The Sound Board

I've just been trying it in anger on a short film. It is quite disappointing really. All this stuff is material dependent obviously. Waves Clarity is better on some stuff. It's another tool in the box, but after Hush Pro I was praying for miracles.

Ah, that's too bad. I'm rooting for RX since it's my tool of choice, though I don't really do much dialogue editing (odd bits here and there). But for everything else, I do heavily rely on it. Thanks for the feedback, Guy. I think at this point I'll wait to upgrade.

I got the Standard version yesterday via Plugin Discounts for £70.

Again, it's not a huge update from 9 and I skipped 10 completely. There's a few features missing from Advanced, but I have all those in RX9. There are only two enhancements I don't have, which is the improved Dialogue Contour and RX10's updated Spectral Recovery with a low end mode. Those would have cost me an extra £200.

Here's all the new stuff I have with this upgrade.

NEW DIALOGUE ISOLATE
This is a better than RX9, but nowhere near on a par with Hush Pro. I like having noise and de-reverb on their own separate faders, this is probably the single biggest benefit.

MUSIC REBALANCE
Honestly I can't tell much difference here though they say they've improved the algorithm. On some material it is still miraculous, but it's typical to hear those artefects if you are totally separating stems out. The buffer time on BEST has gone up considerably, it takes something like 8s to hear a change you've made when previewing

REPAIR ASSISTANT
Still haven't actually used this.

LOUDNESS OPTIMIZE
This is essentially is a one-trick pony - if your quiet stuff falls below a certain threshold, it doesn't count in the loudness reading. By boosting the quiet stuff a little, it then gets calculated and drops the overall reading - meaning you can make the overall track louder. It seems to do this elegantly. It's a one-trick pony, but a fairly cool trick. Many tracks aren't affected at all, however.

STREAMING PREVIEW
This feels like half a feature. You can compare codecs between the different streaming platforms, which is good I guess. But what would make this far more useful is if it also adjusted the loudness figures for each platform, giving you an instant A/B/C between each one. You have to generate previews for each manually.

ARA INTEGRATION
At the moment this is only available for Logic in Rosetta mode, so pretty useless. It is coming for Pro Tools however (and Studio One) so that might be useful.

MID/SIDE MODE
This exists, I haven't had cause to use it yet.

NEW IN RX10

TEXT NAVIGATION
This is utterly useless - word recognition is appalling. If it ever got accurate, and it could find uses of repeated words or phrases etc, it might be useful one day. The fact it hasn't been improved in 11 when it is so awful makes me not hold my breath.

DYNAMIC ADAPTIVE DE-HUM
I tried this on an old much-used sound effect that has hum all over it. On the full frequency it did change the sound of the thing itself so you have to be careful, but restricted to the right range it worked great.

MULTIPLE SPEAKER DETECTION
Like Text Navigation, this just doesn't work on my test material - it couldn't tell the sonic difference between a man and a woman. Would be very useful if it worked and you could select all of one speaker.

So a very mixed bag. They are still clearly falling behind the competition when it comes to innovation, but as a complete bag of tricks it is still formidable. The pricing on Advanced to Advanced is insane, but Advanced to Standard is okay.

RX11.1 is out, with ARA compatibility with Pro Tools and Studio One. Note - that Pro Tools will only work with the next update, 2024.6, due out this month.

A few more observations on music rebalance...

While the swishy artefacts are still largely present and incorrect, it's fascinating to see what the algorithm can and can't do. Arcade Fire's Black Mirror from Neon Bible is full of vocal effects - long reverse reverbs, short echos and delays all over the place. RX11 was phenomenal here - it put them all into the vocal stem which almost defies belief. On the other hand, Kraftwerk's The Robots confused the hell out of it. Bits of the classic bassline got separated into Bass, Drums and Other, the snare went into Other too at times. Vocoder vocals go into other. The sensitivity controls do vary things, but not very usefully. This is a very hard test to be fair, Kraftwerk's pure electronic source generation is tough.

RX11 is a step forward quality-wise, but I don't think it's the best in class. Lalal.ai gets consistently rave reviews from colleagues, so I tried The Robots on that. It offers a staggering selection of possible stems, but it was as confused as RX as to what goes where. I then gave it some easier fare, Daft Punk's Get Lucky, to see how clean the drums were. And the answer is - considerably cleaner. It's certainly not perfect - yet - but it is impressive.

RX11 - Get Lucky drums
https://www.dropbox.com/scl/fi/xzt3ofid ... yb47g&dl=0

Lalal.ai - Get Lucky drums
https://www.dropbox.com/scl/fi/xzt3ofid ... yb47g&dl=0

Still. It's all miraculous, but how fast we become blase to it.

Here's a little factoid. I've been running a Pro Tools session with RX11's Dialogue Isolate in live mode to reduce the reverb in a recording. It works tolerably well if not overdone. However I noticed the CPU meters and latency was pretty high so I glanced at the power meter I have across the rig. Making that single plugin inactive made the CPU drop to near-zero - and 30w got taken off the power consumption figure.

So if you want save a bit of money and the planet into the bargain - render the results of very power-hungry plugins.