You are here

Need advice on best practices for batch-processing VO audio

For everything after the recording stage: hardware/software and how you use it.

Need advice on best practices for batch-processing VO audio

Postby VillainsOfLeisure » Fri Dec 08, 2017 12:07 am

Hello! Part of my job is to create post processing presets for the VO talents that my company works with. The job fell to me as I was the only one around with audio recording experience. The goal of the presets is to normalize, reduce noise, de-ess, and compress each file uniformly. The set of presets essentially form an algorithm that must deal with anomalies such as extreme transient spikes.

I currently use only the stock plug-ins available in Adobe Audition. A typical preset is composed of the following plug-in sequence:

1. Single-band compressor set to eliminate anomalous transient spikes as determined by a manual examination of VO talent's average peak decibel level. (This is the weakest link because it heavily relies on consistent recording levels from the talent, which is out of my hands.)
2. Normalize to -3db
3. De-esser
4. EQ
5. Single-band compressor to improve dynamic consistency
6. Noise reduction
7. Normalize to -3db (set as such to match desired Premiere Pro output levels)

I've found this works decently as long as the VO talent keeps a consistent recording setup project to project. I would be very glad to hear feedback on how I could either improve or replace this process to better deal with anomalies. All plugins are stock. Thanks!
Posts: 2
Joined: Thu Dec 07, 2017 11:45 pm

Re: Need advice on best practices for batch-processing VO audio

Postby Mike Stranks » Fri Dec 08, 2017 11:07 am

Welcome! :thumbup:

I'm afraid that this is a "Well, I wouldn't start from here..." type answer...

I do a lot of work on single voice recordings and do use batch-processing for certain things. However, I wouldn't trust 'automated/preset' approaches for, for example, de-essing/EQ, compression and noise reduction. To be effective and not introduce processing artefacts I find I do need to listen and then adjust controls as appropriate so that the correct amount of 'processing' is applied.

Equally, to get the best compression/limiting I find that it's often more effective with a 'limit-normalize-limit' approach, but again that's dependent on the specific recording.

Of course, you are where you are and the only option open to you in response to "Jump!" may be to respond "How high?". FWIW, I don't find anything alarming in your sequence of processing. And if you have to go the automated route then Audition is as good as any other option IMHO.
User avatar
Mike Stranks
Jedi Poster
Posts: 5565
Joined: Fri Jan 03, 2003 1:00 am
Location: Cirencester, Gloucestershire, UK

Re: Need advice on best practices for batch-processing VO audio

Postby The Elf » Fri Dec 08, 2017 11:19 am

I'll defer to Mike's vast amount of experience in this field. He knows his stuff.

The only additional I'd offer for consideration is dynamic EQ. I like the benefits of dynamic EQ at the best of times, but when you're having to deal with audio of widely varying quality it may help to mould a degree of consistency.

This said, it depends on whether a dynEQ is available in your stock tools.
User avatar
The Elf
Jedi Poster
Posts: 11466
Joined: Tue Aug 14, 2001 12:00 am
Location: Sheffield, UK
An Eagle for an Emperor, A Kestrel for a Knave.

Re: Need advice on best practices for batch-processing VO audio

Postby Wonks » Fri Dec 08, 2017 12:02 pm

I'd start the process with a high pass filter, (typically in the 100-Hz to 200Hz range), cutting out any unwanted low-end sounds that may be in the recording and could unnecessarily trigger the compressor etc. The setting will very much depend on the voice (deep/high pitched, male/female etc.), so it's not something you can automate, but you could have a series of presets per VO talent, if that would be acceptable to the powers that be (rather than a one-size-fits-all solution). You could probably then tailor some of the pre-sets to better suit the VO talent's style.

Otherwise, at least use a high pass filter that fits the lowest common denominator e.g. say 80Hz, as it probably will clean up a lot of low-end noise that you aren't aware's there.
User avatar
Jedi Poster
Posts: 6084
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Now even grumpier than Ivan in his heyday.

Re: Need advice on best practices for batch-processing VO audio

Postby Mixedup » Fri Dec 08, 2017 12:14 pm

What Mike said — this is not the route to the best results, whcih require manual input. But you may be able to take yourself so far with carefully crafted artist+mic presets. As for your signal chain, the one thing I'd say is that you should achieve better (or at least more consistent!) results if you have more consistent material going into your compressor, because it will mean the threshold (which you really should set manually) is likely to be that bit closer to its optimum setting.

That means I'd do any HPFing, de-noising and de-essing first (rumble and loud esses can obviously trigger gain reduction... while noise will be brought up in level relative to the loudest bits), or if not then at least tweak the compressor's side-chain to be less sensitive to LFs/essing/other noise.

If you're dealing with different mics for different takes that need editing together, you could try match-EQ at this stage too. As the frequency balance is more uniform, that's a good thing for the compression stage — again, the side-chain circuit is reacting to more uniform input material.

Now think about *loudness* (LUFS) of the material going in to the compressor, because that affects where you want to set the threshold. You can normalise loudness, preferably after any high-pass filtering/de-essing/de-noising, but before compression. That way, the compressor will always 'see' similar levels, which makes it less likely that the threshold setting is wildly out — though how consistent this is will depend on why the levels vary in the first place (eg, is it preamp gain, or proximity to/distance from the mic, or a different mic, or a wildly different frequency response for some reason) and what long-term level variations there are (does the VO talent get louder/quieter as the recording goes on?).

Essentially, now I come to think of it, something like that goes on in the first stage of Sound Radix's Powair compressor, so maybe investigate that. Maybe also think about a compressor with a gain-reduction limiter, that prevents it getting overenthusiastic (eg Elysia Mpressor has this).

You might then try another pass of match-EQ post-compressor (to a different, post-compression reference). I don't know if that will be worth it, but it's probably worth investigating.

And finally, some limiting of errant peaks (there's no problem the peaks going into your compressor earlier, as long as the compressor is set to react to average levels, which I'd definitely recommend given that it will sound more forgiving) to give you enough headroom for a final stage of loudness normalisation. Then dither as you bounce.

I can't guarantee this would work... as (a) it's all based on theory and not a huge amount of practice, at least not for professional VO, and not in terms of batch-processing... (b) there may be rather too many stages in here (eg. one match-EQ pass should be sufficient) and (c) even with the same artist, same mic, and same recording space, you're potentially going to be dealing with very different performances, with a voice sounding different at different times of day etc, so all the EQing and compression in the world may not atone for that.

But the above is where I'd start my own investigations if set the same task!
User avatar
Jedi Poster
Posts: 4201
Joined: Wed Sep 03, 2003 12:00 am
Location: Laputa

Re: Need advice on best practices for batch-processing VO audio

Postby VillainsOfLeisure » Fri Dec 08, 2017 5:03 pm

Thank you all for the great information! I am grateful for the post-processing education. I will be experimenting with these suggestions for sure. For those interested, I am currently creating individual presets per VO talent.

This forum is great. Thanks again!
Posts: 2
Joined: Thu Dec 07, 2017 11:45 pm

Re: Need advice on best practices for batch-processing VO audio

Postby James Perrett » Fri Dec 08, 2017 9:13 pm

The one thing I would add is to change the position of your noise reduction from 6th to first. Audition has a very powerful noise reduction system which, in expert hands, can make a massive difference but it relies on the noise level being constant throughout the recording. As soon as you start compressing or making other level changes you run the risk that you won't be able to reduce the noise as effectively as you would if you were working on the original material.

To achieve your aim of a semi automated process you will also need to ensure that the VO artists don't change their setup without telling you and that they give you a consistent sound and level from day to day. Maybe it would be worth having them say the same stock phrase at the start or end of each session which you could use to match up the settings you need (the vocal equivalent of line up tones).
User avatar
James Perrett
Posts: 7533
Joined: Mon Sep 10, 2001 12:00 am
Location: The wilds of Hampshire
JRP Music - Audio Mastering and Restoration. JRP Music Facebook Page

Who is online

Users browsing this forum: No registered users