Best practices for processing vocals

E

eco_bach

Well-known member
Hi
Can anyone suggest some best practices for processing vocals if I have a high noise floor?
If I need to increase the volume, what do I do...amplify, normalize, use the multiband compressor or...

And if I am applying a graphic equalizer, should that come before or AFTER the volume has been raised?
 
I apply eq after I have raised volume (normalize), mainly so I can see how it affects the waveform, not because I hear a difference.
Volume nearly always gets adjusted again anyway. I prefer limiters to compressors unless its an instrument.

If its narrative and actors and your actors are trained then ADR can sometimes work out cheaper - you just got to do the math on man hours.
 
I use - in this order - noise reduction, EQ, noise reduction, EQ, compressor/limiter. Of course, there are many situations where I will change the order, that's just a starting point. There are no "cookie cutter" solutions, every situation is on a case-by-case basis.

The best solution is to capture the sound properly in the first place.
 
'The best solution is to capture the sound properly in the first place.' agreed but you work with what you have.
Certainly no shortage of options. Curious why you apply noise reduction and EQ twice?

After going thru the Audition manual, have a better overall understanding. The goal is to keep everything as much as possible in the Multitrack editor so your effects and edits are non destructive and only saved as a session file. The exception would be any diagnostic fixes such as pop click removal and noise reduction in the waveform editor. Chadfish thanks, but surprised if iZotope RX can do anything not possible using the Spectral Viewer in Audition

As well 2 non-destructive voice specific effects (Vocal enhancer and Speech Volume Leveler) as well as the Parametric EQ effect can be used for vocal enhancement and noise reduction mitigating the need for destructive noise reduction waveform edits.

So I guess one path to take would be to see first if sufficient noise-hiss reduction can be accomplished using non-destructive means first.
 
Last edited:
eco_bach said:
Chadfish thanks, but surprised if iZotope RX can do anything not possible using the Spectral Viewer in Audition

As well 2 non-destructive voice specific effects (Vocal enhancer and Speech Volume Leveler) as well as the Parametric EQ effect can be used for vocal enhancement and noise reduction mitigating the need for destructive noise reduction waveform edits.
Click to expand...

All noise reduction algorithms are not created equal. Audition processes audio as Azden mics pick up sound. Yes it does it, but it's not as good as a more professional solution. Do you like the sound of crickets chirping?
 
eco_bach said:
Thanks...Will have to give IZotope a try[FONT=arial, sans-serif]...[/FONT]
Click to expand...
Trial Rx doesn't allow you to do as much as I remember. But yeah, like Chad said, audition is no where near the caliber of Rx. Only problem Rx 2 Advanced is around 1400 bucks.
 
No reason to get advanced. Well there is some but if you need advanced then you probably also need a sound editor. Most of those "vocal enhancer" type plugins you mentioned are really more suited to pop music and quick and dirty ENG than film sound. You don't really want to "level" the sound, you also don't want to do a one pass fixes all noise reduction pass. And your right you want to keep the destructive bits to a minimum if possible. I don't know the app your using but in Pro Tools what I would normally do is duplicate the region (non destructive) and put it in sync on another track then disable that region, and then do the destructive work on the original region. In PT anytime you do something destructive you have the option to create a new file rather than overwrite the original and I always work with that set. So I can always go back to the original if things go wrong.

You also want to A/B noise reduction with the original. The rule is to keep asking yourself "is it better, or just different?".
 
eco_bach said:
Curious why you apply noise reduction and EQ twice?
Click to expand...

Noiz2 said:
... you also don't want to do a one pass fixes all noise reduction pass.
Click to expand...

That's why I use NR twice. The first EQ is used as a notch filter. Then I EQ for sound. So it's noise reduction, EQ/notch filter for noise reduction, more noise reduction, EQ for sound, compressor/limiter (99% of the time its limiter only). Every situation is different. Sometimes the first EQ and/or second NR isn't needed.
 
Last edited:
Thanks Bob!
Is it recommended to avoid normalizing or amplifying the sound signal, and instead use a compressor?

My understanding is that normalizing-amplifying increases the entire dynamic range, hence raising the level of any noise as well.


Notice you leave compression for last.
Also, if adding some reverb, where in the post chain would you put it?
 
Eco_bach,

99% of the time your goal in production sound is make it intelligible, then make it fit in. This is why there is no fixed chain of FX. You will use EQ and Noise reduction and possibly multi band expansion to cleanup and reduce noise, but mostly a lot of editing. Then you may use EQ and possibly some compression and volume automation to make it as understandable as possible. While doing the first two you have to keep in mind that it needs to sound like you didn't touch it and you had the best mixer boom person on earth.

In the make it fit in category you might use a bit of reverb to match takes, maybe. In the final part of the mix, after everything sounds really good then you may end up using limiting, compression and possibly a small amount of reverb depending on the scene. As a rule the less you do the better. But it's not music. Your not putting reverb on a voice because it sounds better your putting reverb on because they shot the close up in a different room or it's shot in a living room but it's supposed to be the flight deck of the enterprise.

The exception is some weird flick where your screwing around with the dialog for artistic reasons.

At the end of the day the right way to do it 99.9% of the time is the way where none of the audience thought you did anything and wonder why you even got a screen credit.
 
Voice operates within specific frequencies. So first off you can roll off everything below 120 Hz and then gradually roll off everything above, say, 16500. Too much roll-off and it will sound dull or tweaky. Adjust those settings, applying watever EQ you like to bring out the voice sound--the more "voice" in the EQ, the less of everything else. Female voice will be roughly an octave higher, so you can start your roll off even higher if you like. After that, the more expensive a program you use, the better the results, many of these use a noiseprint from a sample where there is "just noise"--the space between events.
 
I'm gonna go ahead and disagree with the idea of automatically rolling off everything under 120hz, and that not affecting the human voice. 80hz maybe. I say there is no set rule. You must use your ears.
 
120 would probably work with a bunch of voices but that is in the zone where you will start to hear it on most people. But as Chadfish kind of says there are plenty of voices that go below 120 so it really depends on the voice.
 
Hey something happened! I think I'm having a flashback or something. My color is all different! Anyway, I can hear a difference at 80hz on my voice, or most males.
 
You must log in or register to reply here.
Back
Top