Digital Summing

Discussion in 'rec.audio.pro' started by TYY, Aug 30, 2003.

  1. TYY

    TYY Guest

    I am mixing a little project "in the box" using Sonar. I was wondering
    if there is anything I can do to improve the quality of the summing
    buss. Should I keep the levels low, or do I want to keep levels hot as
    possible?

    My only other option is using the mix buss on a b******r board, so
    thats out of the question. I don't run any signal through the mixer,
    only used it for monitoring.
  2. Get the output level as hot as possible without hitting 0dBFS, just like any
    other digital gain staging. If you are exporting a 24-bit file and plan on
    doing some mastering after mixdown, you don't need to get it as hot. If your
    peaks are hitting at least -12dBFS, you're in great shape.

    -Matt M

    "TYY" <tyler@dhiw.com> wrote in message
    news:b6b7a391.0308301152.4b5120ec@posting.google.com...
    > I am mixing a little project "in the box" using Sonar. I was wondering
    > if there is anything I can do to improve the quality of the summing
    > buss. Should I keep the levels low, or do I want to keep levels hot as
    > possible?
    >
    > My only other option is using the mix buss on a b******r board, so
    > thats out of the question. I don't run any signal through the mixer,
    > only used it for monitoring.
  3. I've never used Sonar, but I understand it sounds OK. I wouldn't worry about
    it as long as it sounds good to you. Get the level as hot as you can without
    clipping.

    --
    Bill L


    "TYY" <tyler@dhiw.com> wrote in message
    news:b6b7a391.0308301152.4b5120ec@posting.google.com...
    > I am mixing a little project "in the box" using Sonar. I was wondering
    > if there is anything I can do to improve the quality of the summing
    > buss. Should I keep the levels low, or do I want to keep levels hot as
    > possible?
    >
    > My only other option is using the mix buss on a b******r board, so
    > thats out of the question. I don't run any signal through the mixer,
    > only used it for monitoring.
  4. I'm always concerned when I see a generalized statement like "get the level
    as hot as you can without clipping" because it addresses nothing in terms of
    the type of music it is (a ballad doesn't need to be running at 0 dBFS all
    the time), nor what type of processing is being done, nor what the average
    RMS power of the music is. Barrelling down the road at 0 dBFS all the time
    is simply the best, quickest way to get flat sounding music with no dynamics
    nor stereo field, simply because one can place a brick wall limiter at .01
    dBFS and have .1 dB of dynamic range. Now that's not going to sound very
    good no matter what type of music it is. There's way more to discuss than
    simply cramming all the signals into one small space at the top of the
    dynamic range.

    --


    Roger W. Norman
    SirMusic Studio
    Purchase your copy of the Fifth of RAP CD set at www.recaudiopro.net.
    See how far $20 really goes.




    "Bill Lorentzen" <lorentzn@verizon.net> wrote in message
    news:8x94b.5079$vm2.3162@nwrddc03.gnilink.net...
    > I've never used Sonar, but I understand it sounds OK. I wouldn't worry

    about
    > it as long as it sounds good to you. Get the level as hot as you can

    without
    > clipping.
    >
    > --
    > Bill L
    >
    >
    > "TYY" <tyler@dhiw.com> wrote in message
    > news:b6b7a391.0308301152.4b5120ec@posting.google.com...
    > > I am mixing a little project "in the box" using Sonar. I was wondering
    > > if there is anything I can do to improve the quality of the summing
    > > buss. Should I keep the levels low, or do I want to keep levels hot as
    > > possible?
    > >
    > > My only other option is using the mix buss on a b******r board, so
    > > thats out of the question. I don't run any signal through the mixer,
    > > only used it for monitoring.

    >
    >
  5. mark-h

    mark-h Guest

    i agree. becauseof summing, if your individual tracks,submixes, aux etc are
    reading hot on their track meters, you will have little dynamic range at the
    final output. i liken it to driving a car at 60 miles an hour in 3rd gear:
    if ya shifted to 4th, you'd get much more responsive performance. itsa gain
    stage thing, and sonar has a "gain" control on each track--use it, andthe
    volume fader,correctly and you'll have a controlable mix at the main faders.
    mark-h

