Shell script that downmixes and transcodes audio into high-quality AAC tracks

transcode-audio.sh

#!/bin/sh

# Converts audio from BluRay or lossless source into high quality
# AAC files with custom stereo downmix parameters
#
# Use like this:
#   ./transcode-audio my_movie.mkv ger51 eng51 skip jpn20
#
# Each parameter after the movie name tells the script which kind
# of audio channel to handle.
#
# Dependencies:
#   * ffmpeg
#   * qaac (working = w/apple's core audio toolkit)
#   * wine (to run qaac)
#

# File from which the audio is taken
inputFile="$1"

# Quality, 127 is best, 95 is about equivalent to 320 kbit mp3
#
#   Channels | Quality | Expected bitrate
# -----------|---------|-------------------
#      2     |    95   |     160 kbps
#      2     |   127   |     260 kbps
#     5.1    |    95   |     480 kbps
#     5.1    |   127   |     720 kbps
#
qaacStereoQuality=127
qaac5dot1Quality=127
#qaac5dot1Quality=95

# Whether to even encode the 5.1 tracks
#
# If this is false, all 5.1 audio is downmixed to stereo audio.
# And the 5.1 audio tracks are not provided at all
#
want5dot1=true

# Whether to delete the *.flac files
#
# These are created via ffmpeg and contain the processed audio
# tracks before they're transcoded to AAC (using qaac instead of
# ffmpeg's own transcoder due to qaac supposedly being higher quality)
#
# If you want to do some checking and/or processing (i.e. normalize
# volume levels) in Audacity, set this to false.
#
deleteFlacs=false

# Standard downmix formula (also used by ffmpeg with -ac 2),
# this is sourced from ATSC 7.8, but lacking mastering data,
# fixes 'clev' and 'slev' to 0.707. Makes dialogue too quiet:
# -af "pan=stereo|FL<1.0*FL+0.707*FC+0.707*BL|FR<1.0*FR+0.707*FC+0.707*BR"
#
# Nightmode dialogue formula from Doom9. This results in voices at
# levels more typical of traditional stereo recordings, but it's
# mostly just guesswork and some listening tests:
# -af "pan=stereo|FL=FC+0.30*FL+0.30*BL|FR=FC+0.30*FR+0.30*BR"
#
# The middleground between ATSC and decent dialogue volume.
# I don't perceive the dialogue as much quieter than nightmode's,
# but this should yield much better stereo separation:
# -af "pan=stereo|FL=0.85*FC+0.65*FL+0.5*BL|FR=0.85*FC+0.65*FR+0.5*BR"
#
volume350="pan=stereo|FL=2.98*FC+2.28*FL+1.75*BL|FR=2.98*FC+2.28*FR+1.75*BR"
volume325="pan=stereo|FL=2.76*FC+2.11*FL+1.63*BL|FR=2.76*FC+2.11*FR+1.63*BR"
volume300="pan=stereo|FL=2.55*FC+1.95*FL+1.5*BL|FR=2.55*FC+1.95*FR+1.5*BR"
volume250="pan=stereo|FL=2.13*FC+1.63*FL+1.25*BL|FR=2.13*FC+1.63*FR+1.25*BR"
volume200="pan=stereo|FL=1.7*FC+1.3*FL+1.0*BL|FR=1.7*FC+1.3*FR+1.0*BR"
volume100="pan=stereo|FL=0.85*FC+0.65*FL+0.5*BL|FR=0.85*FC+0.65*FR+0.5*BR"
volume90="pan=stereo|FL=0.77*FC+0.59*FL+0.45*BL|FR=0.77*FC+0.59*FR+0.45*BR"
volume80="pan=stereo|FL=0.68*FC+0.52*FL+0.4*BL|FR=0.68*FC+0.52*FR+0.4*BR"

germanStereoDownmixFormula=$volume100
stereoDownmixFormula=$volume100

# -------------------------------------------------------------------
# Figure out where we will get the 2.0 audio files from
# -------------------------------------------------------------------

