gain structure while mixing in the digital domain.....

Discussion in 'Tracking / Mixing / Editing' started by briefcasemanx, Sep 23, 2004.

  1. Usually what I do during a mix on my computer is just keep turning each wave that is low in the mix up, and by the time i'm done I end up having the main mix level at like -15, and each waves volume at like +2 to +12 or something like that. In the analog world I know doing this might create problems, but in the digital domain does it even matter at all? Is it going to effect the quality of my mixes?

    I use Adobe Audition.
     
  2. zemlin

    zemlin Well-Known Member

    I think all of the native effects in Audition will work fine like this - internally I think everything is 32-bit float so no clipping will ocur even if a track is peaking above 0dB.

    Some third party effects I have WILL clip on a track if the signal is pushed too hard - even if it isn't hot going into the "effects rack" - if I push it too much in the rack it might be clipping by the time it comes out.

    Also, with some effects the amplitude of the incoming signal will affect what comes out of the other end.

    So as long as you are feeding a good level to your effects and not pushing things enough that you're getting clipping on an individual track or bus, I think you're OK.

    Make sure you mix down to 32 bit float - you won't get clipping even if the mixdown goes over 0dB. You just have to normalize, or limit if appropriate, to bring things to proper levels.
     
  3. Krou

    Krou Active Member

    I always leave my Master Fader at Unity and lower the individual tracks where needed. My average reading on tracks is -6 to -12, I very rarely go above zero, it's just how I'm used to doing it. I also never normalize...
     
  4. zemlin

    zemlin Well-Known Member

    You say that (or I "hear" it) as if normalize is a bad word.
     
  5. David French

    David French Well-Known Member

    Yes, he says that because every calculation, even simple volume change, is technically a degradation, so normalization is just one more.
     
  6. zemlin

    zemlin Well-Known Member

    OK - you win. When I got home I sat down at my DAW and took a short sample of a recording - 24 bit file which loads into AA as 32-bit floating point.

    I made a copy - cut by 6dB - boosted by 6dB - paste/invert and was quite surprised by how much difference there was. Not that I'm a normalizing adict or anything - but I will think twice next time. I had assumed that with floating point the differences would be quite minimal - granted it is low level stuff that my ears would not be able to detect - but such things would accumulate.

    [edit] The 6dB CUT & BOOST presets in Audition had the DC BIAS ADJUST option default to ON. That's where MOST of the difference came from. With that shut off the difference is down in the -150 dB range. My stupid goof - Not a total null now, but makes me feel better that Normalize and Amplify ARE pretty inert. - sorry about the brain fart. :oops: [/edit]
     
  7. Crap, with the card I'm using right now I have to DC bias adjust when I record. I am getting a Presonus Firepod though tomorrow so hopefully with the new card I can shut that option off and not degrade my sound that way.

    Should I be using 4 byte pcm or 4 bye IEEE float, and what does the box "try as WDM" mean? I have that checked in the options, i think by default.
     
  8. zemlin

    zemlin Well-Known Member

    I don't think the DC Bias Adjust is a bad thing, it just explains why I was seeing such a measureable difference between my before and after samples. The "Try as WDM" is a driver thing - some cards work better with it - some without. I think you just have to try it to see what works best for you.
     
  9. sorry to bring back an old thread, but by having tracks go above 0 and then bringing my master fader down will that change the dynamic range of my mixes? Or will there simply be the small amount of distortion incurred from all the volume changes?
     
  10. tomtom

    tomtom Guest

    Hi briefcasemanx,

    You need to specify the scale of the numbers you're talking about.
    When you say "by having tracks go above 0..." I assume it's 0dBFs (for Full scale) since you're working in the digital domain, but when you talk about "waves volumes of +2 or +12..." What do you mean?
    Are these the numbers you read on the internal mixer?

    Well, that said, I don't know what Adobe Audition does bad when performing destructive gain changes, probably not a lot, but I think that you should be more concerned about 2 things in order to preserve quality:

    Do not ever go above 0dBFs as you will clip your signal digitally, and though you might not hear it at the moment, it does get nasty somewhere else down the audio chain.

    Record your signals at a healthy level. Boosting level of your files in order to bring them to a good mixing level will raise noise and conversion artifacts. So think about it before you record.
    Record them hot, use all the dynamic range that you can afford but leave some headroom for further processing. or you'll have to lower the level of your file before applying EQ or compression or whatever you need.

    Thomas
     
  11. I understand the idea of volume adjustments creating distortion. I don't understand what it does to dynamic range. I'll try to explain it more thoroughly.....

    Each track is set at default at zero. I'm talking about the gain on each channel, not have anything to do with clipping. I don't know the technical term, dbfs whatever, but it just means that i have changed the volume of the track. When it's at zero it is the same exact level as when i first recorded it.

