How to Rip an Audiobook CD Set Into a Single M4B

The box is on the desk. Fourteen discs in cardboard sleeves, bought for a car that no longer exists, and the plan is one file on your phone by the end of the evening.

The ripping is not the hard part. Feeding discs is tedious, but a Mac does it without argument. The other half of the evening — one chaptered file out of fourteen discs — is what AudioBo is built for; the page below is the whole route, discs to phone.

The hard part is that fourteen discs arrive as roughly two hundred small files, all named Track 01, in fourteen folders, and something has to turn that into a book that plays in the right order and remembers where you stopped.

The rest of this page is how that turn gets made: what to rip to, why the seams between discs cost more than the format does, and which decisions disappear the moment the box is closed.

Apple Music rips the discs and then leaves you there

Start with the tool that is already installed, because it does more than people assume — and because where it stops is not where people assume either.

Music reads discs perfectly well, and it even knows what an audiobook is. What it cannot do is put the result where your audiobooks live. That is a second app with a separate library, and it has no idea what a CD is. So the import ends with a shelf of tracks in one program and a bookshelf in another, and nothing walks between them.

Music on macOS 26 still imports CDs, and Apple still documents it in the current Music user guide: insert the disc, click Import CD near the top right, and change the format under Music → Settings → Files → Import Settings. The default encoder is AAC. The app also carries an Apple Lossless encoder, an AIFF encoder, and a WAV encoder, plus a Join CD Tracks command for gluing selected tracks together as they come in.

So as a CD ripper, Music is serviceable.

As a route to an audiobook, it stops one step short of anywhere useful — and not for the reason usually given. Music does understand audiobooks. There is a real Audiobook media kind on macOS 26.6. Check it on your own Mac: select a track, open Get Info, and set Media Kind to Audiobook — the window retitles itself “Audiobook Info” the moment the kind changes. Build a smart playlist and one of the criteria you can pick is Media Kind is Audiobook. Nothing in the import sets any of that for you; set it yourself and Music knows what your tracks are.

You will then have exactly what you had a minute ago, filed differently. Fourteen discs of separate tracks, tagged as audiobooks, sitting in Music.

What you wanted is one file with chapters, and it wanted to live in Books, because that is the app that remembers where you stopped. Books keeps its own library in its own sandboxed container, and tagging something in Music puts nothing into it. Books also shows no sign of CD code — I found nothing in the application bundle containing the words “Import CD”, and no audio-disc strings in its binary.

That is the whole shape of the problem. The app that can read the disc hands you a shelf of tracks; the app that holds books cannot read the disc, and does not see the shelf.

If you have two discs and all you want is the audio on your phone as tracks, import them in Music and stop here. You do not need a converter for that, and everything below is about the case where two discs are fourteen.

What the rip should hand you, before anything else sees it

This is the part of the evening that quietly decides how the rest of it goes, and none of it is technical yet. While each disc is in the drive you settle three things: how much quality you keep, whether the disc comes out as one file or as its separate tracks, and whether you write down which disc this is. The last one is the one people skip, and it is the one nothing later can work out for you — fourteen folders of identically named tracks look exactly alike once the box is closed.

Three decisions get made at rip time, and only one of them is easy to change later.

Rip to a lossless format. WAV, AIFF, FLAC, or Apple Lossless — the choice among those barely matters. What matters is that the audio is not yet compressed, because the finished book will be compressed exactly once, at the end, in a single pass over a single timeline. Ripping straight to MP3 spends a lossy generation on files you are about to dismantle and reassemble, and a second generation always subtracts something.

Budget the disk space. CD audio is 16-bit, 44.1 kHz, stereo — ffprobe reports 1,411 kbps on a CD-spec file, which works out to 176,400 bytes a second. A 70-minute disc is about 740 MB. Fourteen of them are about 10 GB, sitting on your drive until the book is built. It is temporary, but it is real, and a full disk halfway through a rip is a bad evening. The build step at the end no longer doubles that bill — AudioBo cleans its temporary files as the build runs — but the rips themselves still need their 10 GB.

Choose per-track files or one file per disc with a cue sheet. Both work. One file per disc plus a .cue is tidier — fourteen big files instead of two hundred small ones, with the track boundaries preserved as text rather than as file breaks.

Set the disc number while the disc is still in the drive. This is the decision people skip, and it is the one that costs the most. Restarting track numbers and unpadded disc numbers are both sorting problems with worked answers elsewhere — audiobook chapters in the wrong order has the ffprobe dump and the renaming loop. The rip-time half is the part no later tool can recover for you. The disc number is a real tag field and it survives both containers: I wrote disc=2/14 into an MP3 and into an MP4, and ffprobe read it back from both. Write it while the disc is in your hand and you know which one it is. Reconstruct it afterwards, from fourteen folders of files all called Track 01, and you are guessing.

For the ripping itself, XLD is the long-standing macOS answer — a free download that asks only for an optional donation — and its own site is specific about what it does: it has been ripping CDs for well over a decade, checks tracks against the AccurateRip database, splits by cue sheet, and outputs WAVE, AIFF, Apple Lossless, FLAC, AAC, and MP3 among others. That covers every decision above without anything else installed.

The joins are where a box set actually breaks

Here is the part nobody warns you about, and it is the reason the format decision above matters more than it looks.

In ordinary words: glue two hundred already-compressed tracks end to end and you do not get the book back. You get the book plus a sliver of extra material at every seam, put there by the encoder to round each file off. One of them is far too small to notice. Two hundred of them, spread across fourteen discs, land in the middle of sentences — at exactly the points where the publisher happened to end a disc.

Compressed audio does not end where the audio ends. Encoders work in fixed-size frames, and the last frame of a file gets padded out — an AAC track’s tail is rounded up to a whole 1,024-sample frame whether the audio wanted that or not. Join two hundred separately-encoded tracks and you join two hundred padded tails.

