Skip navigation
Currently Being Moderated

How to normalize only HIGH vocals?

Nov 6, 2013 9:39 AM

Hi. Is any clue, any chance to got low vocals in the same height while normalizing whole audio?

I mean: I want loud vocals to be smaller but I want a quiet vocals don't chance their volume.

 

in other words:

http://img208.imageshack.us/img208/7204/oxun.png

 
Replies
  • Currently Being Moderated
    Nov 6, 2013 10:01 AM   in reply to Wermount2323

    You need to use one of Audition's Compressor effects from Effects/Amplitude and Dynamics menu.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 8, 2013 5:19 AM   in reply to Wermount2323

    Unfortunately without hearing your vocals it is impossible to say which would be best. You could try the Rock Vocals preset to start with but you will have to tweak the settings for best results with your material. You could also try some of the vocal presets from the Single Band Compressor effect where the Threshold control is probably the one that needs the most tweaking.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 8, 2013 8:49 PM   in reply to Wermount2323

    Let's go into some basics here.

     

    Normalise isn't the effect you want.  All it does is find the loudest sample in your entire file, move it to the level you've set as the maximum you want, and moves everything else the same amount.

     

    The compression effects tthat have been recommended are exactly what you want.  It will leave any level below a certain threshold complete unchanged, but then lower things above that threshold as you want/tell it to.

     

    The problem is, every file is different so there's no single preset that can do what you want.  You'll need to read up on how it all works so you can make the appropriate adjustments to the threshold, gain, reduction, etc.

     

    For your application, I'd probably try Effects/Amplitude and Compression/Dynamics Processing.  Depending on what version of Audition you have, something like "Vocal Processing" might make a good starting point but, as noted above, you'll almost certainly have to change the setting to suit your particular recording.

     

    FYI, if you're not used to compressers, you may find the graphical control...the one with the diagonal blue line...easier to understand.  If you read the scales on the graph, it should start to make sense which levels are being left alone and where (and how sharply) higher levels start to be limited.  You can also add "make up gain" to raise the whole file to where you want it but with the dynamic range modified as you want.

     

    Finally, if you want to read up on the whole theory of dynamics processing, there's a good article here on Rane Notes:  http://www.rane.com/note155.html

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 9, 2013 4:20 AM   in reply to Bob Howes

    As Bob says in essence once you have squashed the high level vocals down nearer to the lower level ones with the compressor you then need to raise the whole vocal to the final level that you want. It can all be done in the Dynamics Processor if you understand how it works. But for you it might be easier to use the Normalize effect to bring the levels to what you want after you have done Compression.

     

    Compress then Normalize.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 9, 2013 11:05 AM   in reply to Wermount2323

    Noise Reduction first always because Dynamic Processing may make the Noise levels move up and down. But the Dynamic Processing can bring back some noise as you are increasing all the lower level audio signals. So it is always good to make the best recording you can with as little unwanted noise in the first place. So look for the noise that your microphone is picking up and find ways to reduce that first to produce cleaner recordings that don't need Noise Reduction.

     

    In your Step 1 picture of the Dynamics Processor screen there is Output Gain which in the default Voice Over preset is set to 0dB. Click on that and try increasing to 6dB and use it with that setting on your original vocal file.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 9, 2013 9:53 PM   in reply to Wermount2323

    Definitely noise reduction first, otherwise your dynamics processing will make the noise level pump up and down (and much more noticeable).

     

    However, just to be an old nag, it always worries me when somebody has noise reduction as a standard part of the production process.  If the noise is that consistant, there will be ways to reduce or eliminate the noise before you even record it.  It's always better to start with clean tracks and keep noise reduction for emergency problems.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 11, 2013 12:39 AM   in reply to Wermount2323

    The trouble is that the dynamics processing is automatically raising and lowering the level of your material.  Think of it as a man sitting at a sound mixer pulling down the levels when things get too loud and putting the fader back to normal when levels are lower.  This fixes your problem with the loud bits being too loud but also means the background noise goes up or down every time there's a level change.

     

    This does two things to the noise.  First, it makes it more obvious.  Think of an air conditioner droning away in the background--you soon learn to ignore it.  However if the air conditioner fan is going faster then slower, each change stands out more.   It's the same with background noise.  It's more noticeable as the level pumps.

     

    The second thing doing dynamics before noise reduction does is make it harder for the NR feature to work.  As you know, you grab a noise sample and Audition works out from there how to reduce the noise.  If the noise level is constantly changing, Audition can't find a single reduction setting to suit you.

     

    ...but, as I said, you're better off eliminating the noise before recording anyway.

     
    |
    Mark as:
  • SteveG(AudioMasters)
    5,602 posts
    Oct 26, 2006
    Currently Being Moderated
    Nov 11, 2013 1:51 AM   in reply to Bob Howes

    Bob Howes wrote:

     

    Definitely noise reduction first, otherwise your dynamics processing will make the noise level pump up and down (and much more noticeable).