    "Roger W. Norman" <rnorman@starpower.net> wrote in message
    news:birqq8$sm8$1@bob.news.rcn.net...
    > I'm always concerned when I see a generalized statement like "get the

    level
    > as hot as you can without clipping" because it addresses nothing in terms

    of
    > the type of music it is (a ballad doesn't need to be running at 0 dBFS all
    > the time), nor what type of processing is being done, nor what the average
    > RMS power of the music is. Barrelling down the road at 0 dBFS all the

    time
    > is simply the best, quickest way to get flat sounding music with no

    dynamics
    > nor stereo field, simply because one can place a brick wall limiter at .01
    > dBFS and have .1 dB of dynamic range. Now that's not going to sound very
    > good no matter what type of music it is. There's way more to discuss than
    > simply cramming all the signals into one small space at the top of the
    > dynamic range.
    >
    > --
    >
    >
    > Roger W. Norman
    > SirMusic Studio
    > Purchase your copy of the Fifth of RAP CD set at www.recaudiopro.net.
    > See how far $20 really goes.
    >
    >
    >
    >
    > "Bill Lorentzen" <lorentzn@verizon.net> wrote in message
    > news:8x94b.5079$vm2.3162@nwrddc03.gnilink.net...
    > > I've never used Sonar, but I understand it sounds OK. I wouldn't worry

    > about
    > > it as long as it sounds good to you. Get the level as hot as you can

    > without
    > > clipping.
    > >
    > > --
    > > Bill L
    > >
    > >
    > > "TYY" <tyler@dhiw.com> wrote in message
    > > news:b6b7a391.0308301152.4b5120ec@posting.google.com...
    > > > I am mixing a little project "in the box" using Sonar. I was wondering
    > > > if there is anything I can do to improve the quality of the summing
    > > > buss. Should I keep the levels low, or do I want to keep levels hot as
    > > > possible?
    > > >
    > > > My only other option is using the mix buss on a b******r board, so
    > > > thats out of the question. I don't run any signal through the mixer,
    > > > only used it for monitoring.

    > >
    > >

    >
    >
  6. Ethan Winer

    Ethan Winer Guest

    TY,

    > I was wondering if there is anything I can do to improve the quality of

    the summing buss. <

    My guess is the very LAST thing you need to worry about is the quality of
    Sonar's summing buss.

    --Ethan
  7. Mike Rivers

    Mike Rivers Guest

    In article <sQj4b.30755$Qy4.7274@fed1read05> goodluck@cox.net writes:

    > i agree. becauseof summing, if your individual tracks,submixes, aux etc are
    > reading hot on their track meters, you will have little dynamic range at the
    > final output. i liken it to driving a car at 60 miles an hour in 3rd gear:
    > if ya shifted to 4th, you'd get much more responsive performance. itsa gain
    > stage thing, and sonar has a "gain" control on each track--use it, andthe
    > volume fader,correctly and you'll have a controlable mix at the main faders.


    That's the way to use it if it doesn't work right.

    One of the primary differences between a digital and analog system is
    that there's really no such thing as "lack of headroom." You have all the
    headroom you need until something comes along that requires more bits
    than you have the ability to store or process, and they you have no
    headroom whatsoever.

    A properly designed digital summing bus would have sufficient word
    length to add everything it's supposed to add without running out of
    bits. If you add two full-scale 24-bit numbers, you get a 25 bit
    number. I don't know if it extends forever, like if you add 24
    full-scale 24-bit numbers (highly unlikely if you're actually mixing
    music instead of theory) you get a 49-bit number, but just let's say
    you do. If you have a 52-bit accumulator (which some systems do) in
    which to store that sum, then you have not run out of headroom. Of
    course you need to get that 52-bit number back to 24 bits (or maybe 16
    bits) eventually, and there are good and bad ways to do that. But
    there is no reason to run out of headroom on a properly designed
    digital system.

    There's really no reason to run out of headroom on a properly designed
    analog system either, but there are practical reasons. Here, if you
    add two signals that peak at 2 volts each, you need a bus that can
    pass 4 volts. And if you add 24 of them, you'll need a bus that can
    pass a 48 volt signal. Few are designed to do so, for practical
    reasons, so you need to reduce the level of the signals going into the
    summing bus so that the sum is no larger than a signal with the
    greatest amount of distortion that you consider to be acceptable.




