Why do digital tracks clip?

Discussion in 'Recording' started by jwschmidt, Mar 10, 2009.

  1. jwschmidt

    jwschmidt Guest

    Ok, so if I used an analog mixing board or in an analog recording environment, I can understand how pushing a fader up too high would lead to clipping. Physical components can only take so much of a signal's volume until hitting some kind of breaking point (I don't know the math behind it, but I know it exists).

    So, why does this happen in my DP5 digital mixer on my laptop? The signal is digital. Not analog. (I'm not talking about clipping during recording, but during mixing, after the mics and interfaces are unplugged and its just me and the software)

    Why am I not able to push the volume of a given track as loud as I want? At a certain point, it still hits the red. But what, in reality, does that red zone represent in the digital world? Again, I understand that analog circuits can't take an overly large amount of volume running through them, but aren't we just talking about ones and zeroes? How is digital clipping like this even possible?

    - This is not causing me any immediate problems, its just been a brain teaser for me for a while.
  2. Guitarfreak

    Guitarfreak Well-Known Member

    The Engineer I used to work for told me that the Analog To Digital Converters could only take so much information so fast, and at higher levels they begin to be overcome with information leaving holes in the signal.

    Not sure if this is the most accurate way of describing, as it has been a while and I'm going on memory here. He definitely said it had something to do with not being able to convert the signal to digital properly.
  3. BobRogers

    BobRogers Well-Known Member

    Blacksburg, VA
    Suppose you were going to do digital recording with two bits. You have four numbers to represent your signal 00 (0), 01 (1), 10 (2), and 11 (3). You want to design a converter to represent an analog signal with these numbers. You arbitrarily pick a range and split the range up evenly. Say 0 to 1 volts = 00, 1 to 2 volts = 01, 2 to 3 volts = 10, and anything above 3 = 11. So that a signal that is bouncing around like crazy between 3 and 10 gets "clipped" to a flat string of 11 11 11 11 11.
  4. IIRs

    IIRs Well-Known Member

    Your digital signal has to be converted to analog before you can listen to it!

    Most modern DAWs use floating point numbers to represent audio, so actually have enormous amounts of headroom and can represent numbers far higher than 0dBFS. I assume DP is no different, but you can check it if it allows 32 bit exports: import a normalised audio file, then boost the gain by a huge amount like 50dB so that its clipped beyond all recognition (turn your monitors down first!). Then export a 32 bit float file, re-import, and turn the gain of the imported file down by the same amount you boosted the original: the result should be exactly the same as you started with, a clean undistorted signal that peaks at 0dBFS.

    The distortion you heard when you originally boosted the track was occuring when the 32 bit floating point audio stream was converted to a 24 bit integer stream suitable to send to your DAC.
  5. jg49

    jg49 Well-Known Member

    I am not certain that JWS was talking about speaker distortion in playback but just clipping either in the waveform or "red lining" the master bus during mixing. In which caser what he is asking does not have to do with conversion. Maybe JWS you could clarify the question a little.
  6. jwschmidt

    jwschmidt Guest

    Hm. I'm not sure if my question has been answered by these responses? It may have been, but I'm not sure I'm bright enough to know!

    Ok, let me try and clarify.

    First thing I do is record some stuff with mics into my computer/DAW via an interface. No clipping problems, everything is in the green, all good. I finish recording sound. Time to mix.

    The audio files are now in digital form, sitting on my hard drive and appearing in my DAW's on-screen representation of a mixer, complete with little imitations of sliders. Everything is fine when I mix normally.

    Now lets say that, on a whim, I decide to drag one of those faders up higher than normal. The track gets louder, until it starts clipping. This is the scenario that I'm talking about. I am simply confused how it is possible to boost volume in an all-digital environment and cause clipping at all. (And its not my speakers clipping, because I can play stuff even louder out of them with no problem. Plus the monitor is showing red, so its clipping "in my computer")

    I just don't understand how a digital signal can clip on a simulated digital mixer just by boosting its volume. For a while, I was thinking that maybe its because the software designers were hardware purists and intentionally built in the limitations of analog equipment, though I'm pretty sure thats not the case.

    If that didn't properly explain it, let me ask my question another way:

    I understand that the meters on a digital console are there to indicate how hot the signal is running. When analog meters are reading red, it is because the signal has been boosted to a level at which the physical components of the workstation can no longer clearly transmit the signal, thus, it clips.

    Apparently, I don't understand what the meters in a digital console are indicating, because I know it has nothing to do with physical components. So when my meters read red on my computer screen, what is actually happening to the digital signal which is causing it to clip?
  7. hueseph

    hueseph Well-Known Member

    It has been answered by Bobrogers and IIRS. The clipping occurs in the conversion to 24bit as a result in the Digital to Audio conversion. The clipping in the DAW is just to let you know that when you finally go to 16bit 44k it will not sound good. It may be fine in the DAW but once it is downsampled, you will hear it.
  8. IIRs

    IIRs Well-Known Member

    Ok I'll try again.

    Imagine you have to write a number into one of those forms that gives you a box to write each digit in. If the form provides only 3 boxes, then the highest value you can represent is 999.

    Similarly, a 16 bit audio signal is comprised of individual samples, each of which is a string of 16 0's or 1's. The highest value that can be represented with this setup is 1111111111111111, or 65535 in decimal. If you try to represent a number higher than this it will be clipped to that maximum possible value: there's your distortion.
  9. BobRogers

    BobRogers Well-Known Member

    Blacksburg, VA
    So have you got the idea of a digital signal as a string of 8 or 16 or 24 bit numbers? And that there is a maximum possible number that that this system can represent? OK, well everything you do inside the DAW, eq, compression, etc., creates a new set of numbers. Well the simplest thing you can do in the DAW is push a fader. That takes your string of numbers and multiplies them all by a number determined by where you pushed the fader. If there was a nice peak in your digital signal where you had a nice waveform with a height around one half of the biggest number allowed in the system and you pushed the fader until it wanted to multiply everything by 3 then all of the numbers in that complicated waveform would be converted to 1111111....clip.
  10. Boswell

    Boswell Moderator Distinguished Member

    Home Page:
    Most of the respondents have talked about the mix bus output of the DAW mixer holding too large a value to fit in the range of the D-A converter. This can indeed be a problem, but I don't think it's what is happening here.

    From what jwschmidt says in his posts, I'm guessing his DAW software has an artificial limit imposed on the internal buses, trying to simulate a real (analog) mixer. It could be that the DAW mixer uses fixed-point calculations and there is real numerical overflow occurring, but I think that it's more likely that it uses floating-point internally and simply tests the summations against a limit value, limiting the sum and showing red if the limit is exceeded.

    So I think the answer is that the DAW software has been programmed to mimic an analog mixer, and limits at times and places where a real mixer might do. Don't worry about it, jwschmidt - work within the constraints of the tools you have, as with any piece of gear.
  11. rockstardave

    rockstardave Active Member

    dude just pull down the faders and turn up the speakers!
  12. RemyRAD

    RemyRAD Member

    Just as in a mixing console, one can physically overload the summing bus. It doesn't matter that the tracks themselves are not clipped. This is what mixing is all about. You have to work within certain parameters. Not to blow away all of the very smart & mathematically savvy folks here but what we're talking about is trying to stuff 10 pounds of crap into a 5 pound bag. It's alchemy, turning lead into gold. We know how to do it. Sometimes it's also known as "polishing turds", which means you have to be very good with playing with dirty.

    Dirty girl
    Ms. Remy Ann David

Share This Page