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

0dBfs

Discussion in 'Recording' started by MightyFaulk, Feb 22, 2009.

  1. MightyFaulk

    MightyFaulk Active Member

    is hitting 0 dBfs considered clipping? The reason I ask is because I just purchased a presonus central station and the LED meter considers 0dBfs a clip.
     
  2. IIRs

    IIRs Well-Known Member

    Yes. Some meters will register an over after a single full scale sample, while others will wait for a certain number of consecutive full scale samples. Others will oversample the audio to detect possible inter-sample peaks in oversampling DACs.

    Until floating point audio came along 0dBFS was always the highest possible value, so the only way to detect an 'over' was to wait for at least 3 consecutive 0dBFS samples, and deduce that the middle one was originally higher... the clip lights on your DAW mixer channels probably work the same way for historical reasons, even though overs are not actually clipped.

    In other words, the 'clipping' LED is really a 'you have no headroom left' LED.
     
  3. soapfloats

    soapfloats Well-Known Member

    IIRs:

    Would this be true for analog pres as well?

    I have a Focusrite 428 and a Sytek 4ch.
    The 428 is clearer, since I have a VU meter as well.
    But the Sytek only has LED signals.

    Reason I ask is, while I can hear the clipping occur, it's difficult to pinpoint the source in a 16 channel live (yes, PA and monitors) session.
    Since I'm trying to make a good recording and provide a good PA mix at the same time, those clip indicators really help me fix it quickly.

    Side note: how accurate/useful are things like VU meters on more budget gear. The pres above are my biggies, but I use some cheaper Presonus (SS & tube) stuff if necessary.
    The tube pres have VUs, but I've heard that some of the cheaper products include relatively useless features (like VU, visible tube - w/ glow!) to create some curb appeal. Correct?
     
  4. RemyRAD

    RemyRAD Guest

    Originally, the "VU" or, more correctly the Volume Unit meter specification had very specific rise & fall balistics. They were designed to show you average levels and are fairly insensitive to displaying any peaks. Whereas in England & Europe, PPM meters or, Peak Program Meters were quite a bit different in their operation along with their display. Today, with the advent of LEDs & computer displays we are provided with greater visual information as to what is actually being recorded. I personally use to prefer VU meters with LED peak indicators. Those little VU meters are not necessarily made to the same standards as the old original large meters were. When we recorded things in analog, we knew that with VU meters, what ever you were looking at had peaks that were 10 to 15 DB higher than what the meter could indicate. Peak meters just used to get me rather nervous since all they really ever did was constantly quiver. Sort of like an inbred Yorkshire terrier. Then they would fall in slow motion just like when you came home from a hard day's work recording a lousy band and you flop into bed. Plus, when we recorded in analog, banging VU meters while recording drums, provided us with the visual knowledge that the tape was providing some natural limiting along with a satisfying saturation to the sound. It was a lot different if you were trying to use ppm meters. So that's why you used to find those on tape recorders mostly. But with digital it's a whole different ballgame. Although I am one of the crazies here that believes a very slight amount of peak clipping on drum transients can provide for a punchier sound if only done very slightly. In that respect, it's harder to record that way. But after recording, with manipulation of software, can be accomplished in a more controlled manner with better predictable results that can be chosen or not. This probably has not answered your question? But I believe when dealing with an analog preamp an averaging meter system reliably provides you with the most practical of operating parameters. Anything digital is better represented by a peak program metering since the highest indication indicated, is in fact the end of the line. And you really don't want to cross that line. Bottom line is both are important meters to understand and some are just plain fun to watch. Interestingly enough most analog tape recorder alignments can be performed with the onboard VU meter indications. Not so with the ppm types. So you'll generally find the ppm types on the console. They were even some mechanical meters produced that had dual mechanisms in a single enclosure providing you with a consistent visual indication of both average & Peaks. But those cost a bundle and LEDs are much more affordable and can be either or both, simultaneously or separately. Some high-end consoles featured "plasma" metering which are in fact actually neon lights and you only found those on the high-priced spread of consoles. Not necessarily prudent to have a couple hundred volts of metering in an audio console. LEDs generally can't kill you. Plus they are more consistent & last longer than plasma metering but not as high in resolution. Everything is a trade off compromise in our business and you have to learn how to interpret what you're looking at or hearing. A true recorded peak will generally be "flattopped" of which there is no good way to undo. It will be clicky sounding & richer in third harmonic distortion which is dissonant sounding. Analog & tubes when overloaded, generally are richer in the second harmonic distortion which is musical sounding and occurs naturally in life. Third harmonic distortion does not occur naturally in life and therefore is generally not anything anybody wants. Sort of like the way your clock radio sounds when it's turned up too loud. Just plain bad. If your preamps only had a single peak LED, that's really all that's necessary. So, anything lower would be just, lower. I generally still record at 16-bit but the argument for 24-bit actually makes sense since it provides for an extra cushion of comfort in the lowest volume levels. Some people think that they can hear greater resolution at 24-bit but I say that's a bunch of whoee. It's not quite connect the dots in the sense that everybody believes it to be. But if you think you can hear a difference and believe that difference is important to you? Then go for it. I don't believe in it. I believe in working in the format in which most will listen to it in. And especially when it comes to rock-and-roll. Orchestral is a bit different. The fact is, I don't like the way PCM sounds to begin with. But it's convenient & an established standard so we all use it. And it's probably different for the folks who have been handed substantial financial contracts. But today that's more rare than it used to be. I think all PCM, regardless of sample rate & bit depth all sounds the same, i.e. it ain't analog sounding. I hate fluorescent lighting and I find PCM to be the audible analogy to fluorescent lighting. It ain't an incandescent bulb. But then one is more efficient & one is more wasteful.

