iZOTOPE RX 6
The sixth full version of the spectral audio editing and repair package has used machine learning to make its algorithms better than ever…
Scoring 10/10 in all five of its reviews to date, iZotope’s spectral audio editor and restoration suite is now well established as an industry standard. Facilitating targeted processing across both the time and frequency domains via a Photoshop-style spectrogram editor, it offers the usual cropping, cutting, pasting, gain-changing and all the rest of it, as well as a potent array of restorative processing modules: De-hum, De-noise, De-click, etc. Many of these are also included as VST/AU/AAX/RTAS plugins, and each new version of RX has added more to the list. The last, RX 5 (10/10, 225), brought us the De-plosive, Leveler and Signal Generator modules, for example, as well as the Instant Process tool (immediately applying the settings from a chosen module when a selection is made) and module chaining.
With machine learning now part of iZotope’s developmental technology base – thanks to its
COMPUTER MUSIC
first outing in the development of their Neutron mixing assistant (9/10, 238) – version 6 introduces a range of previously unrealisable corrective tricks to the RX box. It’s available in the usual Standard and (much more expensive) Advanced versions, as well as the all-new Elements edition for those on a budget. We’ll touch on the differences between them in this review, but for the full breakdown, check out the iZotope website.
The machines are coming
In their ability to separate unwanted noise and the signal you want to hear, the machine learning algorithms go deeper than the previous FFT processing did, enabling the software to ‘understand’ which is which, regardless of their dynamic profiles, frequency range crossover and chaotic complexity. In RX 6, the fruits of this particular algorithmic approach are two of a trio of new dialogue-rescuing tools: De-rustle and Dialogue Isolate, available – along with the third, De-wind – only in RX 6 Advanced.
Dialogue Isolate does a very impressive job of pulling spoken word material out of its ambient environment, even if that includes other people talking in the background. A slider sets the ‘strength’ of separation between Dialogue and Noise, while Gain sliders control the loudness of both; and depending on the source material, you’ll get at least profound reduction of the background signal if not total voice isolation. At worst, it’s incredibly effective; at best, it’s truly astonishing.
De-rustle targets the rubbing noises and thuds of lavelier mics on clothing, fading them away seamlessly via the manipulation of just two sliders: Reduction Strength and Ambience Preservation.
As great as De-rustle is, we’re – if you’ll pardon the pun – even more blown away by De-wind, which does an incredible job of
removing wind noise and, particularly, rumble. As the Crossover Frequency slider rises, high frequencies in the voice start to suffer a little, but the most important thing is that the sonic turbulence below that cutoff is essentially switched off.
Let it De-bleed
The De-bleed module (Standard, Advanced) is dedicated to the removal of headphone leakage from recordings of vocals and acoustic instruments, but, alas, not drum kit multitracks. With the audio to be de-bled and the timealigned source of the bleed loaded up, RX 6 learns the profile of the latter (an offline process) in order to extract it from the former. Once again, the results are flawless.
Further oral salvation comes in the shape of the new Mouth De-click and De-ess modules, and the integration of Nectar 2’s (9/10, 200) Breath Control module (Standard, Advanced). Mouth De-click nails lip and tongue noises with aplomb, while De-ess expands greatly on the Leveler module’s Ess Reduction slider, offering Classic and Spectral algorithms, so you can banish sibilance the traditional ‘broadband’ way, or get more frequency-specific. Breath Control (superseding the Leveler’s slider of the same name) lets you apply up to 100dB of attenuation to those inhalations and exhalations, which are detected with uncanny accuracy.
Last but not least in the vox department, the De-plosive module (Standard, Advanced) has been improved under the hood (the difference is small but sometimes noticeable) and plugin-ised, and the old De-noise module is now split into separate Spectral and Voice De-noise modules (Standard, Advanced and, for Voice De-noise, Elements – yay!), the second with discrete Dialogue- and Musicoptimised algorithms.
6 is the best
Yes, it’s full marks once again for iZotope’s unassailable restoration suite. How could it not be? Everything that was already in place is still as powerful and relevant as it was when RX 5 hit the streets, and the new stuff added for RX 6 is simply knock-out. For those working in post production and dialogue, particularly ‘in the field’, this is an utterly essential upgrade, even taking into account its continuing lack of support for multichannel audio. And while there’s perhaps less here of direct relevance to musicians, the vocal clean-up modules are superb, and, of course, the spectral editing at the heart of RX is still a huge selling point.
What we love most about RX 6, though, is that it’s just so easy to use, despite its potentially intimidating remit. Balancing the inevitable artefacts that come with any serious repair algorithm against quality of results is a theme throughout, but you always feel empowered rather than intimidated by the possibilities, thanks to RX’s approachable, brilliantly aggregated interface. And as for those results, when it comes to audio fix-up, just like versions 5, 4,3, 2 and 1 before it, RX 6 represents the gold standard.
“What we love most about RX 6 is that it’s just so easy to use, despite its potentially intimidating remit”