OWC 12.4

Review: iZotope RX4 and RX4 Advanced

This working audio pro puts these plugins through their paces.

By Ron DiCesare

Just over a year ago, I had the opportunity to review iZotope RX 3 for postPerspective. So when I was offered the chance to test the latest version of the plug-ins, the RX 4 and RX 4 Advanced, I jumped at the chance.

Before you dig into this newest review, I encourage anyone who is not familiar with iZotope’s RX series to read my previous review advanced to get up to speed.

The new RX 4 and RX 4 Advanced offer all of the great plug-ins and features from the previous versions with some important additions. It is available for Mac or PC, including 64-bit AAX, which is compatible with Avid Pro Tools 11, RTAS/AudioSuite for Pro Tools 7.4 to 10, VST, VST 3 and Audio Unit. The list price is $349 for the RX 4 and $1,199 for RX 4 Advanced. It is also available as an upgrade for previous RX and RX Advanced users. In this latest version of RX 4, the standard version now offers one item that was only included in the Advanced bundle, the Dialogue Denoiser.

Introducing RX Connect
The iZotope RX 4 and RX 4 Advanced are all about improved workflow. That is due to a new feature called RX Connect, which has completely changed the way I work and how I think about plug-ins. RX Connect enables you to shuttle regions between iZotope’s RX 4 standalone program and your DAW (which for me is Pro Tools 10) with great ease. By following a few simple “send to” and “send back” steps, it works like a plug-in on steroids, allowing more flexibility and power than ever before.

In the past, I have only accessed plug-ins as inserts for realtime processing or via the AudioSuite, particularly for noise reduction and audio restoration. With RX Connect we can leave the AudioSuite behind and harness the power of iZotope’s standalone program. This changes everything and raises the bar on how we all work.

Understanding RX Connect is essential for maximizing iZotope’s power since not all of the RX 4 tools are accessible via the AudioSuite and inserts alone. In RX 4 Advanced when you select the AudioSuite or mixer inserts, you will see Declicker, Declipper, Decrackler, Denoiser, Dereverb, Dialogue Denoiser and Hum Removal. These offer more power and capabilities than anything I have ever used in the past. But, just like I never thought I would need a telephone that could take photos, launching RX Connect opens a whole new world of features and capabilities, and I am now wondering how I ever lived without them.

Using RX Connect is pretty simple when working with sound files and regions in your DAW. Select a region and launch RX Connect via the AudioSuite menu, which opens the standalone program. There you will see all of the items found in the AudioSuite and inserts, but more significantly you will see tools found only in the standalone program such as Time and Pitch, Equalizer and Spectral Repair. For this review, let’s concentrate on the four new ones found in RX 4 Advanced when using RX Connect, which are Leveler, EQ Match, Ambience Match and Loudness.

Ambience Match
To illustrate how innovative the folks at iZotope are, let’s start with their powerful tool designed to add back noise. That’s right… a tool designed to add back noise to a sound file rather than remove it. That took me by surprise, too, but here is how it can be used to solve an everyday real-world problem.

I do a fair amount of work for Vice Media, who are known for their intense news reports from some of the world’s most dangerous places. On the front line in Afghanistan, for example, survival takes priority over getting proper audio levels. As a result, I am often cleaning up dialog and trying to pull the words out of some very noisy situations. But another problem I am faced with is when the ambience cuts out between the lines of dialog. This can make the noise more obvious by drawing your attention to the on/off of the background noise as the dialog regions play out. To me, this is more distracting than a constant noise running in the background.

Ambience Match is a way to seamlessly blend the empty spaces between separate dialog regions with consistent ambience. This eliminates the annoying cut on-cut off from ambiences found on many dialog tracks. This is how it works — by selecting two or more regions in your timeline and launching Ambience Match via RX Connect, it samples, or learns, the ambience from the selected regions. Once learned, Ambience Match will render a continuous sound file with the gaps filled in with a consistent and natural sounding ambience. Incredibly, Ambience Match does this process automatically by regenerating ambience, not by copying and pasting it. From what I’ve seen, no other plug-in offers this feature.

That means we no longer need to resort to the patchwork of copying and pasting ambience and room tone segments “stolen” from what is often a very short and hard to find segment of clean ambience or room tone. Audio mixers are very familiar with having a dedicated audio track just for these stolen bits of ambience filling in random gaps in the dialog or production tracks. Thankfully, this is no longer an issue.

EQ Match
EQ Match is something that previously existed in iZotope’s Ozone 5 Mastering software, but now is part of RX 4 Advanced. Just like the name implies, it is a tool used to calculate and match differences in EQ. There are countless situations where EQ Match can be used with impressive results. I can see it being the “go-to” tool for matching the sound of ADR to production dialog.
EQ Match can take the guesswork out of mixing ADR and can be especially effective if the same type of mic from the set is used (a shotgun mic, for example) when recording ADR. Another great application is if the talent moves his or her position on the mic while recording. EQ Match can reduce the sonic differences resulting from such movement, even within a single take.

