Reduce Ambuient noise

I am using Audacity 2.0.3 installed via the dmg and with OSX 10.6.8.

I have been asked by a friend if I can fix a video he made. He recorded it on an airliner in flight and the ambient noise from the engines, wind, etc. nearly blocks the retirement speech being given to the captain. Is there a method in Audacity 2.0.3 to select the ambient sound so I can up the level of the lady doing the talking?

This would mean a lot to my friend and to me, too.


Jim Smith

The only way I know to do that in post production is with Effect > Noise Removal. It works in two steps: The first step is drag-select a portion of the show that’s noise-only and no voices. Run the tool and press Get Noise Profile. Then select the rest of the show and run the tool again and this time apply the tool using the settings on the bottom. The tool will try to subtract the first step from the second.

We can’t do forensics and we’re not CSI. It’s probable the best you can do is make it sound like a bad callphone call.


Thanks for the reply. I have seen that feature and I think my real problem is going to find a place with the ambient noise only. I’ll try searching the entire video to see if I can find a couple of seconds of jet noise only.

A bad cell phone call would be a huge improvement, though. If I can get it that far, I should be able to work on that part.

Could you attach a short sample?
Preferable with some “silence” and some speech.

The response to this kind of project is normally “no.” If the forum poster can’t give me the transcript of what the performer said, the show is usually trash. As you noted, once the show is that bad, you can’t perform the “profile selection” step and without that the tool is pretty much useless.

By “bad cellphone call,” I mean a call where the person keeps saying, “What!?!”

Another note: If you have the FFMpeg software installed… .

…then Audacity will directly open many video formats and allow you to edit the sound track. But. It won’t put it back. You need a video editor to fold the revised track back into the video show. Create a WAV at 48000, 16-bit, stereo. That’s video sound standard.


I hope this is what you’re after. I have no idea how large of a sample of the ambient-only noise you need. It seemed to me that someone was always talking.
DSCN9818a.aif (789 KB)

You’re not going to be able to make much improvement to that. It’s like when James Bond turns the shower on so that SPECTRE can’t overhear him when the room has been bugged.

Actually, I was afraid of that when I listened to the entire 3 minutes of the audio and couldn’t really fins a time when there was no talking. I’d think that Audacity needs at least two or three seconds of the noise to be suppressed, am I right?

Thanks anyway for all of the help.


Ideally you would want 5 seconds or more of “noise only”.
The Noise Removal effect can be used with shorter periods of noise for the profile, but with reduced effectiveness.
The main problem for reducing the noise in the sample that you posted is simple the amount of noise compared with the voice. The Noise Removal effect can work very well with low level constant noise on an otherwise good recording, but in this case there is more noise than voice, so you will not get better than a very small improvement at best.

Thanks, Steve. That’s mostly what I expected. It doesn’t help that the original recording was rather low quality anyway. I thought I would give it a try for a friend. Then saw I would have to take it to experts.

Perhaps, if I can get a transcript of what was being read, I can take it to y video editor and overlay the text as if it were a translation service. Better than nothing, I think. :wink: