Skip to content

Optimized AAX-to-MP3 conversion

Audible is great, and the app is not bad, but I find Plex and other apps more attractive to use for audiobooks. Now, if only I could take my books home...

Audiobooks collection in PLEX

Added Bonus: pick your own choice of book covers!

Audible AAX/C formats

I didn't know much about this, until I found the KrumpetPirate/AAXtoMP3 tool that taught me the basics, so I quote that source here:

Audible uses the AAX file format to maintain DRM restrictions on their audio books and if you download your book through your library it will be stored in this format.

Downloading audible.com/library provides the same AAX files, and nothing else. There used to be an option to download lower quality files in a non-DRM format, which would only show when visiting the library from Linux, but that appears to have beem removed.

Worse yet, newer (upcoming?) AAXC files will be even harder to use in Linux:

The AAXC format is a new Audible encryption format, meant to replace the old AAX. The encryption has been updated, and now to decrypt the file the authcode is not sufficient, we need two "keys" which are unique for each audiobook. Since getting those keys is not simple, for now the method used to get them is handled by the package audible-cli, that stores them in a file when downloading the aaxc file. This means that in order to decrypt the aaxc files, they must be downloaded with audible-cli. Note that you need at least ffmpeg 4.4.

Prerequiste: activation_bytes

To decode the AAX files you need your own authentication code, that comes from Audible. To obtain this string, I used inAudible-NG/audible-activator and saved the bytes in ~/.authcode to use later.

Performance comparison for short and long books

KrumpetPirate/AAXtoMP3 works very well so, for a while, this was the tool I used.

After a while, it started to bother me that the conversion took a long time and only used 1 CPU core, all the time. Having to convert a boatload of books and having a 12-core CPU, I was motivated to create my aax2mp3 script to parallelize the encoding of chapters.

The performance diffences is thus most notable with long books, but there is quite a visible difference even with short books, e.g.

Performance of AAXtoMP3

Short book: 3h 18m converted in 3m 50s

$ time AAXtoMP3 \
  HowtoTalkSoTeensWillListenandListenSoTeensWillTalk_ep7.aax
...
real    3m50.693s
user    3m58.307s
sys     0m13.704s

$ id3v2 -l Audiobook/Adele\ Faber\,\ Elaine\ Mazlish/How\ to\ Talk\ So\ Teens\ Will\ Listen\ and\ Listen\ So\ Teens\ Will\ Talk/How\ to\ Talk\ So\ Teens\ Will\ Listen\ and\ Listen\ So\ Teens\ Will\ Talk-01\ Chapter\ 1.mp3 
id3v2 tag info for Audiobook/Adele Faber, Elaine Mazlish/How to Talk So Teens Will Listen and Listen So Teens Will Talk/How to Talk So Teens Will Listen and Listen So Teens Will Talk-01 Chapter 1.mp3:
TRCK (Track number/Position in set): 1
TPE1 (Lead performer(s)/Soloist(s)): Adele Faber, Elaine Mazlish
TPE2 (Band/orchestra/accompaniment): Adele Faber, Elaine Mazlish
TALB (Album/Movie/Show title): How to Talk So Teens Will Listen and Listen So Teens Will Talk
TYER (Year): 2005
TIT2 (Title/songname/content description): How to Talk So Teens Will Listen and Listen So Teens Will Talk-01 Chapter 1
TCON (Content type): Audiobook (255)
TCOP (Copyright message): ©2005 Adele Faber and Elaine Mazlish (P)2005 HarperCollins Publishers
TSSE (Software/Hardware and settings used for encoding): Lavf58.76.100
CTOC ():  frame
CHAP ():  frame
CHAP ():  frame
APIC (Attached picture): (Album cover)[, 3]: image/png, 272335 bytes
Audiobook/Adele Faber, Elaine Mazlish/How to Talk So Teens Will Listen and Listen So Teens Will Talk/How to Talk So Teens Will Listen and Listen So Teens Will Talk-01 Chapter 1.mp3: No ID3v1 tag

Long book: 28h 42m converted in m s

$ time AAXtoMP3 \
  ElantrisTenthAnniversarySpecialEdition_ep6.aax 
...
real    47m11.226s
user    52m53.673s
sys     4m37.468s

