Noise removal using stored noise profile

Hi all!
I am not an Audacity expert at all.

Recently I had to record a lot of Desktop sessions (with me speaking and explaining), and no matter what, I end up with a lot of background noise.
I found the noise removal audacity tool to be just perfect for my needs (I think it’s just some form of electronic noise, so the tool is very efficient).

But the whole process is a bit long:

  1. open file with audacity
  2. select region for noise analysis, and analyze it
  3. apply noise reduction
  4. export audio file, for example as .mp3
  5. open original video with kdenlive
  6. split video / audio
  7. delete old audio, put in new audio cleaned with audacity at step 4
  8. export everything as .mp4

Since the noise profile is alwasy the same (same laptop, same mic, same software) is there a way to store the noise profile in some way?
In a perfect world it would be supernice to export the noise profile and then using it from batch, with some switch like “–noise-profile=noise1.prof --reduce-noise=24dB” (just to give you an idea).

In this way I could script the entire process…

Thank you!

I believe Audacity holds onto the last known good noise profile. So if nothing changes, you should only have to get the profile once.

Try it. Don’t make a new profile. The second and third, etc corrections should continue to be useful.

Several tools work like that. If you open the Equalizer, you should see the last curve you used.

Post a sound test. We may be able to do corrections some other way—or tell you where it’s coming from. Blue Yeti microphones have a particular noise that takes a custom tool we made rather than struggle with Noise Reduction.

https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

Koz

export audio file, for example as .mp3

Any reason you’re using MP3 for this step? Since your goal is a compressed video file, you are taking your high quality performance and compressing it twice, adding compression distortion at each step. We recommend WAV (Microsoft) for production and steps in the middle to avoid at least one distortion step. If that’s overkill, you might boost the MP3 quality as high as possible and listen for any difference in the show.

You’re listening for bad cellphone, talking into a wineglass compression distortion. It’s tricky. You may need to switch back and forth between the original raw performance and the show files to hear the difference.

If there is no obvious distortion, then you’re good to go as is. Just a note this compression distortion is cumulative. If someone wanted to rip your sound and do something else with it, they are much more likely to have distortion problems on the third pass, say through MP3 again.

MP3 production is a time bomb.

We get complaints where people complain bitterly they can’t make a custom playlist from highly compressed, on-line MP3 music. Yes. That’s correct. iTunes recently offered better quality music at higher cost so you could do that.

Koz

First of all let me clarify: I am not trying to get professional results, I am just doing videos for my IT collegues (internal training).
But the noise coming out from the registration is so bad that I cannot give them the video as it is.

Let me give some details.

  1. The noise stays the same both from internal mic (it’s a laptop) and external mic (for example, the one integrated in my Sennheiser headphones, cheap model but with quite a good noise cancelling mic). I think that the noise is coming from some source I… didn’t think of :smiley: . When I use skype or gotomeeting (always from Linux) the audio quality from my mic is good or very good as far as others tell. It’s just when I register the full desktop that I get noise. Tried with kazam, RecordMyDesktop, and a few other apps - nothing changes.

  2. I used .mp3 as audacity audio output since is the format I am more familiar with, the default one was .aiff that I don’t know… so I picked from the list one that I knew. Distortion is not an issue here, there is just my voice saying “click here, restart service…” and so on.

  3. About the noise profile that stays the same between different files, it’s a good idea, this saves me some time… but my original idea was to script the whole process, so if I cannot launch it with some command line switches… from my point of view, is almost the same…

I am not trying to get professional results

Thank you. I haven’t been able to use this quote in a long time.

I don’t need professional audio. I just need clear sound at good volume and no distortion or noise.

Most pros I know would kill to be able to do that.

I think that the noise is coming from some source I didn’t think of.

Both microphones sound the same…

Of course, if they’re both communications microphones with heavy environment cancellation, you would never know. They’re both beating your voice with a stick.

good or very good as far as others tell.

Nobody I know ever complains about received audio. As long as they can derive occasional meaning from the voice, it’s good. A friend has a really bad cellphone that deletes every third slice. It sounds like sandpaper, but I can just make out I’m supposed to be there after lunch. I wouldn’t use that as a quality control measurement unless they can record it and send it back.

.mp3 as audacity audio output since is the format I am more familiar with, the default one was .aiff that I don’t know

Odd AIFF is an Apple format. I thought Audacity defaulted to WAV on all platforms… Oh well. You can continue to use MP3, just do it at a high quality. I think the highest in the default list is 320?

but my original idea was to script the whole process,

Maybe the difference between scripting and Audacity Chains.

http://manual.audacityteam.org/man/chains_for_batch_processing_and_effects_automation.html

Koz

I wouldn’t use that as a quality control measurement unless they can record it and send it back.

I actually heared a recording made by someone else, and my voice was much much better, but I don’t know if GoToMeeting applies any noise cancelling.

Following your link I found also the scripting page, so maybe a combiantion of this two:

http://manual.audacityteam.org/man/scripting.html
http://manual.audacityteam.org/man/chains_for_batch_processing_and_effects_automation.html

could be what I am looking for.
I will do some testing for it and revert back to you, and I give you some samples also, so you get an idea on what I mean :wink: