Hi all,
 
I am using some spoken audio in a project and I want to bring up the quieter parts of the dialog. I didn't record these and can't recreate them, so I have to do the best I can to make them intelligible in the song. I have been playing with Waves MaxxVolume, but I wonder if there is a better way...
 
Also, using MaxxVolume brings up the noise floor and I need to address that. I'm thinking maybe a pass with a noise suppressor before using a volume tool?

not to be obtuse,
but couldn't you simply ride the fader volume with automation, and just mix it higher?
 
maybe that, and a combination of a limiter used as an upwards compander?
 

Absolutely - I could do some volume automation. I guess I was looking for a quicker fix, but that is the fallback plan! :)

When voice overs are recorded elsewhere and they vary a bit in their levels and dynamics what I usually do is make a copy of the voice over and open it in a separate editing program.  eg Adobe Audition.  Then I put a VU over the voice over and watch what the meter is doing on the louder passages or the most normal sections.
 
I adjust the gain of the overall track so the VU reads 0 dB VU in the normal sections.  Then I go in and edit softer areas.  By adding gain here and there as needed.  You do have to be slightly careful because bringing a soft section right up to the same level as a normal section can sound unnatural and you want to avoid that if you can.  Level variations are there for a reason as they add to the overall effect.  So a really soft phrase might only hit -10 dB VU compared to the 0 dB VU for other parts.  I might add enough gain to make this maybe reach -4 to -5 dB instead.  Noise can come up I agree.  But that depends on what is happening also at the time in your final mix.  Any increase in noise may be masked if you are lucky.
 
I prefer to get a vocal track ie either sung or read pretty right first in the editor before dragging that edited version out into your session.  The reason is once you have done this you basically don't need any automation and especially putting processes over the vocal track eg such as expanders etc.  These will most like just ruin stuff and make it worse.
 
All that is often required when you edit properly is some EQ to add some tone and some very light compression performing only light conditioning to the overall sound.  I get the best result doing it this way.

Thanks Jeff!
 
I'm actually using some recorded audio from an interview with Steven Hawkings, so no chance of him redoing it for me! ;)
 
I think the manual technique is the best way to go. I'm not using a huge amount of it, so it will be worth the effort to do it right and not lean on quick-fixes that don't quite work.

For parts that are too loud or quiet on a track I like to do gain adjustments as needed from the menu option when possible.  It's time consuming and tedious but very precise and "clean" in my view.  I also use it for transients on recorded instruments that are off the chart so that compressor doesn't get crushed for that short time.

Thanks! :)

smallstonefan
Absolutely - I could do some volume automation. I guess I was looking for a quicker fix, but that is the fallback plan! :)

Waves Vocal Rider

My workflow for this sort of thing is as follows.

1. Compression. Use a good compressor to automatically roll off the louder bits thus allowing you to set the basic level higher which makes the quieter parts not so quiet while keeping the loud bits from getting out of hand.
2. Brick wall limiting. If compression isn't enough to get it done, I add a limiter. This prevents the loudest bits from being any louder than most of the much quieter bits. Used with compression, you can effectively control the volume of most source material while still achieving a relatively natural sounding result.
3. If compression and limiting don't get it done, I bring up the rear with fader automation for the worst bits. Sometimes the fader I automate is the volume fader. Sometimes its one or more eq bands. Whatever it takes to get it done in a natural sounding way.

JohanSebatianGremlin
My workflow for this sort of thing is as follows.

1. Compression. Use a good compressor to automatically roll off the louder bits thus allowing you to set the basic level higher which makes the quieter parts not so quiet while keeping the loud bits from getting out of hand.
2. Brick wall limiting. If compression isn't enough to get it done, I add a limiter. This prevents the loudest bits from being any louder than most of the much quieter bits. Used with compression, you can effectively control the volume of most source material while still achieving a relatively natural sounding result.
3. If compression and limiting don't get it done, I bring up the rear with fader automation for the worst bits. Sometimes the fader I automate is the volume fader. Sometimes its one or more eq bands. Whatever it takes to get it done in a natural sounding way.

If you do the editing approach as I suggested in my post you don't need any of this.  It will sound better and not use a single plug-in.  No automation needed most often.  Hitting a compressor with varying levels from soft to loud is not getting the best out of the compressor.  It is working over a higher range.  When you do light conditioning over a well edited track like this the compressor has an easy time of it and sounds good in the process.  In voice over work it is not even a bad thing to bring quieter parts up to close to the ref level.  So it does not get lost.  When a shift in level is wrong though it sounds unnatural and quite easy to detect.  You just back off the gain changes a little and it will end up perfect.
 
If a voice over is happening over music and effects it needs everything it can get for a nice even and clearly audible level.