Yellow image with lots of grainy small black marks to represent noise.

What is the best noise reduction plug-in for voice work?

Noise reduction is an often talked-about topic for voice artists, and I’ve made my feeling clear about it. I don’t believe in it. Yet here I am writing a blog about various noise reduction plug-ins.

WTF? Read on, Dear Reader, Read on.

Why I don’t believe in noise reduction

I do believe in noise reduction. In the right place.

Noise reduction is a very powerful tool and noise reduction plug-ins are getting better and better. In my work I do use noise reduction software fairly regularly. But I do not advocate that voiceovers should be using it on a daily basis. As you will hear there are always compromises to be made when using noise reduction, and what is imperative for voice artists is that the audio they present to clients is of the highest quality possible. This means that voiceovers should be concentrating their efforts on soundproofing and acoustic treatment to ensure their audio doesn’t need noise reduction. If you reach for your NR software on a daily basis you need to fix your studio. I can help with that, but that’s a topic for a different day.

So when do I use noise reduction?

Mainly in 3 different areas:

Firstly I do an amount of film mixing where audio has been recorded on location and needs a bit of cleaning to enhance the clarity of the dialogue – these are largely corporate films with vox-poppy or testimonial type dialogue, so ADR isn’t an option!

Secondly I use NR for some podcast work where again audio hasn’t been recorded in a studio.

Thirdly I do have clients who specifically send me audio that needs restoration for one of many reasons.

And fourthly since we’re here I do occasionally use noise reduction as a belt-and-braces measure when mastering audiobooks just to lower the noise floor by a couple of dB – no more than that.

So what am I doing here?

I do have a collection of noise reduction suites, but here and now I simply want to consider the de-noising modules – those plug-ins that remove background noise from an audio clip. I’m going to do it in a specific way – in real-time. I’m doing it like this to emphasise the usefulness of this type of plug-in for podcasters and the like who are much more likely to be working in a multi-track environment in a DAW rather than simply editing a voiceover recording which I’ve already said isn’t where I think you should be using these processes.

I have a piece of audio recorded in our kitchen that I’m running through 4 different de-noise plug-ins in adaptive mode as insert effects in ProTools. All of these plug-ins will work as offline processes as well as real-time (and some of them can be processor heavy, so offline processing can be a good option).

I haven’t tried to totally clean up the clip with these plug-ins as doing so will leave lots of artefacts in the audio, I’ve just cleaned up enough to take out as much of the background noise as I can whilst doing minimal damage to the program material, and I’ve done it as quickly and simply as possible.

With further fiddling I could have done a better job (particularly with the Acon, as it has more controls available for tweakery than the other options) but we’re often working against the clock, so I pretended that I was here. Not much pretending needed really – Helen gets cross if my blog is late.

No other processing is being applied – just the noise reduction. So let’s dive in…

Raw sound

Firstly we have the unpressed audio. Recorded far too close to a couple of other sound sources. This is specifically to demonstrate the adaptive nature of the noise reduction plug-ins. A few years ago you would struggle to find affordable adaptive noise reduction, so all you would be able to do is remove sounds of an unchanging nature – a hum, buzz, aircon unit etc. but the algorithms behind noise reduction are getting better and better and that is no longer true. But here in all its glory is me talking to a kettle.


Noise reduction plug-in comparisons

Let’s take things in (sort of) alphabetical order. All the screenshots are the setting I used in the audio samples.

Accusonus Era Noise Remover 6

accusonus.com

Screenshot of Era noise reduction plug-in.

A very simple to use noise reduction plug-in with most of the work being done by a big knob, but there are a few other controls available to fine tune things, including a second algorithm and some EQ shifting to focus the noise reduction onto different frequency bands.

There’s not much else to add to that other than that I’m not sure if this is still available. I’ve used the Era 4 suite a few times since I bought it 3 or 4 years ago but thought I’d see about a trial of this latest version for this blog and it would appear that Accusonus aren’t entirely trading anymore. That said I easily managed to download version 6 and use it, but I have no idea if I’ve downloaded a trial or the full version, or if the fact that I have an Accusonus account already made my download possible. But if you want to maybe get a free noise reduction suite, maybe it’s worth clicking the link! Let me know what happens if you do and I can update this blog. Anyway, here’s the audio.


