iZotope RX8 September?

[TrentWilliams]

Hey all,

Just wanted to get your thoughts on RX8, which I assume should be just around the corner.

Where to from here? What are you expecting? Having moved some of my workflow to Brusfri and Absentia DX, I’m wondering how can RX improve?

[Benny_Cha_Cha]

I am definitely looking forward to v 2 of the Music Rebalance module. Improved algorithm and preview would be nice too.


Sent from my iPad using Tapatalk

[dr sound]

Where did you hear of RX8 releasing in September?
I couldn't mix Dialog in Features or TV without RX
and Absentia DX.
Can't remember the last time I reached for Brusfri
in a mix (and I bought it the first week it came out).

[noiseboyuk]

September is a pretty fair bet - has been so for the last few RXs.

Here's my wish list:

DeSpill. Would remove ambient component present in a key input from the signal. This would be revolutionary, and I can't believe the tech doesn't exist right now. Essentially you could feed mics into each other to get much cleaner versions of each. De Bleed does this a bit, but its very fiddly and doesn't work very well imo.

ReverbMatch. Capture a reverb profile from a reference file, and apply this to another signal.

DePlane. Jets and prop planes. How great would that be?

DialogueContour (improved). Currently doesn't work very well, but i'd like it to. Could borrow from RevoicePro and DialogueMatch - use pitch info from a reference file and apply to a different file.

SplitLayers. (I'm really dreaming now). An extension of Dialogue Isolate. Think a clever Photoshop. Split any signal into a number of layers based on the machine learning engine. A particular voice vs a cough in the background. A male and female voice. A voice vs a buzzsaw going off in the background. A voice vs a wailing child. So it would analyse a signal, detect (say) 3 component parts and split into 3 layers. Put together, they equal the original file, but you'd get faders for each, or just isolate one layer for particular treatment.

I could go on... basically I can forsee Noise Reduction and sweetening as a virtually bottomless pit.

[BScout]

That last one is pretty much Sony (now Steinberg’s) SpectraLayers.

[filmixer]

@dr sound
"I couldn't mix Dialog in Features or TV without RX
and Absentia DX." Yeh you could. ;-)

[noiseboyuk]

Originally Posted by BScout

That last one is pretty much Sony (now Steinberg’s) SpectraLayers.

Wha wha wha whaaaa?!!! How come I've never even heard of this?!! I've been using Cubase for a decade (and now on the Beta team), PT and RX and I swear I didn't even know it existed.

Currently on holiday so not easily able to trial for a few days. Ive just looked so far at the blurb on the website and the promo video - does it automatically detect the different instruments (or whatever)? If so and it does it reliably I am totally sold. There seems to be a half price deal on at the moment too... anyone know how long it lasts for? Thanks for any infos.

[noiseboyuk]

Oh... on SpectraLayers, been looking at the manual and I can't see any of the magic I was hoping for or described in my wish list. Layers in SpectraLayers seems to work by importing different files, then you can work on them all together in layers. It's missing the critical ability to intelligently split a single signal into multiple layers. THAT would be magic.

There do look like a couple of good features though, it can semi-intelligently select all the harmonics of something. I don't know how clever that would be on speech and other sounds, but something along these lines in RX8 would be great - the selection options in RX are actually fairly crude at the moment.

[BScout]

Originally Posted by noiseboyuk

Oh... on SpectraLayers, been looking at the manual and I can't see any of the magic I was hoping for or described in my wish list. Layers in SpectraLayers seems to work by importing different files, then you can work on them all together in layers. It's missing the critical ability to intelligently split a single signal into multiple layers. THAT would be magic.

There do look like a couple of good features though, it can semi-intelligently select all the harmonics of something. I don't know how clever that would be on speech and other sounds, but something along these lines in RX8 would be great - the selection options in RX are actually fairly crude at the moment.

I haven't tried your specific application (I'm just experimenting with it now as a long time spectral junky from CEDAR Cambridge and RX). You might be able to do better splits with its Melodyne integration -- once they are to separate "tracks", SpectaLayers can layer them it seems. It's worth a spin -- their whole idea from the start at Sony was Photoshop for audio (so it takes a bit to work away from the workflow of RX though some things work the same.)
I have no idea how long the recent trial period is. It seems to be Steinberg's push after just purchasing them.

[simon.a.billington]

Originally Posted by noiseboyuk

September is a pretty fair bet - has been so for the last few RXs.

Here's my wish list:

DeSpill. Would remove ambient component present in a key input from the signal. This would be revolutionary, and I can't believe the tech doesn't exist right now. Essentially you could feed mics into each other to get much cleaner versions of each. De Bleed does this a bit, but its very fiddly and doesn't work very well imo.

ReverbMatch. Capture a reverb profile from a reference file, and apply this to another signal.

DePlane. Jets and prop planes. How great would that be?

DialogueContour (improved). Currently doesn't work very well, but i'd like it to. Could borrow from RevoicePro and DialogueMatch - use pitch info from a reference file and apply to a different file.

SplitLayers. (I'm really dreaming now). An extension of Dialogue Isolate. Think a clever Photoshop. Split any signal into a number of layers based on the machine learning engine. A particular voice vs a cough in the background. A male and female voice. A voice vs a buzzsaw going off in the background. A voice vs a wailing child. So it would analyse a signal, detect (say) 3 component parts and split into 3 layers. Put together, they equal the original file, but you'd get faders for each, or just isolate one layer for particular treatment.

I could go on... basically I can forsee Noise Reduction and sweetening as a virtually bottomless pit.

DeVehicles in general. I think this is a region in which machine learning could benefit a lot.

I could see your SplitLayers working a lot like the Music Rebalance thing.

The BIGGEST thing I want, though, is to bring the upgrade pricing down. Even with the vouchers it’s still one of the most expensive software I own.