$ id3v2 -l Audiobook/Brandon\ Sanderson/Elantris-Tenth\ Anniversary\ Special\ Edition/Elantris-Tenth\ Anniversary\ Special\ Edition*\ 1.mp3
id3v2 tag info for Audiobook/Brandon Sanderson/Elantris-Tenth Anniversary Special Edition/Elantris-Tenth Anniversary Special Edition-01 Chapter 1.mp3:
TRCK (Track number/Position in set): 1
TPE1 (Lead performer(s)/Soloist(s)): Brandon Sanderson
TPE2 (Band/orchestra/accompaniment): Brandon Sanderson
TALB (Album/Movie/Show title): Elantris: Tenth Anniversary Special Edition
TYER (Year): 2015
TIT2 (Title/songname/content description): Elantris-Tenth Anniversary Special Edition-01 Chapter 1
TCON (Content type): Audiobook (255)
TCOP (Copyright message): ©2005, 2015 Dragonsteel Entertainment, LLC (P)2015 Recorded Books
TSSE (Software/Hardware and settings used for encoding): Lavf58.76.100
CTOC ():  frame
CHAP ():  frame
CHAP ():  frame
APIC (Attached picture): (Album cover)[, 3]: image/png, 615956 bytes
Audiobook/Brandon Sanderson/Elantris-Tenth Anniversary Special Edition/Elantris-Tenth Anniversary Special Edition-01 Chapter 1.mp3: No ID3v1 tag

Performance of aax2mp3

Short book: 3h 18m converted in 3m 50s

$ wget -O cover.jpg https://m.media-amazon.com/images/...jpg
$ time aax2mp3.sh \
  HowtoTalkSoTeensWillListenandListenSoTeensWillTalk_ep7.aax
...
real    2m0.036s
user    4m21.804s
sys     0m4.618s

$ id3v2 -l 1.\ Chapter\ 1.mp3 
id3v1 tag info for 1. Chapter 1.mp3:
Title  : Chapter 1                       Artist: Adele Faber, Elaine Mazlish   
Album  : How to Talk So Teens Will List  Year: 2005, Genre: Vocal (28)
Comment:                                 Track: 1
id3v2 tag info for 1. Chapter 1.mp3:
TSSE (Software/Hardware and settings used for encoding): LAME 64bits version 3.100 (http://lame.sf.net)
TIT2 (Title/songname/content description): Chapter 1
TPE1 (Lead performer(s)/Soloist(s)): Adele Faber, Elaine Mazlish
TALB (Album/Movie/Show title): How to Talk So Teens Will Listen and Listen So Teens Will Talk
TYER (Year): 2005
TRCK (Track number/Position in set): 1/10
TCON (Content type): Vocal (28)
TLEN (Length): 526419
APIC (Attached picture): ()[, 0]: image/jpeg, 51854 bytes

Long book: 28h 42m converted in m s

$ wget -O cover.jpg https://m.media-amazon.com/images/...jpg
$ time aax2mp3.sh \
  ElantrisTenthAnniversarySpecialEdition_ep6.aax 
...
real    14m37.037s
user    62m19.688s
sys     0m46.445s


$ id3v2 -l 1.\ Chapter\ 1.mp3 
id3v1 tag info for 1. Chapter 1.mp3:
Title  : Chapter 1                       Artist: Brandon Sanderson             
Album  : Elantris: Tenth Anniversary Sp  Year: 2015, Genre: Vocal (28)
Comment:                                 Track: 1
id3v2 tag info for 1. Chapter 1.mp3:
TSSE (Software/Hardware and settings used for encoding): LAME 64bits version 3.100 (http://lame.sf.net)
TIT2 (Title/songname/content description): Chapter 1
TPE1 (Lead performer(s)/Soloist(s)): Brandon Sanderson
TALB (Album/Movie/Show title): Elantris: Tenth Anniversary Special Edition
TYER (Year): 2015
TRCK (Track number/Position in set): 1/79
TCON (Content type): Vocal (28)
TLEN (Length): 195651
APIC (Attached picture): ()[, 0]: image/jpeg, 89607 bytes

CPU usage of AAXtoMP3 and aax2mp3

Short book: 3h 18m converted in 3m 50s

CPU usage of AAXtoMP3 vs aax2mp3

Long book: 28h 42m converted in m s

CPU usage of AAXtoMP3 vs aax2mp3

Code

The gist of this is using xargs to parallelize the extraction and encoding of individual chapters. This would probably be better using GNU parallel, but I learned about that one a few months too late.

aax2mp3.sh

The main script aax2mp3.sh does most of the work.

Note

This script takes the activation codes from a different file: ~/audible_activation_bytes

aax2mp3.sh
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
#!/bin/bash
#
# Convert DRM'ed AAX audibooks (e.g. Audible) to DRMless MP3 audio.
# Cover art is extracted from AAX files if present,
# otherwise cover.jpg file will be use for artwork.

