How to Convert an eBook to an Audiobook

TL;DR

Recently I've been reading a fictional book that I've really enjoyed. Sometimes I'm in the car and I wish I could read while driving. That is what I love about audiobooks. Driving and listening to a book is one area where I can actually multi-task.

Unfortunately, this particular book I'm reading doesn't have an audiobook available. I've found that I can use Kybook 3 and it has the best text-to-speech option I have found.

It made me wonder though. Would it be possible to convert a whole eBook to an audiobook that I could play in my favorite audiobook player? Certainly, the tools are all there, I just need to put them together.

Step 1: Getting a text-only version of the book

I won't go into details on this step other than to simply say that I used Calibre to convert my DRM protected book to a text-only version. That is not really the focus of this post, so I'll leave it to you to do some research if you have not used Calibre before. If you want to find some free books to try this out with, you can find many on Project Gutenberg.

Step 2: Break the book down into chaptered files

Here a little tip. I had a book with 30 chapters. I ran this command for file placeholders:

for i in {01..30}; do touch "Chapter $i.txt"; done

Then I just cut and paste chapters into those files.

Be sure to remove characters that you don't want read. For example, one book I was converting would use underscores to separate sections. So after converting the audiobook had parts that would say "underscore, underscore, underscore, underscore, underscore, underscore, underscore, underscore."

Note that chapter files will be compiled in alphabetical order and chapter bookmarks will be named after the file name. So if you have named chapters such as Introduction, Prologue, etc., then you will need to prefix these with numbers like 00 Introduction, 01 Prologue, etc.

Step 3: Find or create a cover image for your book

Find a cover for your book. For audiobooks, these are generally square, but it doesn't really matter. I prefer square for consistency in my audiobook player. Name the cover image cover.jpg and put it in the same directory as all your chapter files.

Step 4: Run this python script

Note: I've put this script on github, so check here for the latest version

Wait! Before running the script, you will need to have FFmpeg and mp4v2 installed. You can install with brew.

brew install ffmpeg mp4v2

Run the script and pass in the directory containing your chapter files and cover as the first argument.

./ebook2audiobook.py "Name of your directory"

You will be prompted for the title and author.

What is happening

Here is a basic rundown of what the script is doing:

For each text file in the folder, it will run the say command to convert the text file to an audio file. If you want to try this will a simple text file, run:

say -f filenmame.txt -o output.aiff

If you want to change the voice, pass in the -v option. If you want to see the voices available, run:

say -v ?

This is probably the longest process of the script.

Next, the script will run through all of the .aiff files and compress and convert them to .m4a files using FFmpeg. This is essentiall what is being for each file:

ffmpeg -i inputfile.aiff -c:a libfdk_aac -b:a 64k -f mp4 outputfile.m4a

Once all the audio conversions are complete, the script will loop through the .m4a files and generate a file called FILES that contains a list of all these files to use for concatenation. It will then use ffprobe to get the duration of each of these files. It will use this information to build a METADATA file that FFmpeg will use to create the chapter bookmarks and set other metadata information such as the title, author, and album (album is the same as the title).

Here is the ffprobe command that gets the duration:

ffprobe inputfile.m4a -show_entries format=duration -v quiet -of csv="p=0"

Here is the FFmpeg command that will do the concatenation:

ffmpeg -f concat -safe 0 -i PATH_TO_FILES_LIST -i PATH_TO_METADATA_FILE -map_metadata 1 -vn -y -acodec copy outputfile.m4b

The script will then call FFmpeg one more time to concatenate all the m4a files into an m4b file and create all the chapter makers and add the metadata.

Finally, the script will call mp4art (part of mp4v2) to add the album cover image to the m4b file which looks like this:

mp4art -q --add cover.jpg outputfile.m4b