NOT another "why are my mixes so quiet" post!

Discussion in 'Mastering' started by took-the-red-pill, Dec 3, 2012.

  1. took-the-red-pill

    took-the-red-pill Active Member

    No, today, my question is: Why are THEY so freakin' loud that they're over ZERO and into the red???

    Working on mixing a song. As part of that process, I have loaded a few commercially released tracks into a stereo track each. I'm trying to compare my mix to theirs. I listen to mine-which, oddly enough, sounds kinda like donkeys breeding-then I mute any effects on the master bus and play the commercially recorded songs clean and dry. This is my kinda clunky A/B testing.

    Now I thought that we were to make bloody sure that our end result did not go above the zero line. However, Tom Robinson, by Wide Mouth Mason(a highly under-rated Canadian blues/rock band) is pushing almost +1, and Butterfly by Jason Mraz is between +.5 and +1.5 on pretty much every single kick. Both songs seem to have some dynamics left, especially the Mraz offering, but they're just jacked! Seems to me Riding With The King by BB and Clapton was popping through the ceiling too last time I checked.

    So is this zero line not to be respected, as long as we don't hear distortion? Or have their mastering engineers done a very bad thing, or what?

  2. bouldersound

    bouldersound Real guitars are for old people. Well-Known Member

    Digitally speaking there is no such thing as over 0dB. But it is possible for samples at or below 0dB to represent a wave that goes over. Some converters will distort with such a signal and others won't. Maybe your metering reflects that in your reference recordings. What software are you using?

    Or something you're doing, like panning or routing, is adding a little gain.
  3. took-the-red-pill

    took-the-red-pill Active Member

    Hmm. When I tried it on Cubase the meters showed the same readings. I've since switched to Reaper. I confirmed on both programs that all faders are to 0.0 and all pans are centred, and NO effects on any bus in the chain. I toggled Voxengo Span meters on for a bit and they confirm the Reaper meter's readings, and show them over by the same amount, so it looks like it's the song.

    I am using an MP3 version of the song but I can't see that being the culprit. Why would an MP3 jack up a file from its wav state?

    I just played all of Butterfly, the highest peak says +2.0

    Maybe someone else try it and post what your meters are telling you, because mine are consistently showing me these things are over.

    I am in Canada, so it could be that the cold, dense air is adding sound pressure to the file...he he.

  4. bouldersound

    bouldersound Real guitars are for old people. Well-Known Member

    I assume left and right channels of the stereo track are panned apart since that's probably the default setting. But if you pan left and right to the center with 0dB pan law you'll get overs.
  5. took-the-red-pill

    took-the-red-pill Active Member

    AHA! I assumed the default pan law setting would be -3. It was set to 0.0. When I set pan law to -3 there are no longer overs on the commercial releases. Mystery solved.

    Thanks Boulder.
  6. RemyRAD

    RemyRAD Well-Known Member

    I don't think your mystery is solved???

    Firstly, if you take and download an MP3, it's a stereo track. And no panning is involved. It's a stereo rip. Only if you were to split it as 2 separate channels would you have to utilize any pan pots at all. And then they would still be panned, one extreme left and one extreme right. Which actually negates any particular pan law that you have selected. Now if you were to take both channels and pan them to center Mono, hell yeah, your pan law then is applicable. But that still doesn't really answer your question about the levels you're looking at.

    Every software manufacturer builds in a certain slop factor before peak indicators begin flashing. And there's nothing to say that your meters are calibrated to a red book standard. Simply because, they don't want you to make bad recordings of their software, they'll start flashing peaks at you before the peaks actually occur.

    Then there is that loudness mastering factor. Minor transient clips can go relatively unheard which makes that recording slightly more competitive in the loudness wars of mastering today. So do people ignore peaks? No, some of us encourage them. I've talked about creative clipping. There could actually be sonic benefits to it causing a harder dissonant crack to a drum track. Because of the extreme on order distortion that occurs from clipping peaks. But since I'm not in competition with these other morons, I'll then take my purposefully clipped peak drum transients and then normalize those tracks to -0.3 to -0.6 or .7? And in that application, it works a little bit like an enhancer. It really comes down to how you want your drums to sound? More aggressive than harder driving might benefit from a little transient clipping? But when you start to clip powerful legato lines, it becomes intolerable. So, yeah, some of us throw the meter to the wind, and a slight lead different way and/or magnitude as we did with saturating analog tape with drum tracks. But then you'll also find that consortium of folks that believe in clipping nothing. And that's fine for them. So a lot of this comes down to becoming very intimate with your equipment and learning how far you can push the envelope. So I'm not an envelope licker but an envelope pusher.

    It's easy to understand harmonic distortion and what it can or can't do.
    Mx. Remy Ann David
  7. bouldersound

    bouldersound Real guitars are for old people. Well-Known Member

    Remy, many DAWs have two separate pan knobs on a single stereo track.

    took-the-red-pill, I would think that the pan law would only be a problem if the left and right channels were panned center, or at least somewhat away from fully panned. I suppose if pan law were implemented as positive gain fully panned rather than negative gain at center then it would come into play with the channels fully panned.
  8. took-the-red-pill

    took-the-red-pill Active Member

    I'm not the expert here. All I know is that both tracks were significantly peaking, as noted.