EQ Match works in two basic steps. First, select a sound file as reference file that you would like to match the sound to. Second, select the target sound file to apply the change to. Amazingly EQ Match analyzes the difference between the two files and calculates a new setting for the target file. Simply render the new setting to the target file and send it back to your session in your DAW.

I found another great application for EQ Match when I was working on a mix for a web series featuring multiple actors speaking on-camera. Different actors were speaking in different rooms and some sounded better than others. The resulting sonic shifts between people were very obvious to the listener. I discovered that different EQ Match calculation to the target file could be easily achieved by selecting different actors, or reference files. The various results done by EQ Match gave me different options to choose from. I found that selecting a reference file that I did not intend to use for the match achieved better results than the one I thought would work. Yes, this is more of a trial and error approach rather than a scientific approach, but it can yield impressive results with very little effort.

EQ Match is a great tool for anyone who does any advertising work where the VO is subject to a multitude of changes during a commercial’s approval process. It is possible for a VO to be recorded at different times at different studios using different microphones over the course of time. This leads to the common problem of combining new segments of the VO while retaining parts of the original VO. The goal, of course, is to seamlessly blend them together regardless of the quality or sound of each recording.

This problem has plagued me throughout my career. I can always come close to blending and matching different EQs and mics, but I am rarely completely satisfied with the results. I sometimes request that my clients re-record the entire VO to ensure a sonic match. Unfortunately, there are times when that is simply not an option.

To illustrate how EQ Match can save a recording, let me tell you how I used it to solve one of the most challenging problems I have been faced with. I was working on a TV commercial where it was the classic advertising conundrum — something known as “demo love” (I can hear my fellow commercial mixers cringing as I write this). In this case, a scratch track (or guide VO) was recorded at the advertising agency as a quick rough track intended only for the video’s timings. As a result, the talent was too far from the mic and a non-audio person was at the controls doing the recording. This is perfectly acceptable for audio used only by the production team, but it can lead to the common problem of the client hearing this guide track and loving it. What I see happen time and time again is the final VO performance is judged and approved solely on how close it sounds to the temp track instead of judging it on its own merit.

In this case, the original voice recording was done in a medium sized room with a CAD M179 mic using a Mackie mixer as a mic-pre with the talent about 18 inches away from the mic. The new VO was recorded in a small iso-booth using a Neumann U-87 with an Avalon 737 mic-pre with the talent about six inches away from the mic. Big difference. EQ Match allowed me to be able to switch between the scratch track and the new recording with little variation.

After my first attempt, I quickly realized that there was more to my problem than just differences in EQ. The scratch track had a loud air conditioning noise and a lot of room tone recorded in it. To achieve the best results, I needed to use a combination of RX 4 tools starting with Declip, along with some gain reduction, to give me the headroom needed for any additional processing needed.

Next was Spectral Denoise. In this particular sound file, there wasn’t much noise in the clear included in the dialog due to the video editor cutting around the words as tightly as possible. RX 4 allows for multiple selections to be used when learning a noise profile. This is great when there isn’t a long section of noise in the clear. This is accomplished by holding down the shift key when using the time selection tool allowing me to select multiple small segments. The larger the selection to learn the noise profile from, the more accurate the noise reduction will be.
One really cool option was to select the unlink button below the reduction slider in order to have separate control over the Noise Reduction and the Tonal Reduction. This is an important distinction RX 4 makes. This gave me independent control of the two and allowed me to push the tonal noise reduction harder than the noise reduction helping me eliminate any unwanted artifacts.

The next steps were to use Declick to remove some of the mouth noise and Dereverb to help cut down on the room tone. At this point most of the noise, room tone and other flaws were reduced to a minimum on my target file. Now EQ Match was the last step to really help make the scratch track match my final VO recording. I used the final VO region as my reference for EQ Match to learn and calculate the EQ match to the original recording. Now that I had a cleaner target file, EQ Match was able to work its magic much more effectively than when I first used it on its own.

The next new feature offered by RX Connect is Leveler. This is a great tool for mixing dialog. Simply select a region or file containing dialog, launch RX Connect and Leveler automates the volume graph in the standalone program to smooth out the peaks and valleys. This quickly achieves constant levels with an even sound. For example, I was working on a short film where I had the problem of two people talking on-camera, but only one lav was pinned to the lead actor. This resulted in a clear level difference between the two actors talking. By selecting Leveler via RX Connect, I was able to automatically bring up the volume of the person speaking off-mic to an acceptable level.

I can see two important applications here. First is for any video editor who may not have the time or budget to do a proper audio mix with an audio person. Leveler is a great tool for “one-step” dialog mixing, rather than resorting to something like an extreme compression or limiting setting overall. I have seen this done as a quick fix, but it alters the sound too much for my taste. Using Leveler a great way for someone with little or no audio mixing experience to get dialog to be more even and natural sounding.

Another possible application for Leveler is for any audio mixer working under tight deadlines. Like most mixers, I focus much of my time and effort on the dialog levels because often the dialog is what the entire mix is judged on. But many audio mixers may not want to leave this crucial step entirely to an automated process, so RX Connect offers something very cool. After launching Leveler, you can select View Waveform Data and see the moves as a volume graph like the one used in Pro Tools. This allows for any kind of modifications to the auto-created volume graph in the standalone program which gives the audio mixer full control over the volume moves before rendering out the file.