    --
    I'm really Mike Rivers - (mrivers@d-and-d.com)
  8. TYY

    TYY Guest

    "Ethan Winer" <ethan at ethanwiner dot com> wrote in message news:<FWqdnVWsvpd4UcyiXTWJkQ@giganews.com>...
    > TY,
    >
    > > I was wondering if there is anything I can do to improve the quality of

    > the summing buss. <
    >
    > My guess is the very LAST thing you need to worry about is the quality of
    > Sonar's summing buss.
    >
    > --Ethan



    Why is that? I always hear about how digital summing busses sound like
    shite. Having never compared digital vs analog summing myself, I have
    never experienced the differences firsthand. I simply wanted to know
    if there is a mathematical advantage to hitting the buss with low vs.
    high levels. I intend on having the project professionally mastered,
    so no limiting/excessive compression is taking place at this stage.
    Mastering can bring up the levels. BTW, the project is an
    americana/pop/rock/country thing.
  9. Ethan Winer

    Ethan Winer Guest

    TY,

    > Why is that? I always hear about how digital summing busses sound like

    shite. <

    Don't believe everything you hear.

    > Having never compared digital vs analog summing myself <


    Then let's fix that right away. Here's how:

    Take a track off the best sounding CD you have - one that is absolutely
    clear and sparkling - then rip it from the CD to a Wave file and import it
    into a track in Sonar. Now play that track and raise the volume up and down.
    That "exercises" the summing buss as it performs the math necessary to raise
    and lower the volume. How does it sound? Pretty good, eh? If you want to get
    even fancier, take another track and import it, and mix the two together.
    This works best if the tracks are sparse, as opposed to full dense mixes.
    Maybe you have some Acid or sample CDs that have good sounding drums and
    other good sounding instruments. Mix them together and listen carefully to
    each component. It still sounds great, right?

    > I simply wanted to know if there is a mathematical advantage to hitting

    the buss with low vs. high levels. <

    No there isn't an advantage one way or the other. Again, this stuff is so
    easy to test I'm amazed that people continue to spread rumors rather than
    just spend ten minutes to test it for themselves.

    --Ethan
  10. Arny Krueger

    Arny Krueger Guest

    "Ethan Winer" <ethan at ethanwiner dot com> wrote in message
    news:wrKdncmxsqLK1M6iU-KYuA@giganews.com
    > TY,
    >
    >> Why is that? I always hear about how digital summing busses sound
    >> like shite. <

    >
    > Don't believe everything you hear.
    >
    >> Having never compared digital vs analog summing myself <


    > Then let's fix that right away. Here's how:


    > Take a track off the best sounding CD you have - one that is
    > absolutely clear and sparkling - then rip it from the CD to a Wave
    > file and import it into a track in Sonar. Now play that track and
    > raise the volume up and down. That "exercises" the summing buss as it
    > performs the math necessary to raise and lower the volume. How does
    > it sound? Pretty good, eh? If you want to get even fancier, take
    > another track and import it, and mix the two together. This works
    > best if the tracks are sparse, as opposed to full dense mixes. Maybe
    > you have some Acid or sample CDs that have good sounding drums and
    > other good sounding instruments. Mix them together and listen
    > carefully to each component. It still sounds great, right?


    To do a truly bang-up job of this, one should record a 1 KHz tone duration a
    second or two at the front or back of the sound clip, and be sure to match
    levels with it after your fooling around to add noise. It seems like a good
    test might be to cut the level of a track by say, 20 dB and then amplify it
    by 20 dB, and repeat this a dozen or more times.

    Let me recommend the 24/96 "triangle" test sound at www.pcabx.com for this
    purpose. It has the widest dynamic range of any musical recording generally
    available to the public that I've been able to find. The tail is a
    relatively pure sound. It is recorded quite dry so whatever gets added to it
    will tend to stand out.

    After abusing your test file thoroughly, match levels with the test tone,
    edit out the test tone and compare it to the original using one of the fine
    comparators downloadable from http://www.pcabx.com/program/index.htm .

    Let your ears, and just your ears be the judge!

    >> I simply wanted to know if there is a mathematical advantage to
    >> hitting the buss with low vs. high levels. <


    Mathematically, any operation adds some (usually microscopic) degradation.
    The question that is always relevant is how audible the degradation is.

    The size of the degradation depends on whether the bus is implemented with
    fixed point or floating point. The size of the degradation depends on how
    many bits are used for the calculations. In 16 bit fixed point, its
    conceivable that there could be some audible degradation. In 32 bit floating
    point with a 24 bit mantissa, it seems like it would take an immense amount
    of operations to cause problems.

    Then you have to look at the nature of the degradation. If the operations
    are properly dithered, mathematical degradation in a digital mixing bus can
    be expected to take the form of very low level white noise. Even at 16 bits
    it takes quite a large number of these small doses of degradation to build
    up into something that you'll hear over the usual relatively high levels of
    noise from rooms, mics, mic preamps, etc.

