Adaptive Noise Reduction in Post

[conleec]

Hello Everybody,

As some of you know I'm a picture editor who dabbles in audio post. Mostly shorts, trailers and a few micro-budget features. It's something I'm very interested in, but obviously don't consider myself a professional at. I'd call myself a skilled beginner, with the benefit of lots of watching and listening on others' dubstages (spent the day in stage 2 at Warner Brothers today on a CBS pilot).

Anyway, I was wondering how people use noise reduction tools, such as iZotope's RX2 or RX2 Advanced in their workflows?

Is it always used as a rendered effect, clip by clip, or does anybody use an adaptive noise reduction plugin in realtime? If so, under what circumstances, and do you have any tips for best results?

Just curious, mostly because iZotope is offering a reduced upgrade to to the "advanced" version and this feature is the one thing that I'd find interesting about it. But if nobody uses it that way, it doesn't offer much else I feel I need.

Thanks in advance.

Chris Conlee

[Craig F]

I think, clip by clip yields better results
It's a high latency plug if used in realtime

[Postman]

I upgraded to Advanced a couple of years ago (they had a nice special running), and I've tried several times to use the adaptive checkbox, which is on or off.

It hasn't worked for me well enough to keep the box checked. I have had amazing results with RX overall (as an audiosuite tool, not realtime), but at those times when I wanted adaptive to track a variable noise level, the results have been disappointing. When I use RX I use the standalone for everything except declip and decrackle. I have DNS One which is the most efficient for most of my problems. I use RX for tightly targeted fixes. Adaptive mode does not adapt quickly enough to track my problems. Car-by outside the window. Fail. Steam hiss in a kitchen. Fail. Airplane passover. Fail.

That's my experience. I'm glad to have Advanced so I can try every option when facing a horrible problem. So far though, the upgrade hasn't paid for itself.

Having said that, I'll also say that if my little upgrade fee has helped pay for more genius work from iZotope it is money well spent.

BTW Conleec, you seem to have set a big foot into the audio world. Welcome to the dark side!

edit:
I remembered that I LOVE standalone's ability to suggest a target frequency for dehum. I don't know if that's only an Advanced feature or if it is just in standalone, but it has saved A LOT of time!

[conleec]

Originally Posted by Postman

BTW Conleec, you seem to have set a big foot into the audio world. Welcome to the dark side!

Yeah, my wife would KILL me if she knew how much I've spent on what amounts to an expensive hobby at this point. I guess there's worse ways a guy could spend his money tho.

Chris

[minister]

Originally Posted by Postman

I upgraded to Advanced a couple of years ago (they had a nice special running), and I've tried several times to use the adaptive checkbox, which is on or off.

It hasn't worked for me well enough to keep the box checked. I have had amazing results with RX overall (as an audiosuite tool, not realtime), but at those times when I wanted adaptive to track a variable noise level, the results have been disappointing. When I use RX I use the standalone for everything except declip and decrackle. I have DNS One which is the most efficient for most of my problems. I use RX for tightly targeted fixes. Adaptive mode does not adapt quickly enough to track my problems. Car-by outside the window. Fail. Steam hiss in a kitchen. Fail. Airplane passover. Fail.

This was my experience with adaptive as well. Stopped trying. Cedar is what I use and I adapt it myself. Or go clip by clip and use RX Advanced AudioSuite. Sometimes both. They do different things.

Or, good dialogue editing and clever masking if that is called for.

I do like having advanced and the nasty stuff I can deal with in the standalone. I had some cassette recordings from the 70's that were bad to begin with that having the Advanced was a boon. So, for me, money well spent.

[antonis]

Originally Posted by minister

This was my experience with adaptive as well. Stopped trying. Cedar is what I use and I adapt it myself. Or go clip by clip and use RX Advanced AudioSuite. Sometimes both. They do different things.

Or, good dialogue editing and clever masking if that is called for.

I do like having advanced and the nasty stuff I can deal with in the standalone. I had some cassette recordings from the 70's that were bad to begin with that having the Advanced was a boon. So, for me, money well spent.

Hey Chris

What minister said is what the case is.

RX in adaptive or realtime from within PT does not do it for me at all. For the time-being only Cedar DNS One can pull it through real-time and it is really the best thing out there.

To be honest I never found the adaptive denoise option producing great results. But I currently always use The Adv+Extreme option in the MNS Algo menu under the More tab.

If you think it's any handy, my iZotope RX2 Advance process is:

1. I go region by region when dialogue editing.
2. I expand each problematic audio to its max handles to a work track.
(currently, I request 1min handles for two reasons: a. sometimes I am able to extract room tone from the handles when RT are not available from production, and b. sometimes, useful alt takes can be found in the handles.)
3. open the audio file in RX
4. make a note of the slate/ scene number, close slates might be afflicted by the same noise issues.
5. I make a preset of the noise print based/ named on the Scene/Slate No.
Presets in standalone will appear in the RTAS version.Hence if you do it once you might not have to go again to standalone RX and you might denoise from within PT.
6. I save the new file with RX as a suffix.
7. Import the RX'ed file in Pro Tools as a new track, based on the selection of the region in the work track (that will make the time stamping would be the same).
8. Select the unprocessed region.
9. Use the P or ; to go to select the same region "portion" from the rx file.
10. copy that and past it over the original, unprocessed file.

[lode]

Out of curiosity, Antonis:

Why not simply work with RX inside pro tools with audiosuite?

[antonis]

Because I find that there are a lot of times that I cannot find the best noiseprint possible through PT.

I like the visual aid of the Spectrograph to make quick decisions and accurately pinpoint what is what.

And if you ask me, I would think that would be a killer feature if PT could open up a region in a spectrograph window.

Soundtrack 3 has one built in (along with denoising and EQ cloning! But obviously the RX denoising algorithms are far superior).

[WOTE_Tech]

Originally Posted by antonis

Hey Chris
If you think it's any handy, my iZotope RX2 Advance process is:

1. I go region by region when dialogue editing.
2. I expand each problematic audio to i.................

Antonis, I follow your exact workflow with Izotope. I too find it much more intuitive and powerful standalone.

It would be great however, if standalone IZO could use the same playback engine as Pro Tools without having to quit the application. So for this reason, I am forced to monitor IZO out of my macs built-in line out...less than awesome DAC...

[FlorianE]

Originally Posted by WOTE_Tech

It would be great however, if standalone IZO could use the same playback engine as Pro Tools without having to quit the application.

There are several ways of doing that.
For instance, good old hardware style:
- connect Mac optical out to interface optical in
- OsX Audio setup: optical as standard O/P
- In PT I/O setup, set an input to interface optical SP/DIF in
- in the session, make a solo-safed aux track, passing this optical input to monitoring.

Alternatives would be:
- software bus like Soundflower
- analog wiring instead of optical wiring (if you dont mind the added DA/AD stage)

Same setup obviously works also with other standalone apps.

Cheers
Florian