    Bob's right - that's definitely the order to do it in. NR always works best with a steady noise source.

     

     


    However, just to be an old nag, it always worries me when somebody has noise reduction as a standard part of the production process.  If the noise is that consistant, there will be ways to reduce or eliminate the noise before you even record it.  It's always better to start with clean tracks and keep noise reduction for emergency problems.

    Well if that was completely true, then I spend most of my time recording emergencies!

     

    Virtually all location recording these days is done in conditions that are noisier than they should be, and there's usually nothing that can be done about it at source. And sometimes that ambient noise is quite important, and you want to preserve it (inter-track 'silence' between different recordings in the same environment, etc.). And over long recording sessions, it varies...

     

    There are things you can do, though. Most of them are downright sneaky (to the point where I'm not going to describe them unless somebody comes up with the same situation), but for voiceovers recorded in an environment where you can control the ambience, I agree - you should be able to get the background quiet enough so that dynamic peak compression doesn't bring the noise floor up too much.

     

    Weremount2323, if you want some numbers, then consider this: If you have a noise floor that's about 65dB below the peak level you are recording at the mic (this should be relatively easy to achieve with good mic technique and correct gain settings), then squashing the peaks you have by 6-8dB and increasing the level by that amount is still going to leave you with a -60dB noise floor - which is more than adequate for voice-overs. The other thing I'd suggest is that it most specifically what you need to do is use limiter settings - this is more predictable in terms of the end result, and won't pump the noise floor.

     

    Essentially this means that when you reach a specific amplitude, all audio above that level is limited to that level. So it doesn't affect the dynamics of the whole signal at all, but just the peaks. With speech, and careful settings, you simply won't notice the result - it's just that everything can be that much louder without hitting 0dB. And your noise floor won't pump! Using the Dynamics Processor, what you need to do is use the graphic setting, and at about -8dB, alter the line so that it turns horizontal (to the right). This will limit the output to that level. The other important thing you have to do before applying this though is to normalize your file so that the peaks are at exactly 0dB, otherwise you won't know exactly where it's working. If you do that though, you'll know that the peaks are being squashed to exactly where the horizontal line is.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 11, 2013 3:37 AM   in reply to SteveG(AudioMasters)

    All true if you're talking about location recording, Steve (though, as you say, even there proper mic technique placement can greatly reduce the issues).   However, yeah, on location stuff I often end up with some NR. 

     

    However, I don't think the OP ever mentioned where he's recording and, if it's some kind of home studio, there's always things that can be done to control the usual background noise of computer disks and fans, heating/air conditioning, etc.  Yeah, I'm a bore on this but (in other forums) the number of people who record their own music or audio books or whatever and rely on NR rather than getting the tracks right to start with is a frustration for me.

     
    |
    Mark as:
  • Currently Being Moderated
    Nov 12, 2013 7:44 PM   in reply to Wermount2323

    Yeah, a limiter is totally different from normalising.  A limiter doesn't adjust the general level at all; it just stops your signal from going above a certain pre set level to prevent you getting any clipping.

     

    You haven't told us how you're recording or where you're recording, but generally you're better off just keeping your levels low enough that even the peaks don't get into clipping.  In my personal set up, I have hardware limiters for every channel and, although I put them in circuit, I try to adjust my levels so they never come into play.

     
    |
    Mark as:
  • SteveG(AudioMasters)
    5,602 posts
    Oct 26, 2006
    Currently Being Moderated
    Nov 23, 2013 1:50 AM   in reply to Wermount2323

    Wermount2323 wrote:

     

    I capture noise reductions profile, then I add a noise reduction. Ewerything is OK now. But after I do "Dynamic processing" a noise will be hearable/audible.

     

    Why? You said that noise reduction filter will be still in use.

    We didn't say that at all. We said that if you apply noise reduction first, then you will have a higher signal to noise ratio to start with before compression. But noise reduction is a single process you carry out - it doesn't run continuously. In other words, it's not a filter. So if you compress a signal after applying it, then yes, the noise floor will rise again, relative to the new tops of the signal.

     
    |
    Mark as:
  • SteveG(AudioMasters)
    5,602 posts
    Oct 26, 2006
    Currently Being Moderated
    Nov 25, 2013 2:07 AM   in reply to Wermount2323

    Wermount2323 wrote:

     

    But still "the right way" is do a noise reductions first, before use compress filter?

    Yes, because otherwise the result will be even worse! Whatever is left of the noise will inevitably have far less effect on the end result, whatever you do in the way of subsequent processing, so yes, do the NR first.

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points