1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

Hopefully quick and easy question to answer regarding record levels

Discussion in 'Recording' started by Method2Madness, Feb 1, 2012.

  1. Method2Madness

    Method2Madness Active Member

    Hello all,
    As you can tell, I'm new to this site. I've been into audio and it's surrounding technology for quite a while. I'm a DJ as my second job and have done quite a few remixes, light sound reproduction, as well as my share of home recordings. So I generally know how to make things "work" but what I've never really taken the time to do is understand why they work. Which leads me to my question:

    I'm currently using Adobe Audition to remaster some old songs that were either sampled from tape or were "ripped" from less than stellar sources i.e. youtube, etc. and have a question regarding the levels. I mostly understand the golden rule of not recording above 0db, but the "loudness war" really seems to raise it's ugly head more and more and I'm seeing recordings well above 0db. I also understand this tactic reduces dynamic range and if clipped, sound quality. But my question is this, if, unfortunately, engineers don't really pay much any attention to 0db anymore, exactly how high above 0db can one record without it clipping? I've done a couple of site searches and haven't really found a definitive answer. So please don't think I'm just popping up asking for answers without doing my share of the searching. ;)

    Thanks for any help provided. :)
     
  2. Guitarfreak

    Guitarfreak Well-Known Member

    0dB as an actual signal level is a best estimate really. 0dB is the amount of signal you can get onto analog magnetic tape without distorting (not always a bad thing) and also how much data a digital converter can convert without distorting (generally pretty bad). The line is not a definitive mark though, as lower end units can clip before the signal reaches 0dB and higher end units can sometimes be pushed beyond 0dB without clipping or artifacts.

    Recording levels is a different story. No matter how loud you want to make your mix, you should always record with plentiful clean headroom. Once you clip a signal, it cannot be unclipped, you won't get that dynamic range back through processing. Recording with headroom also helps mixes to sound cleaner and more crisp sounding. There are no rules, and when there are rules, there is always an exception to the rule, but this is one rule that you should really follow, especially just starting out. Play as loudly as you can (whatever instrument the mic is picking up) and make sure that the level stays within the range of -12 to -6dB as a general guideline for digital recording.

    In mastering, a mix may get smashed and compressed closer to 0dB if the producer desires this, but I have never actually seen a mastered mix go above 0dB... The RMS level goes up, but that's not the same thing as going over 0dB. 0dB in a mix and mastering sense is not a loose rule like stated above like when you have higher end gear you can push it harder, the point of mastering is to make the mix sound good on everything, and what good is a song if it clips your car radio from over signal due to poor mastering?
     
  3. bouldersound

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

    There's more than one kind of dB. If you don't specify then you're going to have a hard time discussing the subject.
     
  4. Method2Madness

    Method2Madness Active Member


    Awesome. That's really helpful and kind of goes along with my thinking. I really appreciate you taking the time to help me out. :)

    Indeed. I probably should've specified. In most instances for just sound mastering and "touch ups" I usually just go with Audition, which indicates with straight db.
     
  5. Method2Madness

    Method2Madness Active Member

    While this example doesn't show it going over 0db on the waveform view, the meter at the bottom shows it clipping. This is straight from CD as the engineer crafted it. I'm assuming this is not ideal, as it looks quite compressed. Is this assumption correct?

    Untitled-1.jpg
     
  6. bouldersound

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

    Almost certainly dBFS (full scale).

    When you digitize an analog signal without clipping the converters it can be converted back to analog without overs, but once any processing is applied all bets are off. It's quite possible to create a situation that causes some digital-analog converters to clip if two or more samples in a row are at 0dBFS. The volume wars and the proliferation of amateur recordists has made this common. It's standard practice to limit max levels to a few tenths below 0dBFS to avoid clipping DACs.

    Can you zoom tighter into the timeline to show the actual waveform where it's hitting 0dBFS?
     
  7. Method2Madness

    Method2Madness Active Member

    Certainly. And thanks for taking the time to go over this with me.

    Here is what I'm pretty sure you're asking for:

    Untitled-1-1.jpg

    As stated, the yellow box in the middle of the picture shows the parts that go over 0dbfs. Let me know if you need anything else from me.

    EDIT: It looks as if the forum resized the pic. If you want the full version, here's the link:

    http://i1208.photobucket.com/albums/cc374/method2madnesss/Untitled-1-1.jpg
     
  8. bouldersound

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

    This is what I was hoping to see:

    Clipped.jpg
     
  9. Method2Madness

    Method2Madness Active Member

    Ahhh ok. Sorry bout that. I'll do another and'll post it here in a few.
     
  10. Method2Madness

    Method2Madness Active Member

    Here ya go. Sorry bout that. My brain is sort of mush after working all last night and most of today. ;)

    clip.jpg

    EDIT: And heres another

    clip2.jpg

    You know what? The more I look at it the more it looks like it's only going above -1db and not above 0db. I think you having me get these helped me figure out that I was looking at the scale incorrectly.