If you feel that your editing or production skills could do with improvement then get in touch and we can arrange a 1-2-1 Power Hour session to show you some new tricks.


Acon Digital Restoration Suite 2

acondigital.com

Screenshot of Acon noise reduction plug-in.

This is a more complex noise reduction plug-in, but using the adaptive mode is still quite straightforward. The Help section of the plug-in isn’t very helpful, but head to the website and download the manual and that is a lot better.

You have 4 parameters to control, and by tweaking those you get a better degree of flexibility than you do with the one knob options. There is also an EQ section so you can weight the noise reduction to the frequency range that most needs it.

This is a very good plug-in, but of the 4 options I’m looking at this is the only one I’ve never used on a live job. This is down to one limitation that it has that doesn’t suit my workflow. I have found a workaround to this today, but for the purposes of this test I’ve left things without the workaround so you can hear what I mean.

This limitation is a strength of the plug-in in circumstances different to mine. If you listen to the audio below you’ll hear that it’s about 1 second before the noise reduction kicks in and reduces the kettle noise. This renders the plug-in unusable for the type of job I want it for. But you can change that lag by reducing the Adaptation Time. You can reduce that enough to make the noise reduction kick in instantly if you wish. However, letting the plug-in hear analyse the noise for longer will make the noise reduction more effective, so if you can leave that noisy header on your audio file the Acon will be able to do a better job. You could either chop the head off after you’ve rendered the sound if your project works that way, or use your DAW’s automation so the plug-in can analyse the noise, but you don’t hear it. This would give you a good few seconds in this sample for the noise reduction algorithm to work with.


Izotope RX10 voice de-noise

izotope.com

Screenshot of RX noise reduction plug-in.

This one will need the least introduction I’m sure. This comes with all 3 versions of Izotope’s RX suite. Again the controls are simple to use, but unlike the Acon, the help section is very good – click on the Help icon and it takes you straight to the correct page of the manual. The adaptive mode works very nicely and you can see the GUI changing the noise print and threshold in real-time as the audio plays.

If you want to find out more about RX10 click here and have a look at this blog I wrote previously.


Waves Clarity Vx

waves.com

Screenshot of Clarity Vx noise reduction plug-in.

This one comes in 2 levels. Clarity Vx, which is a one knob wonder and Clarity Vx Pro which adds all sorts of complications into the mix so you can hone in on the noise and get better results. The price difference is quite big, and being a true Yorkshireman I stumped up the cash for the one knob version. But it’s pretty impressive. It’s not exactly a one knob wonder (none of them are) as it does have a choice of 2 algorithms you can switch between (Broad 1 and Broad 2), and if you’re working on a stereo file you also get a Width control so you can alter the stereo field.


Oh, go on there. Seeing you’re still here let’s have a fifth option…

Izotope RX Dialogue Isolation

This one only comes with RX advanced and doesn’t work in real-time, but is easy to use and can do a much more comprehensive job than the regular dialogue de-noise. I do not recommend you rush out and upgrade your RX to advanced as the bits you get in there are largely useless to studio-bound recordists. I just thought I’d chuck it in seeing as you’ve been good.


Conclusion

Out of the 4, the one I use most often is Waves Clarity Vx.

For the stuff I’m doing it seems to give a cleaner result than the others and leaves fewer artefacts. Most of what I’m really using noise reduction for will be used with a music bed, or other audio streams, so a lot of the artefacts will be masked. And for the Vox-poppy type stuff you don’t want total silence anyway. Sometimes in these cases you can afford to be a little rough and ready with the noise reduction. But when it’s dry audio you lose the luxury of hiding the flaws and your audio needs to be crisp and clean.

As you’ve heard, none of these will provide that level of noise reduction with this level of background noise. If the background noise is other people talking you will have even less success. They will help the dialogue stand out from the noise enough for a lot of projects, but for your everyday voiceover recording – as I said earlier and will go on repeating until the sun withers and dies – you need to get your studio sounding good and not rely on processes such as these.