I found your entry interesting thus I’ve added a Trackback to it on my weblog :)…

Why I started fiddling with script is to actually automate the process as much as possible and as many movies as possible, so this way that requires much user interaction is no good (for me). I’m sure I’ll try/use it for single file conversions :)

I call it mkmkv.sh, but feel free to call it whatever you like. It relies only on a single argument (the video file name), and just make sure to invoke the script from the same directory as the video file.

cd /path/to/mkv-dir/
mkmkv.sh filename.mkv

#!/bin/bash
MP3="${1}.mp3"
WAV="${1}.wav"
 
echo ""
echo "Identifying Video File: \"$1\""
echo ""
 
ffmpeg -i "$1" 2>&1 | grep "Audio"
 
echo ""
echo "Which audio track do you wish to convert to Stereo MP3?"
read -p "Enter desired FFmpeg audio stream ID in X:Y format: " STREAM
echo ""
echo "Identifying MKV Track ID's..."
echo ""
 
mkvmerge -i "$1"
 
echo ""
read -p "Enter Video ID to be Included: " VID
read -p "Enter Video ID $VID Description: " VDESC
echo ""
read -p "Enter Audio ID to be Included: " AID
read -p "Enter Audio ID $AID Description: " ADESC
echo ""
read -p "Enter name for your NEW MKV file (e.g. movie.mkv): " OUTPUT
echo ""
echo ""
echo "Verify proposed mkvmerge command..."
echo ""
echo "\"mkvmerge\" -o \"$OUTPUT\" --language $VID:eng --track-name \"$VID:$VDESC\" --default-track $VID:yes --language $AID:eng --track-name \"$AID:$ADESC\" --default-track $AID:yes -a $AID -d $VID -S \"$1\" --language 0:eng --track-name \"0:English Stereo MP3\" --default-track 0:no -a 0 -D -S \"$MP3\" --track-order 0:$VID,0:$AID,1:0"
 
echo ""
read -p "Ctrl+C to cancel, or ENTER to continue."
 
echo ""
echo "Running command: ffmpeg -i \"$1\" -map $STREAM -acodec pcm_s16le -ac 2 \"$WAV\""
ffmpeg -i "$1" -map $STREAM -acodec pcm_s16le -ac 2 "$WAV"
 
echo ""
echo "Running command: lame -V0 -q0 --vbr-new \"$WAV\" \"$MP3\""
lame -V0 -q0 --vbr-new "$WAV" "$MP3"
 
echo ""
echo "Running mkvmerge..."
echo ""
"mkvmerge" -o "$OUTPUT" \
--language $VID:eng \
--track-name "$VID:$VDESC" \
--default-track $VID:yes \
--language $AID:eng \
--track-name "$AID:$ADESC" \
--default-track $AID:yes \
-a $AID -d $VID -S "$1" \
--language 0:eng \
--track-name "0:English Stereo MP3" \
--default-track 0:no \
-a 0 -D -S "$MP3" \
--track-order 0:$VID,0:$AID,1:0

I was, too, aware that such simple script can’t cover all aspects, but for me it does cover 90% of mkvs :)

As you surely know, this line does ripping of audio:

ffmpeg -i $1 -acodec pcm_s16le -ac 2 audio.wav

and (as you also surely know) with this command you can inspect the streams:

ffmpeg -i

Example result:

Stream #0.0(eng): Video: h264, yuv420p, 1280×720, PAR 1:1 DAR 16:9, 24 tbr, 1k tbn, 47.95 tbc
Stream #0.1(eng): Audio: ac3, 48000 Hz, 5.1, s16
Stream #0.2(eng): Audio: dca, 48000 Hz, 5.1, s16

So, you just need to add this to choose a different (in this case, the 3rd, dts) stream:

-map 0.2

Solution: add another input parameter for script specifying the audio stream mapping (tho you have to check the file beforehand with ffmpeg -i ).

It just depends on the number of files you are working on so it can turn out that u spent more time on script than you would on converting each file manually (via e.g. gui) :)
But at least we polish our scripting skillz ;)

For example, many of my MKV’s have multiple audio tracks, and the default isn’t always English. So simply running ffmpeg without specifying which to use, might cause ffmpeg to pull the wrong source audio track. Also, many foreign movies end up needing subtitle tracks, and excluding them all by default can suck. :-) Again, choosing the right one when there are multiple subtitle tracks is also important.

It will take some work, but you’ve inspired me to get a public version going. Thanks again!

#!/bin/bash
echo
echo $1
echo
mkvmerge -i $1

echo
echo “####### STARTING AUDIO EXTRACTION #######”
echo
ffmpeg -i $1 -acodec pcm_s16le -ac 2 audio.wav

echo
echo “####### COMPRESSING AUDIO #######”
echo
lame -V0 -q0 –vbr-new audio.wav audio.mp3

echo
echo “####### APPENDING AUDIO TO CONTAINER #######”
echo
mkvmerge -o new.$1 \
$1 \
–language 0:eng \
–track-name “0:2-Channel Stereo (MP3)” \
–default-track 0:yes \
-a 0 -D -S audio.mp3 \

echo
echo new.$1
echo
mkvmerge -i new.$1

Just call it from (unix) command line with mkv filename as argument.

Regarding AC3, the whole point in the mentioned scenario was that certain media players won’t decode the AC3 DD audio for you. So unless you have a receiver in between the player and the TV, you can’t hear anything. Unless I’m missing something about 2-Channel AC3, wouldn’t that still hold true? Sounds like a test is in order. I’ll add it to the post if so.

Thanks again!