Skip to main content
K
kdoc2
Inspiring
August 3, 2017
Answered

Loudness and Clipping

  • August 3, 2017
  • 2 replies
  • 6830 views

For a voice-over (narration) of a slideshow, once the audio clip has been edited initially, how is it best "finished" for acceptable loudness and avoidance of clipping or overly soft (inaudible) speech? Should I use Normalize and/or Hard Limit? Should I simply use a broadcast standard (even though I'm not broadcasting), such as in the Match Loudness for ATSC or ITU. Note that I don't know precisely which settings to choose, so please include exact recommendations.

If there is a written or video reference which answers these questions, I'm also willing to read or watch it/them.

Thank you

Keith

    This topic has been closed for replies.
    Correct answer SteveG_AudioMasters_

    Two questions related to that:

    I only asked about that because Audition hinstruction seems to make a big thing out a final submix and Master, and i didn't know if that would help or was necessary.

    Regarding the music track. I note that Essential Sound doesn't do much with Music. No tweak of EQ or Compression. The music I was given to use needs at least to be increased in volume (perhaps Normalized to -3 or -2). Could you suggest a slightly more helpful approach if I feel the music sounds sub-par.


    kdoc2  wrote

    I only asked about that because Audition hinstruction seems to make a big thing out a final submix and Master, and i didn't know if that would help or was necessary.

    The thing about Audition is that it does a heck of a lot of things - and the chances are that for any given job, you won't have scratched the surface of it, so I don't really know why they picked out that bit in particular either. The reason for this is dead simple - no two audio jobs are ever the same! It's the same with EQ and compression - there's no absolute law says you have to use it - you make decisions based on what your ears tell you.

    There's nothing to stop you normalizing the music track to -2dB in Waveform view, and then doing the final level tweaks in Multi-track. I can't actually tell you specifically what to do, because I don't have access to your audio...

    Probably the clearest explanation of what to do with Essential Sound is here: Get started with audio mixing |

    It's intended for video editors who claim to have little knowledge of sound treatments, but the steps seem pretty easy to carry out.

    2 replies

    S
    ScooterVoice
    Participant
    February 22, 2023

    HELP!  I'm probably posting in the wrong place and apologize but . . .  My audio is clipping and becoming distorted in Audition at about -3dB.  Why is this happening?  I can't even get the audio to hit -1 or 0 unless I normalize, and then I'm detecting mild distortion on tracks normalized to -1.  It's almost like a setting, but I can't find the answer.  Please help.    

    ryclark
    ryclark
    Participating Frequently
    August 4, 2017

    Your final operation would be to Normalize your audio to a reasonable level, say around -3dB. However before that stage it is possible that a bit of Compression be used to reduce the dynamic range of the commentary. Try some of the Voicover presets from any of Audition's Compression effects.

    See also the last post in this thread Editing from start to finish help

    K
    kdoc2Author
    Inspiring
    August 4, 2017

    Thanks, I've kinda come to that too, without understanding all the nuances.

         1, Does one compress and reduce dynamic range first so that a normalization will give the quieter parts more volume after normalization?

         2. Do we use -3 db Normalization mainly because sound on other tracks might result  clipping?

         3. If there is NO audio when the narration is going on, would one use -1 or -2 db Normalization preferentially?

         4. Does one then finish off yet further with Match Loudness or Radar in order set ITU or ATSC levels, or isn't that necessary in a non broadcast production? Will either or both of these themselves take care of what compression is doing?

         5. I note, btw, that the Radar does not contain the ITU etc-1770 preset. Is the main advantage of Radar over Match Loudness that one can see if a part EVER becomes too soft or loud? (But then, without using Match Loudness, is there still fairly good matching?)

    I appreciate all comments

    kdoc

    SteveG_AudioMasters_
    Community Expert
    SteveG_AudioMasters_Community Expert
    Community Expert
    August 4, 2017

    Oh, where do we start...

    Generally you don't need to compress a narrative as such - if you want to make it effectively louder you limit the peaks, which means that the overall level can get closer to 0dB without overloading. By and large, you can take 6-9dB of peaks away without anybody noticing, and that's just using a flat limit at that point - no fancy compression ratios necessary. When you've squashed those peaks, you can just normalize up to about -1dB, and that will be fine initially. Yes you can use the Hard Limiter, but I would use it without makeup gain. So the correct process for doing this is initially to normalize your file to 0dB (because this is the reference point that the Hard Limiter uses), and then apply Hard Limiting so that the Maximum Amplitude is -9dB, Insert Boost is 0dB, the Lookahead time is 7ms and the release time is 200ms (fine for speech). After you've done this you can normalize again up to wherever you want the max level to be - as I suggested earlier, about -1dB.  You don't want to use the Input Boost, really - this tends to 'harden up' the sound somewhat.

    The huge advantage of doing it this way without actual compression of the whole of the audio is that the noise floor doesn't bounce up and down, which is really distracting. All that happens is that it moves up uniformly by 8dB. Also, if you now think that's too high for a noise floor, it's possible usually to reduce this using NR without a load of artifacts.

    And if you aren't having to be straitjacketed by a broadcaster, leave the whole business of ITU, ATSC etc strictly alone, and use your ears; you will end up with a better, more consistent result. In general, you need to keep the narrative at a consistent level, and alter everything else around that - altering narrative levels is pretty much a no-no. So if you set the narrative peak to be -1dB, everything else will follow suit either beneath this, or, if there's no narrative, at no higher a level than -1dB either. That way, you don't have people leaping for the volume control all the time.

    HTH...

    Terms & ConditionsAccessibility statement
    Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices