OT - audio challenge

Arthur.S wrote on 12/6/2015, 6:00 AM
I've heard a lot of positive comments on here re software such as iZotope for noise removal. I have the below audio from a wedding recently where many times when the bride spoke a door somewhere at the back of a (large!) London church was crashing shut. I've tried to improve it with my limited noise removal skills using Audacity, but without anything like success. So...wondering if anyone here can do anything with it? Frankly I'll be amazed, but then I'm no audio guru! Link below.



JJKizak wrote on 12/6/2015, 6:43 AM
Try in Sound Forge to just expand the door shut sound then eliminate it and see what it does. Usually noone can tell the difference if you lose a small blip in the voice. Done this many times and it is almost impossible to detect. If this is like a click or a pop from a record it is no problem.
Another way is to copy and paste a tiny noise section from another portion of the track to the door slam. Piece of cake if you have Forge.
Arthur.S wrote on 12/6/2015, 9:22 AM
I don't have Sound Forge, but It's as she's saying her vows JJ (or Nuptials as it's a Catholic ceremony) have you listened to the file?
rraud wrote on 12/6/2015, 9:58 AM
You could take a similar phrase "as a sign" from some place else in the service, and replace it. .. or try iZotope RX or SCS SpectraLayers, though it will not eliminate it completely. If the picture is wide lip sync won't be an issue, otherwise cover it with audience b-roll shot.
Arthur.S wrote on 12/6/2015, 10:33 AM
A similar phrase to a bride giving her vows!!?? For this part of the ceremony I'm always close in. ;-) Interesting that you don't think it can be eliminated rraud (my own feeling too) I'm sure I read on here somewhere a similar thread re a guest coughing while the bride was speaking.
Arthur.S wrote on 12/6/2015, 10:36 AM
Ha! found it!! About halfway down from Johnmeyer:

"If someone coughs just as the bride is saying, "I do," and you own iZotope RX which can easily remove this flaw, would you leave the cough?"
JJKizak wrote on 12/6/2015, 11:51 AM
Yes. It's a piece of cake for me. Even the chair scrapings on the floor. You take a piece of environment (contains the proper room echo and noise level) from a non speaking section, copy and paste it into the chair scrapings or door closings and nobody will be able to discern the difference as there is no dialog during the door closing. Sometimes you can mask stuff with just blending in another copy and paste of what you want. Because you are copy and pasting like amounts there are no frame problems with the track. Takes about 30 seconds if you have Forge. Probably the same with other apps.
Sneddy wrote on 12/6/2015, 3:44 PM
I've fixed the audio issues that occur in the first minute or so.
I used the copy/past method described by JJKizak using exclusively Vegas 10.
I tried iZoptope RX 4 Advanced Spectral Repair but wasn't satisfied.
I used Adobe Audition 3.0.1 spectral repair on the door slam
and got acceptable results but chose the copy/past method in both noises.
I could likely get rid of the groom's throat clearing and the sniffle
but didn't since they're organic :-) Do you want the ambient drop out
shortly after the door slam? I'm delighted to fix anything else you need.
I'm a big fan of marriage - sounded like a wonderful event.
Arthur.S wrote on 12/7/2015, 2:39 PM
I understand moving a piece of environmental noise to cover a NON speaking piece of noise - It's something I've done many many times, but I'm blowed if I can be accurate enough to do that without it affecting her speaking. (such as 5:46) This piece is just a section taken from the Groom's lav mic. I usually just fade out the sniffle/cough type audio from the lav and fade in more from my camera mic. I guess I should have put up a .wav of my final audio mix rather than from just one mic. I really appreciate the work from you all though.
Arthur.S wrote on 12/7/2015, 3:28 PM
Have to say you've done an excellent job at 1:19 "I am". :-) it'd be interesting to see what you can come up with at 5:42 "as a sign of".
Sneddy wrote on 12/7/2015, 6:54 PM
Arthur, I've put up versions 2 and 3 (3 is the best IMO) with a 5:42 fix. I think a Spectral Layers 3 (SL3) expert user might have saved this - I have SL1 and never could get a grasp of it. The V3 5:42 fix uses copy/paste with some adjustments in the ambient filler in addition to grabbing the second part of another word she spoke and including that with the original on a second track (this where I faded the noise and took with it her finishing consonant). It's at www.bobcroquet.com/vegas.htm
Arthur.S wrote on 12/8/2015, 12:41 PM
Great job Bob! Thank you so much - there are some lovely people on this forum!
Sneddy wrote on 12/8/2015, 1:18 PM
Arthur, my pleasure! If you have other stuff that needs fixing, please let me know.
I really enjoy 'attempting' to solve audio issues. If you want a screen shot of what I did, just let me know. I have some spectral repair examples on