    Wasted engineer
    Ms. Remy Ann David
     
  5. MightyFaulk

    MightyFaulk Active Member

    Remy,
    Your posts are always great to read. Packed with a lot of information, saturated with humor, just how I like it.

    IIRs,
    Thank you for the low down on 0dBfs. To me, it almost seems like allowing 0dBfs peaks (or cutting off anything above that) is a general practice in digital audio recording.

    But is this good practice during a mixdown? I can understand why a CD would constantly hit 0dBfs after it's been normalized and compressed to all hell so that it can be a #1 hit on the loudness wars chart. Should I practice mixing at a lower level, say -4dBfs to raise the intrinsic loudness of a piece without using all the headroom?

    To me it's bothersome to always see the LED meter light up red at 0dBfs on my presonus central station, because it's not an over. I've considered calibrating it a few dBs lower than the meter on my DAW but I have a hard time justifying two different levels on the meters just because I don't want to see it hit red all the time.
     
  6. IIRs

    IIRs Well-Known Member

    You shouldn't really ever be hitting 0dBFS.

    When mixing I would suggest aiming for average RMS levels of around -20dB. This will put your peak levels somewhere between -8 and -4 dBFS for most material, and means you never need to worry about clipping.

    When mastering you still shouldn't peak at 0dBFS, as this is likely to cause inter-sample peaks in oversampling DACs. Try the free SSL X-ISM meter, which shows both 'digital' and 'analog' overs. Ideally you should leave a tiny bit of headroom and only peak to about -0.4dBFS.
     
  7. MightyFaulk

    MightyFaulk Active Member

    thanks IIRs, makes sense to me.

    One last thought,
    why would professionally mastered music be hitting the 0dBfs mark? I'm talking rock CDs here, not necessarily classical music or anything that should be softer. I've heard 0dBfs can cause issues in a low quality codec when converted to mp3.

    Anyways, I appreciate the help!
     
  8. RemyRAD

    RemyRAD Guest

    I think the reason for that is that you'll find that type of rock-and-roll to already be rather aggressive. That extra distortion just adds to its aggressive perception & nature as opposed to nurture which good audio engineers generally do. There I go again.....

    I've always joked about telling people that when the peak lights light up. Peak does not always mean at its best. Of course it could be interpreted that way by folks that are not in the know. No what I mean? Or is the know what I mean? But in the English language, it can be effectively mangled by differences in interpretation. That's why my answers frequently include non sequiturs and/or other confusing mangled sentences. Plus my grammar is crap as English was obviously not my forte'. Another reason why I decided engineering was better than being an author.

    In an interview with the news press in 1945 this question was posed.
    "Dr. Oppenheimer, do you know you miss pronounce nuclear?" His response was "Yes, I know I mispronounce nuclear because I can't say nuclear I can only say nukeular...". One of my favorite quotes that just blows me away. I think I'll go out for some sushi for lunch?

    Cy your nara, baby.
    Ms. Remy Ann David
     
  9. cfaalm

    cfaalm Active Member

    When CD players were introduced most of them could not handle 0dBFS and treated it like an error. So when mastering they had to make sure it was kept at -0.3dBFS.

    If an mp3 codec cannot handle 0dBFS it is probably an inferior or outdated codec (perhaps from the -0.3dBFS era). With the loss of sound in converting to mp3, any deficiencies in the mix will become worse. When a CD is already too hot/loud, the mp3s won't sound any better.
     
  10. soapfloats

    soapfloats Well-Known Member

    You actually answered my question(s) quite well Remy.
    Thank you.
    I too don't mind the slight overload (on the 428, for me) on drums.

    Looks like I need to investigate whether those clip LEDs on the BlueTube/TubePre represent the tube, the SS, or the output (and which is last in the chain). The manuals don't help. It'd be nice to a have a diagram, no?
     
  11. jg49

    jg49 Well-Known Member

    You could try the Presonus forum in re: to the Blue Tube meter chain. The tech guys there usually answer questions pretty well.
     
  12. MightyFaulk

    MightyFaulk Active Member

    I'm starting to see a trend in my outboard meter capturing higher peaks than in logic. I'm wondering if the meter in Logic (and perhaps other DAW software?) doesn't have a high enough resolution to capture some transient peaks. Or maybe it's a difference between the analog and digital domain. I'm going to try the SSL X-ISM meter out and see if there's any difference.
     
  13. Codemonkey

    Codemonkey Well-Known Member

    MightyFaulk; odds are, the meter in Logic has a higher attack time - it takes longer to respond to peaks.

    I like TrackMeter from http://rs-met.com/ (freebies page) which has adjustable attack/release times. Works as a VST plugin and beats Kristal's built in displays.
     
  14. MightyFaulk

    MightyFaulk Active Member

    thanks codemonkey, I agree with you on the Logic meter. I'm definitely going to be testing some of these free meter plugins against my analog LED meter to see how they perform
     

Share This Page