Good on you Grazie - yes, I have RX2 - haven't used it for a while, so a bit rusty!

Peter

Grazie I'm having some success with Lassoo / Spectral repair - increase strength to 2.6.

This was the first thing I tried .... I doubt I got it totally right first time!

Yeah . . I was trying with Magic Wand, but that was far to heavy handed, and I wan6t to learn how to modify the "bit" it, Magic Wand, takes from the Waveform.

The trouble with all of this is, is that much of your "wanted" audio is mixed in with the "unwanted" - it's gonna be a close call.

Grazie

"The trouble with all of this is, is that much of your "wanted" audio is mixed in with the "unwanted" - it's gonna be a close call."

Indeed. Wind noise tends to be longish in duration which might make it hard for the sepctral repair to fill in what it taks out. In a previous thread on the wind noise problem I thought the best advice was to err on the side of caution and not try to take too much of it out.
On a windy day, if there's visual clues that it's windy, wind noise isn't really that repugnant. Artifacts always are in my opinion.


I've been on and off grappling with a problem caused by screeching birds that come and go between shots and this thread just got the brain functioning. I just realised it might be easier to add bird noises where there isn't any than remove the ones in the other clips.Thanks gents and thanks to the sound guy from "Night Watch".

Bob.

. .and this from Izo Manual:

"ANR-B is not designed to remove intermittent noises such as horn blasts, explosions, ringing cell phones, microphone wind noise, etc. We recommend using a suitable post-production tool such as iZotope RX Advanced software (www.izotope.com/rx) to remove noises of this nature."

There you have it. Buy the Advanced s/w . . .

Grazie

Yeah - I read somewhere that the magic wand was able to identify harmonics of unwanted sounds, which can be very useful, but I'm not sure it applies to windnoise, which is very bottom end..

I'm finding that lassooing a little at a time, then Spectral repair is working quite well, but of course this way is very time consuming.

Bob, I like your lateral thinking - putting more birds in!

. . "but I'm not sure it applies to windnoise, which is very bottom end.."

Has been in my experience! . . Lol!!!!

Grazie

iZotope will not be able to do anything useful with wind noise. I am not aware of any post-production tool that can do much. You might get a little relief by taking a noise print, and then using the advanced features in RX2 to roll off the noise curve after about 150 Hz so that the noise reduction only gets done at the low frequencies. Most of the noise energy from wind is actually down at low frequencies, even though you may be more conscious of the "hissy" higher frequencies.

You might also be able to use the noise gate in SoundForge to reduce the wind noise in between sounds that you want to retain.

Although several people on this forum have mentioned they do not like the product because they do not like their web site, the Enchanced Audio software "Diamond Cut 8" does an outstanding job on wind noise. I recently had a shoot interviewing a nobel prize winner and the client insisted it be shot in a small courtyard between 3 buildings. Each had a LOUD air conditioner on the roof and there was a 30 knot wind that the foam windscreen with a dead cat over it that would not stop the noise on my AT nor my AKG shotguns. I ran it through the DC 8 continuous noise filter and voila - all the extraneous noise was gone. It's magic. Then I was left with people talking in the background that I had never heard because of the AC and wind noise. I was able to get rid of most all the the talking manually with the DC8 interpolate filter. And, I'm not even using the Forensic version!

Peter -If you want to send me a clip thru the forum email, I'll be glad to try DC8 on it and see if it will work on your problem.

Kent

I'd love to hear the before & after on this using Diamond Cut 8. I am not familiar with that product, but if it can do anything for wind noise, I'll buy it right away. I thought this was one of those "can't be done" things, like taking an out of focus picture and making it look sharp.

I'd like to hear that, too. Despite the fact that wind noise has a significant amount of low frequency energy, it is really quite wide-band as well as constantly varying in level. Most all FFT-based noise reduction is for constant noise at a fixed level. If you recorded the audio with a camera that has automatic gain control, it gets even more complicated; the low frequency energy will modulate the audio levels to the extent that even if there was some miracle software that removes all traces of wind noise, you will still end up with the volume of the remaining audio varying all over the place. I will be (pleasantly) shocked if Diamond Cut 8 can fix the problem.

John,

Your description exactly mirrors my understanding of the problem and is why I posted earlier that I didn't think it was possible to do much, if anything, to reduce wind noise in post.

But, having said that, many of the things that iZotope RX2 does are things I would have absolutely sworn were impossible to do, if you asked me about it a few years ago.

John,

I think Izoptope RX is the closest thing I have ever seen to "miracle software", however, even it doesn't fix wind noise. I won't flatly say that it can't be done, I just haven't heard it done yet with any software.

"You might also be able to use the noise gate in SoundForge to reduce the wind noise in between sounds that you want to retain."

There seems to be some exceptionally better gates than the ones in SF around. They seem to only open based on some form of filtered level detection and as they open they open a bandpass filter. The reverse happens as they close. This seems to eliminate the quite obviouse effect of the gate, you really have to listen to hear that there's still as much noise mixed in with each word to work out how it is being done. Some very problematic location sound that I'd given up on was made acceptable by the use of this.
Unfortunately I couldn't find out any info regarding the name of the tool used to do this.

Bob.

No-one seems to believe that Izotope can do much with wind noise, so I have made a 15 sec sample of a Before and After, and rendered it to an mp3.

This is far from high quality - it wasn't a paying job, and in fact was recorded on a $350 Zoom Q3HD which wasn't monitored - I was playing keyboards at the time - but you can certainly hear the difference. I used the RX2 lasso to select the "orange mountains" at the bottom of the waveform, then used Spectral repair, with strength increased to 2.6

For some reason I can't upload to my server via FTP, so instead of making it universally available, I shall send it to Grazie, farss and johnmeyer, whose emails I have.

Peter

Difficult call.
That "woosh" left in the word "drumming" is a problem.
Listening to both without the matching vison makes it hard to judge.
Certainly getting rid of the low frequencies which are being accenuated by the sub on my office PC is really helping so I'm wondering if just shelfing that out would be better than going further which is leaving some sounds with artifacts which could be rather distracting.


One thing that Rx does better than anything else I've tried is high Q filtering. I've tested asking Rx to cut white noise at 100Hz and looking at the result with SF's spectrum analyser, that's what it does, it's just a cliff. None of the filters in SF come close to that.

Bob.

Thanks Bob
Firstly, as I said, this is far from perfect. For a start I don't think I have a full grasp of how to use the features of RX2, and even though you say it's a difficult call, I don't think it's hard to say which is better to listen to.

I'll have a go at shelfing tomorrow ...

Another complicating factor is that I unwittingly had the audio level of the Q3HD on "Auto", so immediately after the loudest wind noise, which happened over the word "drumming" that you mentioned, the camera dropped the audio level drastically, so I isolated the following portion by splitting and increased its volume by first Normalizing, which made it too loud, then pulling down the Event gain and dragging to make a short dissolve to disguise the sudden change. This could have contributed the "woosh".