# Shift the arguments one down, pushing the input filename away
shift

audioTrackIndex=0

# Figure out if we need to downmix the 2.0 channels ourselves or
# if these are provided by the source video/audio container.
#
# If you want the improved downmix formula from this script, just
# say "skip" instead of "xxx20" for the 2.0 channels and they will
# be downmixed from the 5.1 audio tracks
#
if [[ "$@" == *"eng20"* ]]
then
	englishStereoNeeded=false
else
	englishStereoNeeded=true
fi

if [[ "$@" == *"ger20"* ]]
then
	germanStereoNeeded=false
else
	germanStereoNeeded=true
fi

if [[ "$@" == *"jpn20"* ]]
then
	japaneseStereoNeeded=false
else
	japaneseStereoNeeded=true
fi

# -------------------------------------------------------------------
# Do the actual transcoding
# -------------------------------------------------------------------

# For each argument passed on the command line
while test $# -gt 0
do

	case "$1" in

		skip)

			;;

		eng51)

			if $englishStereoNeeded ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-strict \
					-2 \
					-af "$stereoDownmixFormula" \
					-y \
					english-stereo.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaacStereoQuality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					english-stereo.flac

				if $deleteFlacs ; then
					rm english-stereo.flac
				fi

			fi

			if $want5dot1 ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-y \
					english-5dot1.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaac5dot1Quality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					english-5dot1.flac

				if $deleteFlacs ; then
					rm english-5dot1.flac
				fi

			fi

			;;

		eng20)

			ffmpeg \
				-hide_banner \
				-i "$inputFile" \
				-vn \
				-map a:$audioTrackIndex \
				-y \
				english-stereo.flac

			wine /opt/qaac-2.72/qaac.exe \
				--tvbr $qaacStereoQuality \
				--quality 2 \
				--no-smart-padding \
				--no-dither \
				english-stereo.flac

			if $deleteFlacs ; then
				rm english-stereo.flac
			fi

			;;

		ger51)

			if $germanStereoNeeded ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-strict \
					-2 \
					-af "$germanStereoDownmixFormula" \
					-y \
					german-stereo.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaacStereoQuality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					german-stereo.flac

				if $deleteFlacs ; then
					rm german-stereo.flac
				fi

			fi

			if $want5dot1 ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-y \
					german-5dot1.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaac5dot1Quality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					german-5dot1.flac

				if $deleteFlacs ; then
					rm german-5dot1.flac
				fi

			fi

			;;

		ger20)

			ffmpeg \
				-hide_banner \
				-i "$inputFile" \
				-vn \
				-map a:$audioTrackIndex \
				-y \
				german-stereo.flac

			wine /opt/qaac-2.72/qaac.exe \
				--tvbr $qaacStereoQuality \
				--quality 2 \
				--no-smart-padding \
				--no-dither \
				german-stereo.flac

			if $deleteFlacs ; then
				rm german-stereo.flac
			fi

			;;

		jpn51)

			if $japaneseStereoNeeded ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-strict \
					-2 \
					-af "$stereoDownmixFormula" \
					-y \
					japanese-stereo.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaacStereoQuality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					japanese-stereo.flac

				if $deleteFlacs ; then
					rm japanese-stereo.flac
				fi

			fi

			if $want5dot1 ; then

				ffmpeg \
					-hide_banner \
					-i "$inputFile" \
					-vn \
					-map a:$audioTrackIndex \
					-y \
					japanese-5dot1.flac

				wine /opt/qaac-2.72/qaac.exe \
					--tvbr $qaac5dot1Quality \
					--quality 2 \
					--no-smart-padding \
					--no-dither \
					japanese-5dot1.flac

				if $deleteFlacs ; then
					rm japanese-5dot1.flac
				fi

			fi

			;;

		jpn20)

			ffmpeg \
				-hide_banner \
				-i "$inputFile" \
				-vn \
				-map a:$audioTrackIndex \
				-y \
				japanese-stereo.flac

			wine /opt/qaac-2.72/qaac.exe \
				--tvbr $qaacStereoQuality \
				--quality 2 \
				--no-smart-padding \
				--no-dither \
				japanese-stereo.flac

			if $deleteFlacs ; then
				rm japanese-stereo.flac
			fi

			;;


	esac

	shift
	let "audioTrackIndex = $audioTrackIndex + 1"

done