iZotope by Native Instruments has just released RX 11, the latest version of their audio repair toolkit. RX can tackle unwanted noises such as clicks and hums, it can create spectral based visual guides to pinpoint problem areas inside of an audio file, and its toolset can solve a whole host of audio issues. The past few years have brought a shift in audio processing utilizing machine learning and neural networks. iZotope has reworked some of the modules of their industry workhorse in this latest iteration and in some cases rebuilt the code from the ground up. RX is a tool used for audio post, podcasting, and music, however, this article will focus on some of the changes for audio restoration rather than the music tools.
Probably the biggest advances in audio restoration has been the number of companies who are developing tools with machine learning to clean up audio. The RX kit has many tools that address audio repair from the different required angles and one of the mainstays has been their Dialogue Isolate module. Clearly by its name, its use is to isolate dialogue from noisy recordings. In RX 11, using iZotope’s machine learning, Dialogue Isolate is now an insert plugin which can process audio in real time. If working in Avid Pro Tools for instance, before version 11 you had to either AudioSuite render process the file, or round trip to the RX Audio Editor for processing which you would then send back into the timeline.
Besides the addition of real time processing there are other updates to this module as well. Prior versions had sliders for adjusting Dialogue Gain and Noise Gain, as well as Sensitivity and Ambiance Preservation to control the unwanted noise. RX 11 Dialogue Isolate now offers sliders for Voice gain, Reverb gain, and Noise gain as well as the Sensitivity slider.
With the availability of all of the noise reduction tools available today there can be an impulse to over process sounds and lose the distinct characteristic of the recordings. Hey, we live in a noisy world! Don’t overdo it. The basic rule of thumb with any noise reduction is “do no harm”. Finding the right balance of noise removal to signal is key. The addition of dereverb and more defined controls for each problem area is a welcome upgrade for Dialog Isolate. The Advanced version includes a high-quality offline rendering algorithm that is available in the RX audio editor.
In RX 11 Advanced there are additional controls to define the frequency bands where the noise is located to process. This allows for reduction fine tuning so that if, for instance, noise is happening in the upper frequencies but not the lower ones, you can adjust the bands to only affect those frequency ranges that are problematic.
Another major update to RX 11 is the Repair Assistant module. This is included in the Elements, Standard, and Advanced versions of RX 11. Using their advanced machine learning, the new module has changed a bit and now also includes dereverb as a part of the possible processes available.
Repair Assistant is a fantastic tool for allowing RX to quickly analyze and hone in on the possible problems of your audio and make recommendations on a course of action to create the best sounding result for you. RX 11 Repair Assistant has a cleaner interface than prior versions and iZotope says that under the hood they have further enhanced the audio analysis. You have flexibility with the suggestions that RX makes, the Module Chain gives the user fine tuning of each of the suggested processes. In this chain you can adjust the amount of processing suggested or dismiss any particular process altogether. Repair Assistant has always been a great way to quickly assess issues in the audio and then use that analysis to amend, tweak, or discard the results to taste.
iZotope has also improved the Dialogue Contour module which can pitch and alter the expression of individual words, or parts of words, with a graphical editor, to help change the spoken inflection. Due to the nature of edited dialogue, sometimes we are forced to edit content in a way that sounds unnatural, with mid-sentence phrases that don’t sound complete. Dialogue Contour now has an improved interface to show how you are affecting the needed process with more precision. It has also added Pitch and Variation to the Formant to fine tune the process. As with all of these processes, just a little goes a long way, but this can be useful for those of us dealing in a world of “Frankenbites.” The finer tuning adjustments are a welcome advance to this module.
There are a number of things that have also been added or refined on the music side of RX 11’s capabilities. Music Rebalance, a module that can separate parts of a musical piece into its elements, has been updated. You can pull out vocals or percussion for instance, or as the name implies, boost the vocal in the track, or create a set of stems for the various parts of the cue.
RX 11 has a fantastic Streaming Preview that allows you to listen to your tracks as if it was processed by a streaming service such as Spotify to hear how it would sound. This offers powerful insights into how the various streaming algorithms affect the mix. Something like this would be greatly welcomed in the post audio community. For instance, it would be helpful to be able to analyze a mix and hear how the various streamers, such as Netflix, might alter the final full mixes of a program sent through their algorithms. Mike Rozett of iZotope says that although such a tool is not available for this launch of RX it is something in the plans for later release.
Along with Streaming Preview there is a companion Loudness Optimize module that analyzes how the mix will be measured by the algorithm. By using machine learning it can analyze how the gating affects the signal and then optimize the level of the tracks so that it is not adversely affected once uploaded by a streamer. This is a fantastic tool that will give insight, help make your mixes sound better, and preempt possibly being negatively processed by the streamers.
Here is a video from iZotope that helps explain this process.
https://www.youtube.com/watch?v=SZk5Xn1nDuY
I’ve been a user of RX from the beginning. No matter the project I’ve worked on, be it feature film, feature documentary, or television programming, there is always something to be addressed from the location audio. The wide range of tools in RX has been extremely useful to tackle these problems. On a typical audio post project, I can use just about every module offered in one way or another. I used to rent hardware devices that I sent audio through SP/DIF to process and then to round trip back into Pro Tools to do the noise reduction. They were very expensive, so I only rented, never bought a unit! Then over a long period of years there have been various manufacturers and programs that addressed aspects of the types of issues that are typical from production audio. RX was a comprehensive kit that I embraced from the start.
The immediate advantages of RX was its lower cost and its many modules that could tackle different issues. It’s grown over the years to include more and more modules and as it has done so, iZotope has included more of the Advanced modules down into the Standard editions. The value it provides for the cost is extremely high in my view.
Today there are many new options that offer similar noise reducing features with different approaches to the end result. Some like to use the “as few knobs as possible” approach, some offer varying levels of options to use machine learning to add back lost or missing characteristics to the offending audio. RX has to compete now in what’s become a crowded field of competitors. Typically, and not surprisingly, each sound editor or mixer has their preferred piece of software or workflow for repair and restoration. Ask any sound pro what their favorite EQ or reverb is, what order they process audio in, or for that matter, which monitor speakers they prefer! As is often noted, there is no right or wrong in post audio as long as we achieve our desired result, can work efficiently, and the client walks away happy with a great sounding mix.
As a supervising sound editor and rerecording mixer, I use and own a breadth of these new tools in my own work. I find that some are better than others on specific types of problem material, and I might run a piece of audio through a couple different options to determine which one suits the moment and media best. However, RX has a suite of modules and tools that I reach for most often because it offers so many different processes in one piece of software. If you are a current owner of a prior version of RX, of course whether to upgrade or not is a decision of your own. But if you are not a user, I would suggest the trial demo to see the usefulness of the wide-ranging tools that are available in real-time, offline, and in the audio editor. RX is a very strong tool kit in a world filled with recordings made on smart phones, Zoom, and all the other compromised audio tracks we are tasked with fixing while creating compelling and creative soundtracks.
Woody Woodhall, CAS is a Supervising Sound Editor and Rerecording Mixer and Founder of Los Angeles Post Production Group. You can follow him on Bluesky at wwsm.bsky.social