Accentize dxRevive Pro

Can cleaning and enhancing dialogue really be this easy?

I loved Accentize’s DeRoom Pro 2 — a slick, intuitive, machine‑learning based reverb removal tool for dialogue that’s capable of great results — and their dxRevive Pro is every bit as impressive. It has a more ambitious aim, though: to extract “studio‑like recordings from any source material”. We’re still talking specifically about dialogue, but as well as suppressing ‘room tone’ it tries to remove noises and artefacts from compression codecs, while also improving the spectral balance, by applying EQ in real time and synthesizing new content to counter the effects of filtering, such as the band‑limiting used for phone calls.

Overview

As with DeRoom Pro, dxRevive Pro is authorised by iLok and available in the usual formats for Mac and Windows. There are two different algorithms to choose from, one main ‘amount’ control (to apply as much or as little of the processing as you wish), and input and output level controls, each with a meter. An on/bypass switch is joined at the top by a preset menu and A/B compare buttons. In the middle, the source and processing are visualised, and it’s possible create up to four frequency bands for which the amount of processing can be adjusted individually.

First, you’ll need decide which algorithm to use. The default is Studio, in which the plug‑in does it all: de‑reverb, noise suppression and source enhancement. The other, Retain, focuses on tackling noise and artefacts, and drops the de‑reverb side of things. Unlike something like Descript’s Studio Sound process, this is a regular insert plug‑in (it operates in real time, with a little latency) so you can run several instances in your DAW or NLE project, and can refine the settings after applying further processing — important, since compression and limiting often make artefacts more noticeable.

By comparison with the source files, the voices all sounded full, dry and intelligible.

On Test

Inserting dxRevive Pro for the first time and turning the amount knob brought a smile to my face! A reasonable degree of care had been taken to set up mics for the recording, but the extraction of a nice, radio‑style dry vocal was instant and nothing short of superb. But later, it allowed me to meld some very disparate voices, captured remotely over some pretty crappy mics in some very different spaces, into a coherent, intelligible discussion show: by comparison with the source files, the voices all sounded full, dry and intelligible. Where the speed of result is as important as the quality, then, it could prove a great option. There’s perhaps a little too much latency to be able to use it live when feeding a PA or monitor mix, but it could come in very handy for, say, journalists and news broadcasters for example, or on a streaming feed. Used ‘off the bat’ in this way, dxRevive Pro will also deliver results that will please the average self‑producing podcaster wanting to clean up guests’ phone contributions, or those looking to make the best of the audio from on‑camera or phone mics used when shooting videos for YouTube, Tik Tok and the like. If that sounds like you, you may wish to check out the more affordable dxRevive (without the ‘Pro’), which offers only the Studio algorithm and lacks the multiband facility and a few more minor features.

While dxRevive Pro is shockingly good and can make pretty much any dialogue recording cleaner and clearer, it can’t turn everything into gold in an instant. You’ll sometimes need to work harder on the multiband settings, or use dxRevive alongside other processing to get the best results. In particular, on poor recordings from VOIP systems, I could often hear artefacts in the high frequencies, where dxRevive seemed to be reacting to the effects of filtering. Reducing the amount of processing in just the high band often helped address such gremlins. Also, a little spectral de‑noising at the outset in iZotope RX10 usually enabled me to coax better results from dxRevive generally. Importantly, though, these were better results than I could attain using RX10 alone. It’s also worth noting that dxRevive can’t undo the sort of overcompression borne of some VOIP services’ auto‑gain settings.

Retain mode should not be overlooked, particularly if you own DeRoom Pro, since the latter is a touch better at the specific job of reverb removal (at the cost of greater latency). To be clear, dxRevive is very good at this, but DeRoom Pro is better and having separate control over the noises and reverb can often be helpful.

While I might sometimes choose to combine dxRevive with other tools in search of the very best results, it really is a revelation: no other single processor I’ve used to date has enabled me to ‘clean up’ dialogue recordings anything like as well as this one.