    One thing to remember is that the white noise generated by math operations
    is dependent on the data. Since the data is being changed by each processing
    step, the white noise being added by every step tends to be uncorrelated
    with the noise generated in the previous step. IOW we have uncorrelated
    noise sources.

    In other threads we've looked at what happens when you add a smaller
    uncorrelated noise to a larger noise. Basically, the smaller noise gets
    washed out and adds almost nothing. That means that the noise build-up due
    to many calculation steps builds up very slowly as the number of steps
    actually performed increases.

    > No there isn't an advantage one way or the other. Again, this stuff
    > is so easy to test I'm amazed that people continue to spread rumors
    > rather than just spend ten minutes to test it for themselves.


    I think there have been times in the recent past when badly-designed procedu
    res and small data words were used to implement digital mixing buses.
    There's nothing to say that a programmer's error won't cause a widely used
    product to suddenly have a bus that gets noisy under obscure conditions that
    some people stumble into.

    It's been a long time since I've heard noise being added by the digital
    editing tools I use, but then again maybe I'm lucky or maybe I've been using
    good tools.
  11. P Stamler

    P Stamler Guest

    >There's really no reason to run out of headroom on a properly designed
    >analog system either, but there are practical reasons. Here, if you
    >add two signals that peak at 2 volts each, you need a bus that can
    >pass 4 volts. And if you add 24 of them, you'll need a bus that can
    >pass a 48 volt signal.


    Only if the two signals are identical. If they're not, then to add two of them,
    you'll typically need a bus that can pass about 2.8 volts. Adding 24 of them
    will typically require a bus that can pass about 9.8 volts. Uncorrelated
    signals add by the root-mean-square law.

    Peace,
    Paul
  12. Ethan Winer

    Ethan Winer Guest

    Arny,

    Great stuff as always. Thanks.

    > cut the level of a track by say, 20 dB and then amplify it by 20 dB, and

    repeat this a dozen or more times. <

    That's what I did (once pass only) in the "24 bit test" on my site, to
    reduce Wave data from 24 bits to 16, 13, 11, and even 9 bits. People are so
    obsessed with high resolution they miss the forest for the trees. Just
    yesterday a fellow who downloaded the files admitted he heard no difference
    between any of them. If you listen carefully you can distinguish the 11 and
    9 bit files, but even that low resolution is not as bad as some people think
    it "should" be.

    > Mathematically, any operation adds some (usually microscopic) degradation.

    The question that is always relevant is how audible the degradation is. <

    You mention noise as an artifact, but I usually think of math as generating
    distortion. I guess you get both. What amazes me is how many people fret
    over a mix buss adding noise at -100 dB or adding 0.002% distortion, while
    acoustic interference in their control room causes frequency response
    variations of 15 dB or more throughout the entire low end!

    --Ethan
  13. Bob Cain

    Bob Cain Guest

    P Stamler wrote:
    >
    > >There's really no reason to run out of headroom on a properly designed
    > >analog system either, but there are practical reasons. Here, if you
    > >add two signals that peak at 2 volts each, you need a bus that can
    > >pass 4 volts. And if you add 24 of them, you'll need a bus that can
    > >pass a 48 volt signal.

    >
    > Only if the two signals are identical. If they're not, then to add two of them,
    > you'll typically need a bus that can pass about 2.8 volts. Adding 24 of them
    > will typically require a bus that can pass about 9.8 volts. Uncorrelated
    > signals add by the root-mean-square law.


    Audio tracks, however, are seldom uncorrelated. The purpose
    of the beat is to correlate and it will almost certainly
    cause peaks that exceed what the RMS principle will
    indicate. While statistically less probable, nearly N times
    the per track peak is quite possible.


    Bob
    --

    "Things should be described as simply as possible, but no
    simpler."

    A. Einstein
  14. Mike Rivers

    Mike Rivers Guest

    In article <20030901131817.19898.00000378@mb-m28.aol.com> pstamler@aol.com writes:

    > >There's really no reason to run out of headroom on a properly designed
    > >analog system either, but there are practical reasons. Here, if you
    > >add two signals that peak at 2 volts each, you need a bus that can
    > >pass 4 volts.


    > Only if the two signals are identical.


    If they each peak at 2 volts at the same time no matter what they do
    the rest of the time, you'll still have a 4 volt peak. If you have a 3
    volt power supply, you're into clipping.