    Obviously, when i put a lot of tracks together the volume is going to get louder, and even though none of the individual tracks were clipping when recording, all the tracks combined in a mix may clip. When this happens I take the master fader and lower it below zero and it lowers all of the tracks relatively and keeps me from clipping.

    What I'm asking is, by using the master fader to lower all of the tracks relatively, am I decreasing the dynamic range of my mix at all?
     
  12. Dave62

    Dave62 Guest

    Reducing the master fader just scales back the output. It should not effect your dynamic range. Most daws have a lot of headroom internally and are designed for summing high level signals. You do have to keep an eye on your effects though. Adding 4 db of boost to a track that was only -3 dbfs will result in a clipped waveform and most daws eq's are pre fader. So the eq will output a clipped waveform, the track fader will turn it down and you still could have clipped waveforms at well below 0dbfs.
     
  13. tomtom

    tomtom Guest

    And so, if your master fader is clipping, it would be safer to bring all the faders of your mix down. If all tracks are lowered by, let's say 3dBs, it won't change your balance, neither should it change your dynamics unless you have silent parts in your mix.
    The dynamics of a mix is (forgive my english) how much level there is between the softest part and the loudest part of your mix.
    Dynamic is a term that is commonly misused, because we use it to describe something in terms of feeling...
    There's usually so much level to play with inside a daw, that lowering your overal levels for a clip free mix, won't change your dynamics.
    Inserting a compressor, limiter across your master output or gain riding is what changes your dynamics.

    T.
     
  14. vividsonics

    vividsonics Guest

    Hello,
    Another consideration that hasn't been discussed is signal to noise ratio. As a general rule for both fader and EQ settings, it is best for the noise floor (as well as conversion artifacts, and distortion as previously stated) to cut individual tracks and leave the master fader at "0". If you have, for example, 16 tracks and one track is not loud enough, cut the other 15 leaving the track that needs to be louder alone. By cutting all the other tracks as opposed to boosting the one, you will be lowering your noise floor (and other "undersirables") accross 15 tracks. Less noise and distortion to the mix bus is a good thing. If you find yourself encumbered by a massive amount of tracks, mix the individual instruments/tracks to a subgroup (or aux bus depending on your DAW's terminology). Then instead of tinkering with the mixes of different instruments (this is especially handy for large drum sets that are close mic'ed) you can get the instrument's subgroup mix and then just adjust the subgroup faders throughout the song. This method also works well in the analog domain and makes a huge difference in noise, distortion and headroom for analog mixers (assuming your mixer's subgroup summing is up to snuff). I hope this helps you out.

    Regards,
    Bob
     
  15. Roey

    Roey Guest

    Yes you are, but considering that your dynamic range was higher than the mix bus one (hence it was clipping), you bring the dynamic range to optimum by lowering the master fader (I'm pretty much repeating what Dave62 said).

    With regard to your initial question - it is better to bring down the master fader and not the individual tracks. I can give a simple explanation for this with integers, but a floating-point explanation will be far more complex.

    What vividsonics wrote is a bit misleading - once the audio is in digital form, bringing a fader down will attenuate both the signal and the noise by the same amount of dBs (ie, no SNR improvement on individual tracks); In practice, when you bring your fader down your SNR goes down as well (just take any audio, down by 50 dB, bounce, up by 50dB - lots of noise). Having one track louder and attenuating the others, or vice versa, is a separate issue which, if I understand right, was never your problem in the first place.

    I believe tomtom that you are confusing between the 'dynamics' of a sound and its dynamic range - these are two different things.
     
  16. Bobby Yarrow

    Bobby Yarrow Guest

    I'm completely mystified by this. Analog gain structure makes plenty of sense to me, but where to do my cuts on the digital end is a mystery.

    I find that I have these tracks where I'm boosting 12db in a compressor plugin, then cutting 6 db on the channel fader, then boosting another couple db on a bus compressor, then cutting another 5 db on the bus fader, then cutting again on the next bus, etc. Seems insane.

    I'm half tempted to leave all the faders at zero, and control all the volume thru compressor plugins.
     
  17. Turner

    Turner Guest

    Considering the recording and mixing proces ( in a box )
    as a whole I would suggest the following :
    - When recording try to record as "hot" as possible ( without clipping of course ), avoid using normalization.
    - When mixing get your individual levels as high as possible,
    considering the whole mix, get a good balance.
    ( no clipping of course ).
    Chances are when you do this and your master is at 0 db you
    might get some clipping on the master.
    Simply lower the fader a little bit to get rid of the clipping.
    Always check your prefader levels.
    Don't clip in the digital domain.
    Hope this helps you out a bit.
    Good luck !
     

Share This Page