Remove airplane noise in Adobe Audition ?

[Eric Kovacs]

Hello,

I am trying to remove the noise of an airplane engine in adobe audition. It occurred right as the bride and groom were exchanging vowes.

I am new to audition and hoped someone might let me know if this is possible or since it occurrs during their vowes it will not work.

Any thoughts ?

[Seth Bloombaum]

If you're on a recent-ish release of Audition, yes, there's some help there with the Noise Reduction filter (process).

I've not been through these tutorials, but AdobeTV is usually my first stop:
http://tv.adobe.com/videos/noise-reduction/

I'd look at the two near the bottom of the page for Audition CS5.5.

IIRC, there's also some good info on the NR filter/process in the help file.

Conventional wisdom is to take out small bites, successively, rather than trying to be aggressive and get it all in one application of the process.

Likely that sound will be improved, maybe significantly. When it works NR is pretty magical. But there are lots of situations in which it doesn't help so much.

[Bob Hart]

Seconding Seth's comment.


If it is a jet, it might be a bit easier but a prop plane sound is a big challenge. Both sounds change pitch and intensity. A single noise sample will not hold through an entire recording. You may need to chop your recording up, take noise samples from gaps in the dialogue and apply these uniquely to each segment. There will remain surges and variations. You will probably need to crossfade over each junction.

Look up "Cool Edit Pro" and see if there are any forums on it. Adobe bought it a while back and renamed it Audition and have refined it since. The interface is a bit difference but the principles rermain the same.

My personal preference would be :-

Create segments in your recording. The method might be to first make a working copy of your sound file so you can mutilate it. Keep the original intact and stored safely elsewhere so you do not pull it up and ruin it accidentally.

With each incremental alteration to your recording that you make, save a copy with a unique filename so that you can pull it back up if you spoil the track by going down a dead-end. This is better than having to start over from scratch. Adobe may have refined the software to include preview files which enable "undo" and "redo"across editing sessions but I would not count on it.

I have heard that changing from say 48k to 96k and working in the higher digital sample rate ( excuse my ignorance ) sometimes assists when applying corrections. The process may introduce artifacts. Please heed better advice than mine on this. My comment will hopefully prompt their more relevent responses.

Within each of those segments, there may be wide enough gaps that you can take a noise sample. That you can then apply to the the tail of one segment and the head of the next segment, with the noise file from the previous segment meeting in the middle.

You have the option of saving the noise file but I think the filename is taken from the original clip filename.

You will have to careful not to get yourself mixed up with noise sample filenames. Maybe you can apply two noise samples to each segment, one from head, one from tail. Experimenting maybe the only way.

Once you have it the best you can make it, you may be able to apply a mild dynamic effect like compression. I think from memory, you can apply compression selectively to frequencies by a graphic interface by clicking on a curve, then pulling and releasing the curve line into a custom shape. There are presets you can pull in and then modify and save as a custom preset.

Compression is likely to introduce a "pumping" artifact where you have gaps in the dialogue. You do not want to the "attack" and "decay"settings to be too aggressive. You may need to highlight then manually change the audio level lower between each speech segment where there is a longer break.

You may also be able to use the 30 channel graphic equaliser to take out some unwanted frequencies.

You may try stereo delay effects by converting a copy of your corrected file into a stereo track. Sometimes very subtle delay effects can help mask a background noise but it is very easy to introduce an unwanted ambience into your recording.

Each effect can have a interaction with other effects. You need to develop an organisational discipline in how you save and identify each revision to the recording you alter.

Finally, you may discover that there remain irritating discontinuities and jarring changes in the remaining noise under the dialogue. These will exist because you cannot eliminate noise completely without causing speech to become "pingy" or "bubbly" with digital artifacts. You may need to lay a new continuous background noise track under the recording as an artificial ambience to mask the changes.

Please heed the better advice of others who are specialists in this field. I know only just enough to endanger myself.

Here is an example of an old 78 rpm music record recovery which I found troublesome. There was a high noise level from scratches, surface pitting due to fungus in the paper wrapper over time.

I used a noise sample, a scratch removal preset, selected segments with differing graphic equaliser settings to try and preserve the stringed instruments in the orchestral backing, a bit of delay which you will observe in the voice and lowered gain in extended breaks between dialogue.

The select by highlight tool whatever it is called in Audition, then extreme zoom stretch of the waveform can take you into individual sample points where you can manually redraw the waveform to attenuate really bad transients but it requires much patience.

You will observe a "buzzing" artifact in the voice. This is due to the track on the record being worn and stylus slap which there is not much to be done about.

These corrections are not subtle but were the best I could do with my limited skill set. There is a balance to be sought between eliminating noise and yet not processing the life out of your sound source.

BOB DALE MARY'S BOY CHILD - YouTube

[Jeff Pulera]

Not inexpensive, but I've seen demos of this software doing miracles to remove background issues like an ambulance siren completely - SpectraLayers Pro 3 Overview

Might be virtually impossible with other apps.

Thanks

[Jim Andrada]

Izotope might also be a good one to try - but it isn't exactly cheap either. If Audition is what the OP has, then it's a good place to start.

If the OP would post a clip some of us could take a shot at it with our favorite tools

[Pete Bauer]

Sorry, on the hoof so can't be detailed. But take a look at Audition's spectral view. You might get lucky and find that the airplane noise stands out in that view, in which case you can delete pretty much just that noise.

[Greg Miller]

I am always puzzled that people ask, in words, whether a certain audio trick is possible ... but without posting any audio. If I wanted someone to edit a book I was writing, I certainly would not post an audio recording of my keyboard clicks. I would post the words. So I would suggest to the OP (as I have suggested to many other posters in the past), please post some of the audio in question, and perhaps we can give you more specific suggestions, instead of generalizations based on past (but different) audio files.

Many of the suggestions above are good hypothetical starting points. Your particular file might be easy to repair ... or impossible. Hearing the file will help us make a more realistic assessment.

[Jim Andrada]

+111 to Greg's post