    > If they're not, then to add two of them,
    > you'll typically need a bus that can pass about 2.8 volts. Adding 24 of them
    > will typically require a bus that can pass about 9.8 volts. Uncorrelated
    > signals add by the root-mean-square law.


    You can design a console so that statistically it will work fine, but
    there will always be exceptions. They may not last long enough to hear
    much of the time but they may last long enough to measure.




    --
    I'm really Mike Rivers - (mrivers@d-and-d.com)
  15. GKB

    GKB Guest

    I asked this question a while ago with no replies , but
    has anyone noticed [ or opinionated ] about the digital summing
    of the DAWs Vs digital mixers ? [ like the Yamaha O3D ]

    Regards Greg






    TYY wrote:

    > I am mixing a little project "in the box" using Sonar. I was wondering
    > if there is anything I can do to improve the quality of the summing
    > buss. Should I keep the levels low, or do I want to keep levels hot as
    > possible?
    >
    > My only other option is using the mix buss on a b******r board, so
    > thats out of the question. I don't run any signal through the mixer,
    > only used it for monitoring.
  16. Ben Bradley

    Ben Bradley Guest

    In rec.audio.pro, mrivers@d-and-d.com (Mike Rivers) wrote:

    >
    >In article <20030901131817.19898.00000378@mb-m28.aol.com> pstamler@aol.com writes:
    >
    >> >There's really no reason to run out of headroom on a properly designed
    >> >analog system either, but there are practical reasons. Here, if you
    >> >add two signals that peak at 2 volts each, you need a bus that can
    >> >pass 4 volts.

    >
    >> Only if the two signals are identical.

    >
    >If they each peak at 2 volts at the same time no matter what they do
    >the rest of the time, you'll still have a 4 volt peak. If you have a 3
    >volt power supply, you're into clipping.


    You're right, Mike - both Paul and Bob (who was trying to defend
    linear addition, but based on correlated signals - peaks on
    UNcorrelated signals also add linearly with channel count) should be
    ashamed of themselves for making this error! ;)
    FWIW I've found them both to be very knowledgable, so errors like
    this from them are rare and unexpected.

    Even peaks of truly uncorrelated signals (such as several sources
    of white noise) may happen all at the same time, and such peaks will
    add arithmetically. Admittedly, the probablility of such a peak
    decreases [greatly] as the number of sources increases, but the chance
    is still greater than zero. You may be able to get away with less than
    the theoretical maximum headrooom, but doing so is definitely an
    engineering tradeoff/compromise.

    >> If they're not, then to add two of them,
    >> you'll typically need a bus that can pass about 2.8 volts. Adding 24 of them
    >> will typically require a bus that can pass about 9.8 volts. Uncorrelated
    >> signals add by the root-mean-square law.


    Paul, you must be thinking of the average level (or is it the RMS
    level?) of uncorrelated sources adding that way. As I'm sure you
    know, mix busses (and DAW summing routines) clip at peaks (which don't
    neccesarily add the same way), not average or RMS levels.

    >You can design a console so that statistically it will work fine, but
    >there will always be exceptions. They may not last long enough to hear
    >much of the time but they may last long enough to measure.
    >
    >
    >
    >
    >--
    >I'm really Mike Rivers - (mrivers@d-and-d.com)
  17. tyler@dhiw.com (TYY) wrote in message news:<b6b7a391.0308301152.4b5120ec@posting.google.com>...
    > I am mixing a little project "in the box" using Sonar. I was wondering
    > if there is anything I can do to improve the quality of the summing
    > buss. Should I keep the levels low, or do I want to keep levels hot as
    > possible?
    >
    > My only other option is using the mix buss on a b******r board, so
    > thats out of the question. I don't run any signal through the mixer,
    > only used it for monitoring.