I like to think of Leveler as a first pass of the dialog mix by an assistant before I go in and do the final moves. But like with any rendered file, be mindful that rendering the file will commit the moves to the sound file. Therefore you may need to do additional mixing once it is sent back to your DAW to achieve the best results.

The last of the new features I am focusing on is a tool called Loudness. This one is something that I have been waiting for. The loudness feature is able to take a sound file, particularly a final mix, and process the overall output level to be at a certain spec you assign. For example, if you bounce out a sound file that needs to be -24 LKFS, but is either too high or too low, you can set Loudness to process the file to be exactly -24 LKFS in one step.

At Vice Media, most everything I do is for the web. Most people know there is no real standard for audio levels on the web like there is with movies, TV shows and commercials. Vice is addressing this issue by creating its own in-house audio standard for all mixes for their website to adhere to. They measure the final audio output level by using both RMS and maximum peak values found on the AudioSuite Gain plug-in standard on all Pro Tools systems.

Vice is taking the lead on this issue, but the AudioSuite Gain tool may not be the best choice. However it is something that all Pro Tools systems have in common, which gives them some kind of standard metering system for delivering final mixes. The limitation is that the Gain plug-in only does the analysis, not any processing.

Now with the Loudness feature in RX Connect, you can simply set the target level, render it out and you are done. It’s that simple. One day I can see the web having a standard spec like the one we use at Vice. Whether it’s TV, the web or spots, the Loudness tool will help many people pull mixes into spec. It is especially helpful for people who do not have access to proper audio metering (such as video editors) to be able to adhere to audio level requirements in this nearly foolproof way.

Tools of the Trade
You can expect to see all of the great tools from previous RX versions but some have been improved, such as Declick. But by far my favorite is Insight, which is included in RX 4 Advanced. Insight is one of the most important metering tools I have ever used. I love how iZotope was brilliant enough to include a metering tool as part of a noise reduction package. Because of this, RX 4 Advanced is the only plug-in bundle installed in my system other than the stock Pro Tools plug-ins. It really is the one and only plug-in set I need.

There is a helpful user’s guide available though iZotope’s website that offers many great tips and techniques on how to fully use the program. The guide offers incredible insights, and I would strongly suggest reading it cover to cover. One important thing I learned is that an effective way to use any of the RX 4 plug-ins may be using smaller amounts of processing multiple times rather than one pass of a higher or more extreme setting. Smaller amounts in succession, particularly with Dereverb, are likely to produce better and more natural sounding results. There are many more things to be learned from the guide, far too many to list. And if anyone gets stuck on something they feel has not achieved the best possible results or even not worked at all, I suggest contacting iZotope’s support via email to have them take you through anything that may be giving you trouble.

Audio pro and reviewer Ron DiCesare

Audio pro and reviewer Ron DiCesare

Summing Up
The bottom line is that iZotope’s RX 4 and RX 4 Advanced have gone well beyond the capabilities of the AudioSuite and realtime plug-ins with the introduction of RX Connect. Thanks to RX Connect, my workflow has shifted away from many small repeated steps in Pro Tools to more broad fixes in iZotope’s standalone program, which are easily incorporated back into my DAW.

I urge every mixer who has needed to perform “miracles” on dialog (or any audio for that matter) to explore the capabilities available through RX Connect.

Ron DiCesare is an independent recording engineer working in New York City. He has worked on TV and radio commercials, including Beggin’ Strips, Tidy Cats and MasterCard. He has mixed films, including Con Artist and B.A.M. 150and he has worked with such artists as Lisa Loeb, Aaron Neville, the B-52s and James Taylor. Ron is a member of the Recording Academy and votes in the Grammy Awards each year.



One thought on “Review: iZotope RX4 and RX4 Advanced

  1. Paul

    Well thought out and concise review by Ron.

    Regarding Vice, Ron states “… Most people know there is no real standard for audio levels on the web like there is with movies, TV shows and commercials. Vice is addressing this issue by creating its own in-house audio standard for all mixes for their website to adhere to. …”

    I think the first thing Vice needs to do is to stop Peak Normalizing to Full Scale (0 dBFS). I analyzed one of their most recently distributed videos and it exhibits various instances of Intersample Peaks (above 0 dBFS).

    Besides Peak Ceiling issues, -16.0 LUFS (Stereo), -19.0 LUFS (Mono) Integrated Loudness have been widely documented as suitable targets for internet/mobile distribution. That would include spoken word audio (podcasts), and various types of video programming.

    In any case even if Vice establishes their own in-house set of standards, there’s no excuse for distributing clipped audio. The videos should be Loudness Normalized to what ever Integrated Loudness targets they adopt, and the True Peak Ceiling is an important aspect of the adopted spec., especially when distributing lossy codec audio.

    My recommendation (besides -16/-19 Integrated) … True Peaks no higher than -1.5 dBTP in the lossy audio copy.



Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.