    When I changed the pan law settings to -3(with no other changes, and everything set to 0.0), the peaking ceased on both commercial tracks, as seen by Reaper's meters, and also those in Voxengo Span.

    That's what happened. As for why is beyond the scope of my expertise.

  9. kmetal

    kmetal Kyle P. Gushue Well-Known Member

    when i'm a/b ing mixes to masters, i always try to set the fader on the reference in the ballpark of what i'm hearing of my mix. i mix w/ alot of headroom in on the master fader's meter usually between -6 to -10 db. Its inevitable that the mixes will sound quiter, so i just balance my mix and the references, so that i can hear the spectral balance in context.

    i think your meters were lying to you. it is not possible to go past 0dbfs in digital. clipping is just that, lopping the top's off. so your essentially EQ ing with a volume knob. Most commercial releases are mastered w/ a -.3db, to 0db ceiling. perceptively you could go 'louder or above' what a dbfs meter reports, but anything pushed above that digital ceiling, doesn't get registered, and to the meter, doesn't exist.

    this is not to say you can't hear clipping, it's just how much is ok for the tune. Also, referencing an mp3 is not the best. your judging the mix your trying to beat, by something that is degraded.

    people think that limiting/maximizing is the "secret ingredient" in commercial releases, so if i just set my brickwall limiter to -.3 i'll get it loud. sure it'll be loud. but frequency content and dynamics of the song/mix itself will make it sound "full, and rich" which i think alot of people confuse w/ loud.

    i can't say i can make a record that sounds like bob clearmountian mixed, and doug sax mastered. but i do know that alot more contributes to loud recordings than a limiter.
  10. RemyRAD

    RemyRAD Well-Known Member

    A pair of pan pots for a single stereo track Boulder? Yeah, I had that on both my Auditronics and Sphere recording consoles since they were designed for quadraphonic mixing. So I wouldn't doubt to see dual pan pots in the software for left-right and front-rear or, left-right pan in your rear. I couldn't figure any other better place to put it? But that still actually sounds a little on the awkward side? Where's the joystick? I'm a girl so I wouldn't know? (Even though I do... no I mean know).

    Here's what the different pan laws are all about:
    This was a big issue, back in the days of early rock 'n roll mixing. When you wanted that guitar flying left to right and back again. But this presented a problem. The idea originally was that if you had a 1 kHz 0 VU sine wave signal in the left channel and you started to pan it to the right channel, it should dip in at least three DB at the center 12 o'clock position. The gain then starts to go back up again as you traverse to the right channel where it is 0 VU, again. The reason why it was designed originally this way was if you were to listen to this cut in Mono such as on AM radio or from a common monaural record player. When you would get to the center of the pan pot, level would only bump up a little bit when the left and right channels contain the same signal. But when you listened to it in stereo, as it traversed the center of the pan pot, gain could drop a little. So, they created some pan pots that would not dip the level at the 12 o'clock center position to keep the power of that signaled more constant. But in Mono, nasty six DB burst of sound as you traversed the 12 o'clock position on the pan pot. Three DB you can live with better. So there are also pan laws that dip up to -6 DB when the pan pot gets to the 12 o'clock center position. But flying a guitar like that in a mix is sort of 1970s BS. Today we use stereo time delays and all sorts of other gobbledygook. And if you want to fly those guitars with the pan pot, you get to choose your pan law. I mean we use to create mixes separately for both AM and FM and this is something of a hangover from that era.

    So, what this really means is, maybe that's why these software is that you are using don't cost a bundle? From what you're describing, I would say that they are not exactly precise in how they have written the code to make their software meters work? And as I said, frequently both hardware peak indicators and software peak indicators have some kind of slight level padding, so as to show a peak could be happening, before it happens. Of course blatant overload is blatant overload but were not that blatant overload. We're talking about minor transient peaks that may or may not be actually getting clipped. And even if they are, you probably won't hear it. Because it's not blatant and incompetent overload. Either way, I think your software has provided for you some bogus fool proofing that is confusing a bogus fool? So congratulate yourself. You done good. In short, this is nothing really to worry about. Not much.

    If it sounds good? Print it.
    Mx. Remy Ann David
  11. took-the-red-pill

    took-the-red-pill Active Member

    All good. Not too freaked out about it, just wondering why they were going into the red. I just reports it, and lets you experts figure it out.

    To recap:

    Pan law set to zero, Butterfly, Tom Robinson going into the red by 2.5 dB at the highest peaks. Change pan law to -3 and they no longer show as clipping. Why? Buggered if I know. You figure it out.

    As for meters lying to me, maybe, but Cubase, Reaper, and Voxengo Span all show the same transient peaks going 'over,' and in the same spots, and by the same amounts. So either all three meters are full of beans or those peaks exist.

