probably have to A/B something with the orriginal to hear any effect at all, but it can hurt you anyway.

And I would think that normalizing would give you less headroom and more noise. Less headroom because you are making everything louder, and more noise because you are raising the level of the recorded noise on the track along with everything else. Not to mention that the normalization process itself probably adds noise or at least some kind of artifact. I'm not talking "oh my god, what is that horrible sound" noise. I mean really subtle stuff, but it is the subtle stuff that adds up from beginning to end that can hurt a project.

And again, unless you are doing some sort of "compressing" normalization, it doesn't do anything that moving the fader doesn't do. If normalization makes everything on a track louder by exactly 3 db, why not just move the fader up 3 db instead? What if you normalize a track and make it 3 db louder, but then at mixing time you find it needs to be 3 db quieter so you move the fader down? You just sent your signal through two calculation processes that induced math errors and you ended up right back were you started before normalization.

 

[Blue Bear Sound]

I agree with Chibi 200%........ the minute you change the signal in any way - round-off error occurs, and degrades the signal ever so slightly....

Do it several times and you likely won't notice, but think about how much DSP (digital signal processing) you apply to each track in the course of a project - and don't forget, ANY change (ie, EQ, faders, a low--pass filter) to the signal will cause a certain amount of round-off error.

Though not as bad as analog generation loss, this digital generation loss can easily creep up on you to after several dozen DSPs.......

 

[barefoot]

I just discover that there is a name for the mastering technique I use (and discovered independently). I master from "stems". First I mix down to a few stereo stem tracks - i.e. drums, bass, guitars (synths in my case), and vocals. I lightly "master" the stems before mixing them down to a single stereo track. Then I take another, usually harder, pass at the final mix down.

I recently learned that mastering engineers often take a couple passes at processing a track - especially when it comes to limiting. I like to take my first pass in the "stem" stage.

However, I think you might need pretty good tools for this technique. I use the Waves Masters bundle with 48 bit double precision processing.

Thomas

http://barefootsound.com

 

[Fishmed]

I agree with Chibi 200%........ the minute you change the signal in any way - round-off error occurs, and degrades the signal ever so slightly....

Do it several times and you likely won't notice, but think about how much DSP (digital signal processing) you apply to each track in the course of a project - and don't forget, ANY change (ie, EQ, faders, a low--pass filter) to the signal will cause a certain amount of round-off error.

Though not as bad as analog generation loss, this digital generation loss can easily creep up on you to after several dozen DSPs.......

I don't do much Digital DSP work to the tracks. I'll do a little editing then that is it. My compression, eq, fadder, etc. happen with outboard gear and my board. I don't mind coming out of the digital world because it warms the mix up.

 

[sonicpaint]

Back to basics.

I've heard of editing but mastering the tracks then mixing down? I can't say that I heard of that, though like suggested before not a good idea.

Remember "less is more". Only do what has to be done. "If it's not broke don't fix it" kind of thing.

sonicpaint

 

[Fishmed]

Re: Back to basics.

I've heard of editing but mastering the tracks then mixing down? I can't say that I heard of that, though like suggested before not a good idea.

Remember "less is more". Only do what has to be done. "If it's not broke don't fix it" kind of thing.

sonicpaint

If refered to me, I am by no means mastering the individual tracks, just editing. I do believe my edit process reduces the work during the mastering process.

 

[sonicpaint]

Not referring to you fishmed.

Sorry if you misunderstood but I was referring to Freeztar not you Fishmed.

sonicpaint

 

[freeztar]

Ok, but what about NR before mixing? I know, as stated before, that I shouldn't have too. Well, there shouldn't have to be people starving in Africa, but there are and I have to use NR because I can't afford better mics and mixers. I've found that when I NR and normalize (manually edit peaks and valleys), I have a much better sounding track to mix with. So am I wrong to do any of this, or is there a better way to "edit" my tracks to clean (not noisy/hissy) before mixing. I know I destroy some highs and give everything a "muffled" sound with NR, but right now I don't know a better way (I'd prefer to have a clean track, albeit a little muddy, than to have a hissy/noisy track). Any help would be appreciated (I'm a humble newbie---"back to basics").
Keith

 

[Chibi Nappa]

I've found that when I NR and normalize (manually edit peaks and valleys), I have a much better sounding track to mix with.

