marcsoucie

Q: Difference between audio peak on clip and peak on audiometer.

After applying audio level modifications on the clip and applying compression and limiting to the clip, my audiometer shows around -6db. it does<t clip. However, the audio waves on the clip are coloured red throughout. In other words, my clip audio waves are showing clipping while the audio meters are not. Which one is telling me the truth ?

Final Cut Pro X, Mac OS X (10.7.5)

Posted on May 14, 2016 1:59 PM

Close

Q: Difference between audio peak on clip and peak on audiometer.

  • All replies
  • Helpful answers

  • by Alchroma,

    Alchroma Alchroma May 14, 2016 4:21 PM in response to marcsoucie
    Level 6 (18,916 points)
    Video
    May 14, 2016 4:21 PM in response to marcsoucie

    The waveforms may be your more accurate indicator.

    I've seen issue at times where the audio meters seem to lock off at 0db and never go higher.

    Hitting the J key will play the footage backwards and can make the audio meters read to reflect the waveforms.

     

    BTW: What type of footage is in use?

     

    Al

  • by BenB,

    BenB BenB May 15, 2016 7:34 PM in response to Alchroma
    Level 6 (9,836 points)
    Video
    May 15, 2016 7:34 PM in response to Alchroma

    Waveforms turn red long before they'r actually clipping.  I find the timeline waveforms to be rough estimations, nothing close to accurate.

  • by marcsoucie,

    marcsoucie marcsoucie May 15, 2016 9:14 PM in response to BenB
    Level 1 (4 points)
    Video
    May 15, 2016 9:14 PM in response to BenB

    My impression is that the waveforms indicate what the output would be if no effects (compression + limiting, etc...) were applied to the sound.

  • by Alchroma,

    Alchroma Alchroma May 15, 2016 9:46 PM in response to BenB
    Level 6 (18,916 points)
    Video
    May 15, 2016 9:46 PM in response to BenB

    BenB wrote:

     

    Waveforms turn red long before they'r actually clipping.  I find the timeline waveforms to be rough estimations, nothing close to accurate.

    Agreed.

    However, sometimes the metres will LOCK off at 0db and go no higher yet operate normally below 0db. In this situation it would be unwise to rely on them.

    Not sure why this occasionally occurs, only that it does. This seems to be the marsoucies' predicament.


    Any thoughts Ben??

     

    Al

  • by Tom Wolsky,

    Tom Wolsky Tom Wolsky May 15, 2016 10:47 PM in response to marcsoucie
    Level 10 (118,107 points)
    Apple TV
    May 15, 2016 10:47 PM in response to marcsoucie

    No, the waveforms normally change with effect processing.

  • by fox_m,

    fox_m fox_m May 15, 2016 11:39 PM in response to marcsoucie
    Level 5 (5,502 points)
    Video
    May 15, 2016 11:39 PM in response to marcsoucie

    Which one is telling me the truth ?

     

    They both are.  They're on two different scales. The audio waveforms in the storyline are referenced to -6dB. The output meters are referenced to 0dB. They're both referenced to the loudest sound that won't be clipped (except there's some "headroom" allowed to handle "peaks".) If you're over -6dB, FCPX is telling you your audio is too loud. It is correct. It is. Believe it. 0dB is actually the "threshold of human hearing".

     

    The output meters are there for the total combined output of additive audio signals as well as effects applied to those audio signals. The only audio you should allow over 0dB (output meters) are special sound effects (like explosions, etc.) but you shouldn't have to push those too far. 0dB is *very* loud.

  • by marcsoucie,

    marcsoucie marcsoucie May 16, 2016 2:06 PM in response to fox_m
    Level 1 (4 points)
    Video
    May 16, 2016 2:06 PM in response to fox_m

    If the clip waveforms are showing clipping (are red) but my audiometers show -8db for example, am I clipping ? In other words, which indicator is telling me the truth ?

  • by fox_m,

    fox_m fox_m May 16, 2016 5:11 PM in response to marcsoucie
    Level 5 (5,502 points)
    Video
    May 16, 2016 5:11 PM in response to marcsoucie

    Ok... if you've clipped audio - you can hear it. It is *unmistakable*. If you have imported audio that is already clipped, it's ruined, there is nothing you can do to fix it. If you're worried about clipping, your audio is already way too loud.

     

    The storyline clip waveforms are NOT showing clipping... they're showing you "the warning track". Clipping is a flat line at the top of the waveforms.

     

    Screen Shot 2016-05-16 at 4.54.27 PM.png

    This is seriously clipped audio - it will sound like the audio is routed through a guitar effect "Fuzz box".  It is being distorted and it is ugly.

     

    Here's the same audio at its recorded output: the signal goes all the way to the top, touches it in a few places, but it is not clipped because non of the peaks flatline against the top of the range. [This audio, however, will "pop" the clipping indicators in the output meter - just because it crosses 0dB - it's still within the "headroom" range... but just a little too "hot".]

    Screen Shot 2016-05-16 at 4.56.10 PM.png

    When I apply this music to my projects, I dial down the gain so that there is no (or very little) yellow (and no red):

    Screen Shot 2016-05-16 at 4.59.42 PM.png

    This is easy to do in FCPX - just grab that "0dB" line in the audio and drag it down (you'll see feedback in -dB amounts going down, or +dB amounts dragging upward.)

     

    When I do this simple thing, I know my audio is "golden". It is right where it's supposed to be. FCPX is easy that way. This should place my OUTPUT audio levels at about -6dB peaks. Average peaks should be in the -12 to -10 dB range. This is "comfortable listening" levels, not likely to cause my listeners physical pain (there is an audio pain threshold - somewhere around 120dB - where 0dB is referenced from the threshold of hearing... in the real world.)

     

    FCPX is looking out for you. The waveforms are a visual clue to potential problems. It is telling you that when you see yellows and reds, that audio is getting too loud *for output*. Not that they are too loud as the source. Not that they are clipped (the only indication is the "flat-lining" I pointed out above.)

     

    The Output meter is referenced at 60 times (+6dB) louder audio than the waveform audio. You should not cross the 0dB level on the Output meters. (Your peaks should hover around the -6dB mark, max, which corresponds with the "recommendations" FCPX provides for you with the waveform indicators.) There is, however, approximately another 60 times louder audio "headroom" for peaks over the 0dB mark that *are not likely to clip* — and clipping will definitely and finally occur at any signal louder than +6dB referenced on the output meter.

     

    If you're making a direct mapping between audio waveforms and the output meter, it looks something like:

     

    Waveform                                             Output

         0dB                         ⟹                         -6dB

       +6dB                         ⟹                         0dB

      +12dB                        ⟹                       +6dB (clipping occurs)

    (active control [gain])                   (passive response meters)

     

    Try not to be so concerned about clipping - bring your audio output to under 0dB (output meter). You really should do some research and find out the ranges that are "acceptable" for audio levels for different types of sounds, like "normal conversation", different types of music (not everything you hear is heavy metal head banging), etc. You do not always want to "push the limit". If you need "apparently louder" audio, you need to learn how to use compressors which can raise the "low end" without driving the high end over the limit — the audio will be apparently louder/overall loudness remains the same.

     

    If this is not clear to you, perhaps somebody else can explain it better. Audio is not always an easy subject to grasp (and proving a little difficult to explain). Having the right explanation for you can make all the difference.