    Yeap. Think I was. I was assuming that line was signifying 0dbfs, when it was actually signifying -1dbfs. I know this sounds incredibly retarded and trivial for someone knowledgeable, but you having me get these screen caps, actually helped me answer my own questions! I really appreciate it. :)

    clip3.jpg
     
  11. bouldersound

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

    That helps. If you look closely the reference line isn't actually 0dBFS but perhaps -1dBFS, so technically you don't have any digital overs. But the signal spends a lot of time near 0dBFS so it's likely that there are places where it will go over in the DAC. If I were working on this file I would limit it down to -0.3dBFS peak. You can use the Analyze function to find the actual peak levels.
     
  12. Method2Madness

    Method2Madness Active Member

    Awesome. Once I zoomed in that far, I found the same thing you did. The line is only at -1dbfs. All this time, I assumed the reference line was 0db. Your info really helped and I really appreciate it.

    And -0.3dbfs it is. I'll use that as my maximum from now on. (I was *mostly* using .1dbfs).

    I've given you rep, but it won't let me add any more for the day. I plan to stick around the forum and learn and I'll definitely shoot you some more rep.
     
  13. RemyRAD

    RemyRAD Guest

    This is something I've actually discussed and talked about. While digital recording does not allow for overloads without flattopped clipping, extremely minor transient peaks can frequently be clipped with no audible bad repercussions. That's partially because, the clip can go by so quickly, it is not heard. And yet, the same recording without the clipping may not sound quite as aggressive as the one with the minor clipping. That's largely because flat top/flatbottom clipping produces odd order third harmonic distortion which is dissonant sounding. This dissonance can make the recording sound slightly more aggressive. So sometimes, I'll purposefully take a clean mix, I'll slightly over " normalize ". This causes short duration transient flat top clips to occur. I then take that mix and will re-normalize to -.6 DB. Then nothing plays back with clipping indicated and I get a more aggressive sounding mix due to the odd order harmonic content I have created by minimalist use of clipping. But that's only on things like drum set transients not on vocals nor any other instruments. Even heavy metal guitars don't hold up well when clipped because they are more legato without fast transients like drums. Most folks here won't tolerate any sort of clipping. And a lot of low-end inexpensive digital to analog converters may clip out even without the waveform showing any clips. The least expensive ones just don't have the gusto that premium quality converters possess. And software may indicate clips that also cannot necessarily be perceived or heard. You certainly wouldn't want that on an operatic recording. On rock 'n roll, it's just part of the sound. I mean what do people like about the sound of a distorted guitar anyhow? They love the distortion. But that's mostly second harmonic which happens naturally in life and is not dissonant. So second harmonic distortion without any clipping can be more musical sounding than not having that second harmonic distortion. That's a tough one to actually understand.

    Some of the old-fashioned analog tube-based audio recorders like Ampex 350/351's had softer overload characteristics because of a second harmonic nature of the soft overload of tubes. Whereas the transistor versions caused hard third harmonic distortion components which were not nice. Thankfully, most of the professional-based transistor analog recorders had enough headroom in the electronics so that wouldn't happen. Instead, the tape was overloaded in a successful manner known as saturation. The tape unlike digital, would never cause flat top clip distortion elements. That's because tape was known for its nonlinear soft overload limiting characteristics. Digital just ain't like that. And some of the old-fashioned hardware-based "aural exciters/enhancers" had a knob with an indication of even or odd order distortion meaning, second harmonic or third harmonic. Sometimes things and sound better with more second order harmonic distortion other things would sound better with third order harmonic distortion. So my use of " creative clipping " comes under that definition of enhancement with third order harmonic elements on the transient peaks, only. So distortion isn't wrong when it's used in the correct way. It's only wrong when used incorrectly or accidentally as is most often the case.

    APHEX was actually a good mistake that happened at the right time in recording history with the proliferation of Dolby A. It brought back everything that Dolby A destroyed. So many of these enhancers/exciters, both hardware and in software are the creative use of specially crafted distortion components of both second and third variety. That along with a certain kind of phase shift which was not 180° but closer to 90° of phase shift. And at 90°, things are not summarily eliminated. Those exciters were also not utilized at full level but as a phantom lower-level source added in to the original sound. Unfortunately, people have a tendency to overuse exciter/enhancers to the point of ruining their sound totally. It's something that's supposed to be used in merely a nuance way where it improves the perceived audio rather than destroys it. The resulting effect of utilizing too much is a mostly unlistenable recording so it must be used in very small quantities for maximum effect. Thus, LESS IS MORE.

    Now I'm all excited and things are getting wet... with reverb.
    Mx. Remy Ann David
     

Share This Page