Take mono sound from captured video(s), clean and normalise it, merge back in as centred stereo

clean_sound.sh

#!/usr/bin/env bash
# put output files in subdir "done"
mkdir -p done
for i in *.mp4
do
	# Extract mono (left) channel from captured video into a .wav file
	ffmpeg -i "$i" -vn -ac 1 aud1.wav
	# Learn noise profile from first x seconds of file, which should be "silence"
	sox aud1.wav -n trim 0 1.0 noiseprof speech.noise
	# Noise reduction, with mild strength, normalise file volume to -3dB, mix as identical (centre) L R stereo
	sox aud1.wav aud2.wav noisered speech.noise 0.1 gain -b -n -3 remix 1 1
	# Merge video and audio back together, keeping video identical
	ffmpeg -i "$i" -i aud2.wav -c:v copy -c:a aac -strict experimental -map 0:v:0 -map 1:a:0 "done/$i"
	# tidy up
	rm aud1.wav aud2.wav speech.noise
done