Automatic noise reduction with results like Audacity

Question

I like Audacity because of its noise reduction. With a recording done in studio, a noise profile, and reduction at 16 dB (with sensitivity 6.00 and 3 frequency smoothing bands), I get a perfect result without voice distortion.

I would like to automate this noise reduction process: give the software a file with the background noise, generate a noise profile, and reduce that noise in multiple other files.

Audacity allows scripting; even then, it crashes on my system, and no longer allows noise reduction:

This effect is not currently available from scripting.

I tried SOX from this thread:

sox Take80-noise.wav  -n noiseprof noise.prof
$ sox Take80.wav Take80-clean.wav noisered noise.prof 0.21

With my studio recordings and the same noise input, a level of 0.21 has very little effect, and a level of 0.6 has more effect on background noise without eliminating it and it does distort the voice.

How can I automate noise reduction with similar results as Audacity?

Answer 1

I stumbled on this thread a few years ago trying to do the same thing. I like your Applescript solution, but a few months later I found this Github repo https://github.com/tals/audacity-noise-reduction which when built spits out a standalone executable that provides the same parameters and results as Audacity. I wrote a Python API around it to automate audio processing and uploading on a large scale, and it does exactly what I need. Hopefully this can be helpful to someone else!

Answer 2

For lack of results in two months, I copy the answer from Automate pre-processing of sound (noise reduction, compressor, limiter).

I programmed an AppleScript that controls the Audacity Graphical User Interface and achieves the same result. It loads a track with the noise and gets the noise profile, loads the track with the voice, removes the noise with the previous settings, and I also included a limiter and a compressor.

set noise_track to "/path/to/noise/track.wav"
set voice_track to "/path/to/voice/track.wav"
set export_name to "voice-clean"

set short_delay to 1
set middle_delay to 5
set long_delay to 10

tell application "System Events"
    
    tell application "Audacity" to activate
    
    tell application "Audacity" to open noise_track
    
    delay short_delay
    keystroke "a" using command down
    
    
    tell process "Audacity"
        click menu item "Noise reduction..." of menu "Effect" of menu bar 1
    end tell
    delay short_delay
    
    repeat 2 times
        key code 48
        delay short_delay
    end repeat
    key code 49
    delay short_delay
    
    tell application "Audacity" to open voice_track
    delay short_delay
    
    --------------------
    -- Noise reduction
    --------------------
    
    keystroke "a" using command down
    delay short_delay
    
    tell process "Audacity"
        click menu item "Noise reduction..." of menu "Effect" of menu bar 1
    end tell
    delay short_delay
    
    key code 36
    
    delay long_delay
    
    --------------------
    -- Limiter
    --------------------
    
    tell process "Audacity"
        click menu item "Limiter..." of menu "Effect" of menu bar 1
    end tell
    delay short_delay
    
    key code 36
    delay long_delay
    
    --------------------
    -- Compressor
    --------------------
    
    tell process "Audacity"
        click menu item "Compressor..." of menu "Effect" of menu bar 1
    end tell
    delay short_delay
    
    key code 36
    delay long_delay
    
    keystroke "e" using {command down, shift down}
    delay middle_delay
    keystroke export_name
    delay short_delay
    key code 36
    
    delay long_delay
    
    --------------------
    -- Close two projects
    --------------------
    repeat 2 times
        keystroke "w" using command down
        delay short_delay
        key code 48
        delay short_delay
        key code 49
        delay 2 * short_delay
    end repeat
end tell