Rob Bee from B Double E holding a A3 card that says 'less is more'. He's stood against a plain pale background wearing a dark blue t-shirt.

Trimming the fat: Optimising your voice production for impactful results

What are we thinking about this month? Over-production. That’s it. That’s the introduction. No point milking it, labouring the point, over-elaborating, over-stressing, belabouring or making too much of it for the sake of it.

Here’s a situation

Helen likes cooking and so we have quite an extensive array of herbs, spices and seasonings. What we don’t do is liberally sprinkle a bit of everything in all we cook. Curried kimchi chorizo and custard, anyone?!

Less is more

And so it should be with our audio work. Our editors and DAWs all come equipped with an array of tools we can use to shape and manipulate our recordings. We can bend our audio to sound not like us at all. We can polish it so it’s flawless. We can crush it so it sounds like every little spittle pop is dripping into the listeners ear. We can make what was recorded inside sound like it was recorded outside and vice versa. But the one thing we should do over and above everything else is to use our ears.

Sometimes some of the above will be entirely what the job needs

But most of the time it would be overkill. For example there’s a massive difference between radio and corporate production.

Radio voices are typified – possibly stereotyped – by that heavily compressed old school FM DJ type sound, and that works well for radio. Radio needs to compete for your attention, and so making the dynamic range as narrow as possible so the radio pushes its way into your brain and acts as a constant noise in the background, demanding your attention and trying to pull you away from your job, the kids, the housework or whatever else it is that you’re trying to get done.

Translate that same style of production to a corporate and all that compression just screams at you when it’s something you’re already giving attention to and it actually makes it quite tiring to listen to.

Forum postings

I see quite a lot in some forums people describing their post-production schedule. I see posts about how they normalise, then high pass filter, low pass filter, EQ, compress, de-click, EQ, de-noise, de-ess, compress, de-breath, EQ, normalise, limit and then bounce. I have exaggerated a little for comic effect (come on, who laughed?) but honestly some of the process chains I’ve seen haven’t been far off that. What I say to that is simply ‘Don’t’. If you need all that processing what you actually need to do is sort your studio acoustics and/or your technique out. If you don’t need it then what’s the point of running it.

Here’s the thing…

EQ is one of the most overused processes for audio processing.

If you’ve spent wisely and got a good mic and interface and are working in a well soundproofed and treated room then you already have a good sound. If you EQ it, you’re not improving the sound you’re just changing it. Corrective EQ is one thing and with the best will in the world can sometimes be necessary (my beloved HPF is corrective EQ), using EQ to create your mix is also fine, and you may need to EQ heavily to help the voice sit right in the mix. But you shouldn’t need to – and shouldn’t – heavily EQ a dry voiceover delivery.

Here’s another thing.

Many of us own Izotope RX of whatever generation we bought and many of us do love the mouth de-click module. It is sometimes necessary to use it on a voice over – particularly long-form audio – as it can save hours of editing.

But I hear of voice artists who run it multiple times over the same audio and watch the numbers of clicks removed come down until it hits an ‘acceptable’ number – that’s not how it’s meant to be used. Most of the clicks that get removed are inaudible, so the numbers of clicks removed are always scarily high, it doesn’t really indicate the improvement RX has made to the audio, you can only tell that by listening. Ignore the numbers. Run it once. Have a listen. then, if there are still a lot of mouth clicks, maybe run it just once more. Or, run it once, then manually edit out the few mouth clicks that remain. Better still, hydrate properly etc. to avoid clicks in the first place, then you won’t need to run it at all.

No noise reduction process is perfect – there’s always a chance you’ll damage audio you want to keep, so the more you run noise reduction processes as standard the more chance there is of your audio being made worse than the raw file was to begin with. You should only reach for any noise reduction process if you really need it. And the rule still applies – try to solve the problem at source.

So how should you process?

Minimally.

For most dry voiceover all you should need is a high pass filter to lower the noise floor and make editing the plosives easier, and some gentle compression – aim for a gain reduction of 6dB with a ratio of 2:1. And that should be it.

You may need to normalise after you’ve run that processing and finished editing, depending on what the client has requested. That processing should yield you a natural sounding easy on the ear result. Yes there will be occasions when you need more than that, and that more can be applied as necessary.

The key to it all is to use your ears at every stage of your production and don’t use what you don’t need to use. That way you’ll preserve the quality of your audio and the character of your performance.

Our little peccadillos are what separate us from the AI voices

The temptation is to edit and process until we reach perfection, but people are paying for a human to read their copy. We do need to pay attention and maintain a high quality of work, but we also need to remember that we’re all real boys and girls.

In music production, they came up with a ‘humanise’ option for when programming MIDI sequencers. MIDI is how we program synths, drum machines and other electronic instruments, and we play them to a click track or manually program them. Then we can quantise our efforts so the drums all hit on the right beat and the rhythm parts are all in sync. I can’t remember how many years ago it was that a sequencer creator realised that all that made things often sound a little artificial, so brought in the Humanise function that let the beats drift a little in the way that a real human musician would.

In our VO world we need to ensure that we don’t need a Humanise button in what we produce.

I think I’ve made my point – I hope I’ve made my point.

To summarise, less is more.

You’ll get a better quality job by using your ears rather than using all the tools at your disposal. Need help with that? Give me a call.