Ah ha. Looks like we had a communication problem earlier. So, when you say "normalizing" you mean manually knocking down peaks? That's where the confusion was then. That's not normalizing. Normalizing is when you take the amplitude value of your highest peak, and divide every other sample by that number forcing every sample to have a value between 0 and 1. Then you multiply every sample by a number that will bring the highest peak to a set value (usually the highest it can possibly be, or 1 or 2 db lower than that).

 

[Fishmed]

Ah ha. Looks like we had a communication problem earlier. So, when you say "normalizing" you mean manually knocking down peaks? That's where the confusion was then. That's not normalizing. Normalizing is when you take the amplitude value of your highest peak, and divide every other sample by that number forcing every sample to have a value between 0 and 1. Then you multiply every sample by a number that will bring the highest peak to a set value (usually the highest it can possibly be, or 1 or 2 db lower than that).

If that is the definition, I do not technically normalize either. I will bring down peaks that are above a certain threshold resample them down so that its peak matches the threshold. I will sample only from zero to zero, so this is a split second of audio. Really it is so little time that you cannot even tell that it has been done, but it is these little peaks that can keep you from reaching greater volumes with the track and it reduces the need of compression on many things. If you have too low of a threshold, then your track will loose its dynamics. My resampling these small clips, you real maintain the dynamics of the track. Once I do this task, I boost the track so that the hottest peak is -1 db.

 

[Chibi Nappa]

Once I do this task, I boost the track so that the hottest peak is -1 db.

Right there. That very last thing you said (boosting the track to -1) is normalizing. Try mixing without doing that last step and see if the results are better.

 

[Fishmed]

Right there. That very last thing you said (boosting the track to -1) is normalizing. Try mixing without doing that last step and see if the results are better.

For grins, let's see the senerios:

The ideal situation would be to track as hot as you could without clipping; therefore you would have mucho headroom and a low noise floor.

The other senerios:

1) Track hot as posible with a clip here or there, have mucho headroom and a low noise floor.

2) Track hot as posible with retakes with each clip, have mucho headroom and a low noise floor.

3) Track low enough signal to avoid clipping (and retakes), but have less headroom and a higher noise floor.

4) Track low enough signal to avoid clipping (and retakes), normalize for mucho headroom and a low noise floor.

which of these last four situations is the best?

 

[Chibi Nappa]

I think we are running into more communication problems here. What do you mean when you say "headroom"? If you take headroom by its traditional meaning (the amount of "space" you have before a signal or mix clips), then a hotter signal gives less headroom.

And as for the noise floor issue: If you have a clean signal and record it with the meters well below clipping, you will not have a noisy recording. If you have a noisy signal and record it well below clipping, but then raise the whole thing with normalization, you are raising the recorded noise level by the same amount as you are raising the actual music. Normalization will not clean up a noisy signal.

How many bits are you recording at? Hot levels are more important at 16 bit than 24 bit. If you are recording at 24 bit, you can have the meters well below clipping (like -5 db or more) and still have a very clear and accurate recording with no loss of detail.

 

[Fishmed]

The headroom I am referring to is those on my faders on the board. I do agree on getting a clean signal when I am tracking. The noise floor comes in after the D/A conversion to the board and outboard gear. If I were staying digital, there would not be as big of an issue. I need good hot signals coming off the recorder as possible so I do not have to run analog gear wide open which is what would raise my noise floor. I would like to try to track hotter, but some of these bands are inconsistent in their playing, that their dynamics are all over the place, and that is why I sometimes have to track at lower levels.