Want to see the size of it? I measured it. One continuous 60-second tone, cut into twelve tracks of exactly five seconds — 220,500 samples each, no gaps, no rounding — then each track encoded on its own at 64 kbps and the twelve put back together two ways: glued as-is, and decoded first and rebuilt as a single stream. Eleven joins in each case.

Encoded tracks	Glued as-is	Decoded and rebuilt
ffmpeg’s AAC encoder	42 ms per join	17 ms per join
The macOS system AAC encoder	93 ms per join	41 ms per join
LAME MP3, gapless header intact	42 ms per join	0 ms per join

Read the columns before the rows. Gluing already-compressed tracks keeps every seam the encoder left. Decoding first lets a decoder that knows what was padded throw it away — and LAME’s knows exactly, because LAME wrote the delay and padding into the file. Twelve tracks in, 2,646,000 samples out, exactly what went in.

That last row is a fact about one encoder, not about MP3. I stripped the gapless header off the same twelve tracks and rebuilt them again: half a second longer than the source, about 45 ms a join, worse than either AAC rebuild.

None of these columns is the finished book. The M4B at the end is AAC in every case, and it is fine, because the encode that matters is the one that runs over the whole timeline at once, with no seams to pad.

Scale it to a bigger set — twenty-two discs at sixteen tracks apiece is 352 files and 351 joins. At the non-zero rates in the table, that is somewhere between six and thirty-odd seconds of inserted material, distributed through the book as small hiccups in the middle of sentences, at exactly the points where the publisher happened to break a disc.

The structural fix is not a setting. It is handing the joiner audio that has not been encoded yet, so that there is one encode over one timeline and no seams to pad.

And if your discs were ripped to MP3 in 2009 and the CDs are long gone, that shelf is probably not a loss. If the files carry the gapless header — it sits in the first frame, next to the encoder’s version string, so it is checkable — the rebuild measured above cost nothing. You pay a single generation of re-encoding on the way to M4B and you get a clean timeline out of it.

CD tracks are not chapters

A track boundary is a decision the publisher made about a disc, not about the book. Some sets align tracks to chapters, and those rip beautifully. Plenty split every few minutes regardless of what is being said, and a chapter list built from those boundaries is two hundred entries long and named after none of them.

Three honest ways out. Accept the track boundaries and rename the entries into something readable. Use the TITLE fields from a cue sheet if your ripper wrote them. Or bring chapter data in from a sidecar timestamp file and let the track boundaries go.

Whichever you choose, decide it before you export rather than after.

What AudioBo does with fourteen discs’ worth of files

Everything above is arrangeable by hand. The unpleasantness is not any single step — it is doing all of them across two hundred files without losing the thread.

That is the seam AudioBo sits in, and a multi-disc rip is close to the worst case it was built for.

In plain terms, here is what a fourteen-disc evening looks like in it. You drop in one disc’s folder at a time, in the order the discs are numbered, and each folder lands at the end of the list and stays where you put it — adding disc 8 does not rearrange the seven already sitting there. If you re-rip disc 7 after a bad read and drop the whole folder back in, the copies you already have are not added a second time. To check the order, you hold a key and every chapter title turns into the name of the file it came from, so the real running order can be read straight down the list without opening Finder. The titles themselves get renumbered in one pass rather than two hundred small edits. The book’s own details — title, author, cover — can be looked up instead of typed. And building a set this size takes a fraction of the spare disk space it used to; if the disk does fill up anyway, it says so plainly.

If you do want the mechanical version: import behaves like discs exist. A Settings toggle (since 1.3.0) turns off the automatic re-sort that fires every time you add files, so each folder you drop in is appended at the end as its own group and stays where you put it — discs go in one at a time, in order, and the order holds. The Import tab sets the sort key outright: file name, track number, title, or date modified. Collections of 200+ files import without the wait, and duplicate files are skipped automatically — which matters on the evening you re-rip disc 7 after a bad read and drop the whole folder back in.

Verifying order across hundreds of entries is the other half. Holding Option swaps every chapter title for its source filename, so you can read the actual file order down the list without opening Finder once.

Then the names. Two hundred entries reading Track 01 through Track 16, fourteen times over, need renumbering, not editing: Batch Rename does cleanup and renumbering in one pass — stripping Roman numerals, removing brackets, numbering that can start at zero — and its Regex tab shows a live before-and-after preview. Book metadata comes from Fetch Metadata, which searches Audible, Apple Books, MusicBrainz, and Google Books, so the album name your ripper guessed at is a starting point rather than a life sentence.

One practical note for a set this size: builds got cheaper in 1.3.5, which shrank the temporary files a build creates and cleans them up as it goes, so a build needs roughly 80% less free space than before and says so plainly if the disk fills anyway. On a ten-gigabyte pile of lossless rips, that is not a footnote.

The output is one M4B: single timeline, chapters, cover, tags, exported once.

What it does not do is read the discs. Feeding the drive is still your evening, one disc at a time, and everything above starts at the moment the files exist on your disk.

The part that needs saying plainly

A standard audio CD has no encryption on it. Ripping a disc you own is a read, and there is nothing to circumvent — which is exactly why this is the one audiobook workflow that is unambiguously local and unambiguously yours. AudioBo does not remove DRM and never will; that restriction belongs to store downloads and streaming apps, and no converter here changes it.

Once the file exists, the rest is ordinary. Where multi-disc books should live on disk afterwards is audiobook file naming and folder structure; if your source turns out to be video files rather than discs, turning video into an audiobook is the same idea with a different intake; and the general path from loose files to a finished book is converting MP3 to M4B on a Mac.

The discs go back in the box. You will not need them again.