Guides

How to make an M4B audiobook from MP3 files, with chapters

A folder of MP3s plays, but it does not behave like a book. What an M4B is, how to get order and titles right, which bitrate to pick, and what to do when a player ignores your chapters.

· 8 min read

Why one M4B beats a folder of MP3s

Forty MP3 files are forty separate items to whatever plays them: a position per file, if any, and no chapter menu, because nothing in an MP3 says where a chapter begins. An M4B fixes that in the container rather than in the app: one file, so one library entry and one remembered position; a chapter list, so jumping to chapter 12 is a menu choice rather than a scrub; and a media-type flag that sends it to the audiobook shelf in Apple Books instead of the music library.

Technically, M4B is MP4 with AAC audio, the same container as .m4a; the extension tells Apple software and most audiobook apps to treat the file as a book. What matters is the data inside: chapter atoms with start times and titles, an embedded cover, and a few tags including the audiobook media type. Renaming an .m4a to .m4b gives you none of that.

Get the file order right before anything else

Chapter order is file order. Most tools, Kataribe included, look at track numbers first: if every file has one, the set sorts by disc and then track, which handles multi-disc rips where filenames restart at 01 on each disc. If even one file lacks a track number, the whole set falls back to filename order.

Filename order should be natural sort, which reads digits as numbers so 2 comes before 10 without zero-padding. Kataribe does this. Then watch for files that break the pattern: Prologue.mp3 sorts after Part 12.mp3, so rename it with a number prefix or move it by hand in the list, and look for a missing file 07 or one that appears twice while the list is still short.

Track numbers are read from ID3v2 tags only; M4A, FLAC and OGG inputs sort by filename.

Where chapter titles come from and how to fix them

The best source is the ID3 title tag in each MP3, where a publisher or careful ripper will have written something like Chapter 3: The Letter. When it is present, that text is the chapter title, unchanged.

Without a title tag, the filename is used, after cleanup: extension dropped, a leading track, part, cd or disc number and its separator stripped, underscores turned to spaces, and any run of words shared by every file removed back to the last dash or colon. So Book Title - Part 04.mp3 becomes Part 4, and a name that cleans down to nothing becomes Chapter N.

Both can be wrong: CD rips often carry Track 07 in the title tag, and filenames carry the ripper's shorthand. Either way, edit the titles in the list before you build; they go into the chapter atoms and are exactly what the player's chapter menu shows.

Bitrate, mono versus stereo, and honest size math

AAC at 64 kbps is the usual choice for narration and what many commercial audiobooks are sold at; a single voice gains nothing audible from more. For music beds, a full cast or sound design, 96 kbps is a safer floor and 128 kbps is worth it if music is prominent. Kataribe offers 32, 48, 64, 96 and 128 kbps.

The math is bitrate times seconds divided by eight. A 10-hour book at 64 kbps is 64,000 × 36,000 ÷ 8, about 288 million bytes, call it 290 MB. At 32 kbps it is around 145 MB and starts to sound rough; at 128 kbps, around 575 MB.

Two rules: do not go below the source bitrate (64 kbps MP3s get 64, not 32), and know that the channel layout is passed through. Mono stays mono and stereo stays stereo, with no downmix step, so a stereo rip of mono narration spends half its bits on a duplicate channel.

If every input is already AAC, the audio is copied rather than re-encoded and the bitrate setting is ignored.

Cover art and the metadata players actually display

A square JPEG or PNG is the safe cover. It is embedded as-is, without resizing, so use at least 600 pixels on a side and keep it to a few hundred kilobytes. The first MP3 with an embedded picture supplies the default, and you can replace it.

Few tags matter. Title and author are what every player lists; the author lands in the artist and album artist fields, and the title doubles as the album so music-oriented apps group the file. Narrator goes into the composer field, the convention Apple set and Audiobookshelf reads. Year and genre are written, with Audiobook as the default genre, and the media type is set to audiobook.

Not written: a series name or position, and anything that was inside the inputs; existing tags and chapter lists are discarded before joining, so the output holds only what you set in the form.

Step by step: from a folder of MP3s to a chaptered M4B

  1. Put one book in one folder. Remove anything that is not part of the book, and if the files have no track numbers, check that the filenames sort naturally.
  2. Drop the files into the builder. Kataribe reads the ID3v2 tags, sorts by track number or natural filename order, pre-fills title, author, narrator and year, and picks up the first embedded cover. The engine, about 30 MB of WebAssembly, downloads once and is cached.
  3. Check the order and the chapter titles. Confirm the sequence, using the move up and move down controls on anything out of place, and remove any file that does not belong. Then rename any chapter that says Track 07; whatever is in each box when you build is what the chapter menu shows.
  4. Fill in the book details and cover. Title and author are the minimum; add narrator, year and a cover if the tags did not supply them. Output is named Author - Title.m4b.
  5. Pick the bitrate and build. 64 kbps for narration, 96 or 128 for music-heavy productions. Files are processed one at a time, so a long book works in a desktop browser, but it takes minutes and the tab has to stay open. Check that the chapter count matches the file count.

Other ways people do this, and when to prefer them

A browser tool suits one book at a time. Other options fit other situations.

  • Audiobook Builder (Mac, paid, Splasm Software). A native app that joins files into chaptered M4Bs with cover art and can split long output into parts. Worth it on a Mac if you process many books.
  • ffmpeg on the command line. Free everywhere, and the same engine Kataribe runs in the browser: a concat list plus an FFMETADATA file with one CHAPTER block per file, muxed with -f ipod. Prefer it to script the process, handle files too large for a browser tab, or control every encoder flag.
  • m4b-tool. An open-source wrapper around ffmpeg that merges folders, splits chapters by silence detection, and batch-processes a library. Prefer it for large collections, for splitting one long file into chapters, or for keeping the chapter lists inside existing M4B parts.

Troubleshooting

  • Chapters missing in a player. Open the file in VLC or Apple Books, which both list embedded chapters. If they show them, the other player is a music player that ignores chapter atoms; switch apps. If they do not, rebuild from the per-chapter files.
  • The book shows up as music. The media-type flag is missing or the file was added through a music app. Kataribe writes the audiobook media type and the .m4b extension; add the file to the audiobook app, not the music app.
  • Chapters in the wrong order. The chapter list follows the file list, so the files were in the wrong order at build time, usually because a mix of tagged and untagged files forced filename order. Fix the order in the list and rebuild.
  • A chapter starts or ends in the wrong place. Boundaries are file boundaries, computed from each file's probed duration, so a marker can only fall where one file ends and the next begins. If a break sits inside a file, split that file in an audio editor and rebuild.

Do it now

Frequently asked questions

Can I convert MP3 to M4B without losing quality?
Not strictly; MP3 to AAC is a lossy re-encode, but speech at 64 kbps AAC or above is hard to tell from a 128 kbps MP3 source. Never encode below the source bitrate. Only inputs that are already AAC are copied with no loss.
How do I add chapters to an M4B I already have?
A file-per-chapter tool cannot add chapters inside one long file, because it cannot know where they fall. Split the file at the chapter points with an audio editor or a silence-detecting tool such as m4b-tool, then join the pieces. If you still have the original per-chapter files, rebuild from those.
How big is a 10-hour audiobook as M4B?
About 290 MB at 64 kbps, 145 MB at 32 kbps, 430 MB at 96 kbps and 575 MB at 128 kbps. Mono or stereo does not change the total at a fixed bitrate.