OT: NR2

JimMSG wrote on 7/30/2010, 2:00 PM

Has anyone done a study on what settings will work the best for specific types of situations. I have played around with it a little bit, although I usually end up using "Default for fast computers" as it still seems to do a better job than what I come up with. Would like to learn to customize it so it works even better, and would like to see what others have done in this area, if such a discussion exists.

Also, anyone heard any word on if/when this will be ported over to 64bit? I'm using it on the 32bit version of Vegas running on a 66bit Vista machine. I have a 64bit W7 upgrade, I might actually get around to installing over the weekend. I've kind of caught up with things, and really shouldn't have to worry if some my Vegas stuff stops working with the move from Vista 64 bit home premium, to W7 64bit home premium.

Comments

Spot|DSE wrote on 7/30/2010, 3:00 PM

Default works best in *most* instances.
Wish I had word on any 64 bit port, but honestly...I'd be surprised if it was.

Use small noise slices for best results, and sample multiple points/times vs trying to grab big slices.

JimMSG wrote on 7/30/2010, 3:21 PM

Rats! I was hoping there was a treatise somewhere explaining how to get it to work even better than it already does. I've got a "mountain" video where the camera was way too far from the event being shot, not to mention the speaker was voice only, no amplification. Oh well, so it goes . . . .

Too bad if it never gets ported to 64bit. I suppose it really shouldn't bug me, but it does to run a 32bit version of the program on a 64bit machine. Have the same issue with Wave Hammer. Almost never need the surround, would really like the plain version of it to work in 64bits.

Your comment about small noise slices interested me. That's what I do, but how do you sample multiple points/times? NR2 only allows one sample for any noise contour, or am I missing something? Or, am I understanding not what you meant?

Spot|DSE wrote on 7/30/2010, 3:28 PM

Multiple passes at NR vs trying to do it in one pass. You can also apply multiple instances.

TheHappyFriar wrote on 7/30/2010, 3:40 PM

I normally put the audio that needs to be cleaned up on it's own track & apply NR on the track level. That way I can layer NR (and other FX) as needed.

John_Cline wrote on 7/30/2010, 4:01 PM

Always use MODE 3 and a small FFT size, like 1024 or 2048, will generally minimize the artifacts.

JimMSG wrote on 7/30/2010, 6:19 PM

I do a lot with multiple instances, usually two. Three tends to start aliasing. Otoh, sometimes the client prefers the aliasing with less noise. It bugs me though. I have given some thought to running it on a track, rendering the track, putting it in the place of the original and starting again. I have always suspected x-number of runs or instances would have the same result. Perhaps that isn't the case. What have you found?

farss wrote on 7/30/2010, 7:22 PM

I've never even attempted to use NR2 in Vegas.
Anything that needs it gets bounced to Sound Forge where it's so much easier to use.
One thing I do is firstly to use the spectrum analyser and my ears to try to work out roughly where the noise I want to be rid of lies. Then after taking the noise sample I may modify the noise footprint before doing the next phase of actually filtering the sound.
Another tip regarding the sample size. For HF noise you only need a small sample, for LF noise you need a larger / longer sample.
Also don't overlook using simple Eq to get rid of noise. The other thing is I suspect some people expect miracles and with audio it's not too hard to make good into great. It's generally impossible to make bad into good though. One of the problems I've found with annoying sounds is even though with NR2 I may be able to reduce them 20dB or more which is really quite a dramatic improvement they're still friggin annoying. Sometimes it's possible to mask them with some new sound element that's not as annoying.

Bob.