    Then again, maybe it's the mp3 thing, but that would mean converting to MP3 renders a hotter signal, or at least hotter transient peaks. I'd be surprised if it actually did, buy maybe.

    I'm pretty much over it, and would rather spend energy making music. Thanks for the input folks

    Over and out.
  12. RemyRAD

    RemyRAD Well-Known Member

    The reason why this would happen with your Tom Robinson track, changing pan law to -3, which obviously stops those excessive peaks. Of course it would. You changed the pan law from zero DB center pan to -3 DB, center pan. So anything within that near center region would cause those peaks with pan law set to zero. Most professional recording consoles always had a -3 DB center pan law. Though there were those PA boards that had a zero center pan law. And there were some recording consoles that had a -6 DB center pan law. But that was generally very rare with -3 DB center pan law, the norm. Not sure why your software has a default of zero DB center pan law?? Most don't do that unless you tell it to. So you possibly set your pan law to zero DB somewhere along the line, when you first installed and setup your software? And if all of your software is set up that way, so what? I mean we generally use the panoramic potentiometer or pan pot for positioning purposes. We don't use it usually as an effect. So we don't run into those kinds of problems you observed. When used as an effect, of course you would change your pan law to suit your needs. There is no right. There is no wrong unless you're talking about a stereo mix that has had the left and right channels combined for Mono presentation for AM radio. Where you would get a 6 DB burst of sound, as you traverse the 12 o'clock position. So if you were a good mixer, you would change your pan law then to -3 or -6 to suit your needs. You wouldn't want to leave your pan law at 0 DB, if you're using it as an effect.

    As I indicated previously, meters and your software and on actual standalone multi-track recorders, both analog and digital, that have peak indicating meters, are frequently adjusted to show peaks before peaks actually become clipped. Of course beyond their little extra cheated reference level where they set their meters to show peaks, is up to the manufacturer to decide whether they want to show you a peek that has already been clipped or one that is just about ready to clip. It doesn't necessarily mean that the signal has been clipped. In the case of precision digital audio analyzers, the software can be observed to verify where the actual peak indication begins to show. So the analyzer will likely show that when the peak indicator passes a threshold just prior to wave clipping, the indicator will show. Of course anything beyond this small warning window will in fact be clipped. So the software is not necessarily precisely calibrated to indicate everything precisely. It is precisely calibrated to indicate what the manufacture wants you to see. And when they want you to see it. And it really wasn't any different with analog equipment that had VU meters with an LED peak indicator. You are able to adjust the threshold, of that LED indicator, when it would light. So we knew that the VU meter was just an average volume unit indication. It did not show peaks on our tape recorders. We knew as we had to calibrate our own tape recorders. And we knew that they would begin to saturate the tape at about +15. So we might calibrate that peak LED indicator to start indicating peaks at +14? Meanwhile the meters are only moving around -6 to -10 get you may still see an occasional LED flash indicating the threshold has been crossed at +14. And understanding the nonlinear transfer characteristics of the tape, allowed us to purposefully slightly overload or saturate the tape on fast drum transient hits. This added a very desirable characteristic to how the drum sound was reproduced. A little saturation made them sound really super since the tape acted a little bit like a soft limiter while adding a certain kind of extra harmonic je ne sais quoi to the drum sound. Of course, this kind of technique doesn't quite work the same way it did in digital as it did in analog. Though there are elements of this kind of short transient clip overload that can actually contour the sound of the drums to sounding slightly more aggressive while still not being able to hear or detect this short transient clipping. So it could have been done on purpose? Or it could be sloppy mastering, trying to win the loudness war by allowing those short transient clips to just fly through? I mean did it sound good or did it sound screwed up? Without hearing it and by your description, I would think it probably sounded bitchin'? And of course, there are those like myself, but just love to push the buttons and the envelope. So knowing what you can get away with, legally, where no one gets killed, you're golden.

    Here's something fun to try. If you have some drum tracks? You could take those drum tracks and tell your software to normalize the drum tracks to 105% not 100%. This will cause deliberate clipping. Sometimes these clips come out sounding like " clicks ". That's an indicator that the digital to analog converter has definitely been exceeded in its output capabilities. And you don't want that clicking sound no way. So then what you do is you go ahead and normalize that track again to 97%. This will take away form back down so that the digital to analog converter does not exceed its output capabilities. While at the same time, you'll notice your drum tracks still have that light clipping from over normalizing to 105%. Listen to it back and compare it to the original track that has no clips. Then see what kind of a different drum sound you get that way? You might like this trick or you might hate this trick? It's just another tool in the huge palette of color modifications we have.

    The real trick here is that you don't want to record the drums and be clipping at the same time. Record them clean and then you can experiment with clipping, after-the-fact. Just don't say that clipped wave form without renaming it with what you've done to it, So as not to destroy or corrupt your cleanly recorded tracks. Disk space is cheap today.

    Hopefully you might understand the concept a little better now?
    Mx. Remy Ann David

Share This Page