I’ve used the sequential versions of iZotope’s RX Advanced stand-alone application and component plug-ins on all my post-production sessions for several years. With each new release, it keeps getting more powerful, versatile and, frankly, indispensable. The latest version, RX 6 Advanced, promises solutions to some of post’s most vexing and immutable noises that pollute dialog tracks, including wind and the sound of clothing brushing against a lavalier mic.
Other new processing modules (for the standalone application) and plug-ins aim to isolate dialog and reduce mic bleed, sibilance, mouth clicks, lip smacks and breathing noises. The new Composite View combines up to 16 active tabs into a summed spectrogram/waveform display and is useful when you want to apply the same spectral editing to all tracks at once. RX 6 Advanced can also export files in MP3 format.
Legacy processors—De-click, Voice De-noise, De-plosive, Ambience Match, Center Extract and Deconstruct—have also been improved, as has the Find Similar (event) tool. The Module List has been better or- ganized into Repair, Utility and Measurement categories, and you can save custom views of your go-to modules as presets for faster access.
RX 6 is available in three versions—RX Elements, RX 6 Standard and RX 6 Advanced. The Advanced version has the most features and highest price tag. The component plug-ins are available in AAX, Audiosuite DPM, Audio Units, RTAS, VST 2 and VST 3 formats. I used the AU plug-ins in Digital Performer 9, using an 8-core Mac Pro running OS X 10.9.5.
OFF THE AIR
One of three new modules in RX 6 Advanced that use machine learning and intelligent signal processing (the others are De-rustle and Dialogue Isolate), De-wind will reportedly only attenuate intermittent wind gusts that don’t distort a microphone; it regards constant wind noise to be part of the desired noise floor.
The module’s Crossover Frequency slider delimits how high in fre- quency (up to 1.5 kHz) the processing will act on the signal, while the Reduction slider sets the balance between wind-noise reduction and the original signal (see Fig. 1). The Fundamental Recovery slider re-synthesizes lower voice harmonics that have been obscured by wind noise, and the familiar Artifact Smoothing slider (also included in legacy noise-reduction modules) reduces watery-sounding artifacts that can occur with FFT-based processing.
I tried De-wind on the windiest track I could find in all my past post-production sessions—a female dialog track that was severely tainted by extremely strong low-frequency wind gusts that almost completely obscured the voice at times. Simply put, De-wind is a virtual hurricane shelter for dialog tracks. The module eliminated the track’s low-frequency wind gusts—gusts that, during the original session, the RX 5 Advanced De-noise plug-in could not completely quell.
A moderately loud, airy wind—naturally modulating slightly in intensity—remained. Another new module, Dialogue Isolate, dealt with that in short order. I simply boosted the Dialogue Gain slider to the max and lowered the Noise Gain slider all the way to favor speech over noise, then raised the Separation Strength slider slightly to 1.5 (on a scale of 0 to 10) to attenuate the wind. This transparently reduced wind noise to that of a light breeze (I could’ve attenuated it more), which made sense of waving vegetation in the video without sounding obtrusive.
CLOTHES CALL, MOUTH OFF
The De-rustle module uses machine learning to attenuate the sound of clothing brushing against a lav mic. Two controls are all you need: Raise the Reduction strength slider to attenuate the rustling noise more (at the potential expense of speech clarity at high settings), and raise the Ambience preservation slider to retain more of the noise floor and background ambience (which De-rustle might otherwise remove).
De-rustle was very effective at reducing lav noise that was only about 3 to 6 dB lower than—and overlapping—male speech on the same track. Turning up the Reduction strength slider about two-thirds of the way attenuated the rustling noises almost to the point of inaudibility. The track suffered only a moderate reduction in depth, and clarity of speech was not noticeably impacted—an acceptable tradeoff. Once again, I was extremely impressed.
The new Mouth De-click module and plug-in are optimized for removing the sounds of mouth clicks and lip smacks. Aside from the fact that it includes no drop-down menu for selecting alternate algorithms, its control set and operation are the same as with the De-click module.
In A/B comparisons treating an adult fe- male dialog track, RX 5 De-click and RX 6 Advanced Mouth De-click were equally effec- tive at removing extremely loud mouth clicks and lip smacks vented by an off-mic, hyperactive child (up to 6 dB louder than the adult’s speech!). Mouth De-click, however, better-pre-
served bass spectra in the underlying track and introduced far less audible watery-sounding artifacts—a definite improvement.
The new De-bleed module attenuates or removes bleed in a track from an off-mic source. To use De-bleed, you have to import two tracks into the standalone RX 6 Advanced application: the track containing both desired audio and unwanted bleed (“Active Track”), and a track that contains only the audio from the source of the bleed (“Bleed Source Track”). (iZotope doesn’t recommend using De-bleed in Composite View to treat multiple tracks at once.) The two tracks must have matching sampling rates and be time-aligned within a few milliseconds of each other.
The De-bleed module is used on the Active Track. In De-bleed’s interface, you select the Bleed Source Track from a dropdown menu. After making a time-range selection in the Active Track where the bleed is most apparent, click on De-bleed’s Learn button to initiate the module’s analysis of the problem. After the analysis is complete, the center of the interface shows a spectral display of either the Bleed Source Track or Active Track, depending on which associated button you click (see Fig. 2). Drag the Reduction strength slider to the right to progressively remove more bleed, and ad- just the Artifact Smoothing slider as needed to avoid unnatural sound.
I used De-bleed on multimiked tracks for a male singer who was simultaneously playing acoustic guitar. De-bleed did an amazing job reducing guitar bleed into the singer’s mic. Any higher than a 0.3 Reduction strength setting (very low) thinned the singer’s deep bass register. But even such ostensibly light processing was remarkably effective at reducing bleed to an acceptable level.
Used on the same vocal track, the new Breath Control module was both extremely effective and transparent. The module has two modes of operation: Gain mode applies the same amount of gain reduction to all detected breaths, while Target mode lowers only loud breaths to your specified target level.
I also got fantastic results on the same track using the new De-ess module (also available as a plug-in), which lets you use either a spectral or more conventional mode of processing. Spectral mode uses multiband compression (using dozens of bands) and spectral shaping to achieve ultra-smooth and transparent results—without audible modulation—on complex programs. Love it!
Obviously, De-bleed, Breath Control and De-ess also have uses in post-production. And while there are other de-essers and breath attenuators available (including impressive plug-ins made by iZotope), many of RX 6 Ad- vanced’s other new modules offer veritable breakthroughs in audio repair, effectively eliminating noise that was heretofore impossible to remove. RX 6 Advanced is a stunning achievement and a must-have for anyone working in post-production.