# File names and target destinations in local (output) and remote targets.
input=$1
ext=${1##*.}
mp3=${1/.$ext/.mp3}
output="/home/raid/audio/Audiobooks"
remote="lexicon:/home/depot/audio/Audiobooks/"

if [ "$2" == "wy" ]; then
    echo "INFO: final diretory will include (year)"
fi

if [ ! -f "$1" ]; then
    echo "FATAL: $1 is missing, this will cause encoding to fail!"
    exit 1
fi

# Obtain activation bytes.
$activation_bytes
activation_bytes=$(head -1 ~/.audible_activation_bytes)

# Extract cover art. This will overwrite cover.jpg only if cover art is found.
art="cover.jpg"
ffmpeg -activation_bytes $activation_bytes -i "$input" -an -c:v copy -y "$art"

# Extract uncompressed audio.
ffmpeg -activation_bytes $activation_bytes -i "$input" -vn -c:a mp3 -y "$mp3"

# Extract metadata and chapters.
chapters=chapters.txt
metadata=metadata.txt
ffprobe -activation_bytes $activation_bytes -i "$input" -show_chapters >$chapters 2>$metadata
genre=28 # Vocal
album=$(grep '^    title' $metadata | sed 's/.* : //')
artist=$(grep '^    artist' $metadata | sed 's/.* : //')
year=$(grep '^    date' $metadata | sed 's/.* : //')
echo "$album ($year), by $artist"

# Split and encode chapters (parallel).
track=0
numtracks=$(grep -c title chapters.txt)
tabdata=chapters.tab
grep title chapters.txt | cut -f2 -d'=' | while read title; do
    track=$((track + 1))
    start=$(grep -B5 "title=$title$" chapters.txt | grep start_time | cut -f2 -d=)
    end=$(grep -B5 "title=$title$" chapters.txt | grep end_time | cut -f2 -d=)
    wav="$track. $title.wav"
    chapter=${wav/.wav/.mp3}
    echo -e "$track\t$start\t$end\t$title"
done >$tabdata

# Process chapters in parallel.
cut -f1 $tabdata | xargs --max-procs=10 -n 1 aax2mp3-chapter.sh $mp3 $metadata $tabdata

# Move processed chapters into author/book directory.
author="$(id3v2 -l 1.*.mp3 | grep TPE1 | cut -f2 -d: | cut -f1 -d, | sed 's/^ //')"
title="$(id3v2 -l 1.*.mp3 | grep TALB | cut -f2 -d: | sed 's/^ //')"
bookdir="$(echo "$author/$title" | sed 's/ /./g' | sed "s/'//g" | sed 's/\.\././g' | sed 's/&/and/g')"
if [ "$2" == "wy" ]; then
    echo "INFO: final diretory will include (year)"
    bookdir="$bookdir.($year)"
fi
echo "$bookdir"
mkdir -p "$bookdir"
echo "$bookdir" >.bookdirs
rename 's/^/0/' ?.\ Chapter*
rename 's/^/0/' ?.\ Kapitel*
ls -lh $art *Kapitel*.mp3 *Chapter*.mp3 "$bookdir"
mv -fv $art *Kapitel*.mp3 *Chapter*.mp3 "$bookdir"

# Move the book to destinations.
# Note: DO NOT let destdir end in /
destdir="$(echo "$output/$author" | sed 's/ /./g' | sed "s/'//g" | sed 's/\.\././g')"
rsync -turva "$bookdir" "$remote"
mkdir -p "$destdir"
mv -fv "$bookdir" "$destdir"

# Clean-up
rm -f chapters.t* *Kapitel*.wav *Chapter*.wav *_ep*.mp3 $input

The --max-procs flag is set to only 10 so that a few CPU cores are left for other tasks. In a pinch this value can be increased to the number of CPU threads, if the system won't be used by anybody in the meantime, to further reduce the time it takes to encode chapters.

aax2mp3-chapter.sh

And to process each chapter, here is aax2mp3-chapter.sh

aax2mp3-chapter.sh
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
#!/bin/bash
#
# Split and encode a single audiobook chapter, to use with xargs.
# If available, cover.jpg file will be use for artwork.

# Input parameters.
mp3=$1
metadata=$2
tabdata=$3
track=$4
art=cover.jpg

# Recover book metadata from file.
genre=28 # Vocal
album=$(grep '^    title' $metadata | sed 's/.* : //')
artist=$(grep '^    artist' $metadata | sed 's/.* : //')
year=$(grep '^    date' $metadata | sed 's/.* : //')

start=$(grep "^$track[[:blank:]]" $tabdata | cut -f2)
end=$(grep "^$track[[:blank:]]" $tabdata | cut -f3)
title=$(grep "^$track[[:blank:]]" $tabdata | cut -f4)
numtracks=$(cat $tabdata | wc -l)
wav="$track. $title.wav"
chapter=${wav/.wav/.mp3}
ffmpeg \
    -nostdin \
    -i "$mp3" \
    -vn \
    -c:a pcm_s16le \
    -f wav \
    -ss $start \
    -to $end \
    -y "$wav"
lame \
    -S \
    --tt "$title" \
    --ta "$artist" \
    --tl "$album" \
    --ty $year \
    --tn $track/$numtracks \
    --tg $genre \
    --ti $art \
    "$wav" "$chapter"