    I don't have digital theory locked down as well as a lot of you
    guys so I wont comment on that, but I do use Sonar, and like many
    other people I am disapointed with the sound of the digital mix buss
    not only in Sonar but pretty much all DAWs. A lot of people used to
    rave about Paris but I never had a chance to hear it and it's no
    longer supported so I guess that no longer matters. I can't imagine
    what they were doing that no one else is able to figure out.
    Tyler, I would recommend that you try mixing through your b******r
    board (If you have decent A/D and converters available)and compare
    your results. You
    may be suprised. I have mixes that I did on a Topaz mixer and a Fostex
    1/2" 16 track that sound more dimensional than my internally mixed DAW
    mixes. Granted the tape machine adds another variable to the equation
    but I don't hear many people raving about the sound of Fostex 1/2" 16
    track recorders. The sad thing is that I sold my Topaz in order to buy
    some API and Neve clone preamps so I can't do that comparison myself
    at the moment. I keep finding that when I track to the DAW I can get
    tracks that I am very happy with but when it comes to mix time the
    mixed results sound flat without the sense of space I'd like and it
    takes a lot longer for me to get a mix happening.
    I know this is a tired subject but I think there is something to
    this argument because I consistantly find that the albums that sound
    good to me have been mixed on a console. If the Brendan O'Brians and
    Andy Wallaces are still doing it this way then there must be something
    to it.
    I think that eventually DAW summing will sound as good as an SSL
    console ( maybe even affordably) but how long is that going to take?
    In the meantime I'm saving for a good mixer.
  18. Pooh Bear

    Pooh Bear Guest

    Ben Bradley wrote:

    > In rec.audio.pro, mrivers@d-and-d.com (Mike Rivers) wrote:
    >
    > >In article <20030901131817.19898.00000378@mb-m28.aol.com> pstamler@aol.com writes:
    > >
    > >> >There's really no reason to run out of headroom on a properly designed
    > >> >analog system either, but there are practical reasons. Here, if you
    > >> >add two signals that peak at 2 volts each, you need a bus that can
    > >> >pass 4 volts.

    > >
    > >> Only if the two signals are identical.

    > >
    > >If they each peak at 2 volts at the same time no matter what they do
    > >the rest of the time, you'll still have a 4 volt peak. If you have a 3
    > >volt power supply, you're into clipping.

    >
    > You're right, Mike - both Paul and Bob (who was trying to defend
    > linear addition, but based on correlated signals - peaks on
    > UNcorrelated signals also add linearly with channel count) should be
    > ashamed of themselves for making this error! ;)
    > FWIW I've found them both to be very knowledgable, so errors like
    > this from them are rare and unexpected.
    >
    > Even peaks of truly uncorrelated signals (such as several sources
    > of white noise) may happen all at the same time, and such peaks will
    > add arithmetically. Admittedly, the probablility of such a peak
    > decreases [greatly] as the number of sources increases, but the chance
    > is still greater than zero. You may be able to get away with less than
    > the theoretical maximum headrooom, but doing so is definitely an
    > engineering tradeoff/compromise.
    >
    > >> If they're not, then to add two of them,
    > >> you'll typically need a bus that can pass about 2.8 volts. Adding 24 of them
    > >> will typically require a bus that can pass about 9.8 volts. Uncorrelated
    > >> signals add by the root-mean-square law.

    >
    > Paul, you must be thinking of the average level (or is it the RMS
    > level?) of uncorrelated sources adding that way. As I'm sure you
    > know, mix busses (and DAW summing routines) clip at peaks (which don't
    > neccesarily add the same way), not average or RMS levels.
    >
    > >You can design a console so that statistically it will work fine, but
    > >there will always be exceptions. They may not last long enough to hear
    > >much of the time but they may last long enough to measure.


    Just to get some reality into this - if you were mixing 24 tracks, each simultaneously
    peaking at 2 volts with 0dB gain to the L/R bus, it wouldn't say much for your mixing
    skills ! About 10dB above practical analog levels.

    For practical purposes the uncorrelated method of adding sources works fine.

    I like to sum the bus amp 6dB below channel level which allows for mixing 4 seriously
    hot signals without any loss of overall headroom. Using it in the above example (
    assuming uncorrelated - non simultaneous peaks - in reality fine ) means that would
    mean you would be roughly 10 dB below clip at the bus amp.

    Don't forget that the 'main mix' is what you're aiming for. It's no damn good if it's
    clipped because you're running the channels too hot. This is why we have faders and
    gain controls.


    Graham
  19. Pooh Bear

    Pooh Bear Guest

    Matt McGinley wrote:

    > Get the output level as hot as possible without hitting 0dBFS, just like any
    > other digital gain staging. If you are exporting a 24-bit file and plan on
    > doing some mastering after mixdown, you don't need to get it as hot. If your
    > peaks are hitting at least -12dBFS, you're in great shape.
    >


    Good advice. Run the signal any hotter than that and you're getting
    uncomfortably close to digital peak - eeek ! A couple of bits spare is very
    wise.


    Graham

Share This Page