Making audiobooks from documents

Anansi Studio can read a document aloud and turn it into an audiobook, with chapters, a cover and timed lyrics, so you can listen to it like any other book and follow the text as it is read.

You can use PDF, EPUB, Word (.docx), Markdown (.md, .markdown) and plain text (.txt) files.

Before you start, you may want to choose and download a voice; see Voices. You can also do it during the steps below.

Starting

Any of these opens the Make audiobook screen:

  • Create audiobook from document… in the add menu (the + button), Create from document… in the sidebar or in the More menu.
  • Picking a document with Add files….
  • Dropping a document on the window (desktop).
  • Opening a document with Anansi Studio from another app.

You can choose several documents at once. Each gets its own Make audiobook screen in turn, headed Make audiobook · 1 of 3. Each one starts from the choices you made on the one before.

Making audiobooks isn't available in the web version. See Platform differences.

The Make audiobook screen

First the document is read (Reading … Finding its chapters and its cover.). Then the screen shows how the book will be made, so you can change it before anything starts. On a wide window the sections are listed down the left; on a narrow window they follow one another down the page.

Information

  • Cover: the document's own cover, if it has one: an EPUB's cover picture, a PDF's first page or a Word document's thumbnail.
  • Title and Author, filled in from the document. Change them if you like.
  • Language: the document's language if it says, otherwise your voice's, otherwise English. Use Search languages to find another. If the voice you chose doesn't speak the language, the screen switches to one that does and tells you why. It may ask you to download that voice in Voice. If no voice speaks the language at all, it warns you that the book would be read with the wrong pronunciation.

Chapters

Chapters found lists the chapters the app found in the document, as a checklist. Untick any you don't want read. A contents page, an index and a references section start unticked.

Warnings appear here when something looks wrong, for example when only one chapter was found, or when a PDF without bookmarks fell back to chapters that follow its pages.

Pages (PDFs only)

What not to read shows each page of the PDF. What the app already leaves out is hatched in grey and labelled: page numbers, running heads at the top and foot of the page, and text that is in the file but can't be seen. A count below the pages sums up everything left out of the document.

To leave out something else, such as a chart's labels, a margin note or a watermark, drag a box over it. It won't be read aloud or put in the lyrics.

  • A box leaves out the whole lines it covers.
  • A box over a line at the top or foot of a page, such as a running head the app missed, also leaves out the lines like it on every other page. They are hatched in the box's orange.
  • A box in the body of a page leaves out only what it covers on that page.
  • Tap a box to select it, then × to remove it.

A moment after each change, the document is read again, leaving out your skip areas, while you carry on. The chapters, the footnote count, the estimates and the suggestions all update to match. The status line says Reading the document again, leaving out your skip areas… while this happens.

Skip areas belong to this one document, so Use these settings for the other N doesn't copy them.

Voice

The voice that will narrate the book: the engine (On-device voice or System voice), the voice model, its speaker, its Speaking speed and Preview voice. These are the same controls as Settings → Voice, and voices that speak the book's language come first. A voice model can be downloaded right here. See Voices.

The voice you make a book with becomes your default for the next book.

Reading

How the text is read:

  • Read:
  • By sentence: one lyrics line per sentence, so the highlight follows exactly.
  • By paragraph: one lyrics line per paragraph, read in one flow. Easier on the ear for fiction.
  • Pause between sentences and Pause between paragraphs, in milliseconds.
  • Footnotes (only when the document has some, with how many):
  • Read, introduced as a footnote: read aloud, introduced as a footnote.
  • Read plainly: read aloud like the rest of the text.
  • Shown, not read: kept in the text and lyrics but not read aloud.
  • Plain punctuation for the voice: curly quotes, dashes and ellipses reach the voice as plain ones, which voices read more cleanly. The text you see keeps them.
  • Read aloud: which kinds of text are read: Headings, Paragraphs, Lists, Quotes, Captions, Tables, Footnotes, Code and the Contents page. Headings, paragraphs, lists and quotes are read unless you turn them off.

The pauses don't apply to the system voice and aren't shown for it.

Performance

How fast the on-device voice works: Performance Level, Voices at once and Sentences per request. These belong to the device, not to the book, so changing them here changes them in Settings → Performance too. See Making books faster. Not shown for the system voice.

Suggested for this text

The app measures the text: how long its sentences and paragraphs are, how much of it is dialogue, its footnotes, its punctuation and its chapters. It suggests settings to suit, each with its reason. For example, it suggests paragraphs for dialogue-heavy text with short paragraphs, longer pauses for dense prose, footnotes shown rather than read when there are many, and on the device side several voices at one core each. Apply takes all the suggestions at once.

How long it will take

  • Listening time: about how long the book will be at the chosen speaking speed.
  • Making it: once this device has made a book with this voice, about how long making this one will take. Until then, the screen explains that this is measured after the first book.

Making the book

At the bottom of the screen:

  • Make audiobook: starts making this book.
  • Use these settings for the other N (with several documents): makes this book and all the rest with the same settings, without asking again.
  • Skip (with several documents): leaves this document out and moves on.
  • Cancel: stops here, making nothing more.

If a document can't be read, the screen says Could not read … and you can only skip it.

The app tells you the book is being made and that it will appear in your library as soon as its first chapter is ready.

The settings you chose are kept with the book, so if making it stops part way, it carries on the same way. Your reading choices and pauses also become the starting point for the next book.

Following the making

Books are made in the background, one after another, while you carry on using the app.

  • The banner at the top of the library shows the book being made, how far it has got and how many more are waiting, with Stop. It reminds you that chapters can be played as they are finished. The Open audiobook button beside the title opens the book once its first chapter is ready.
  • The book's tile has its cover from the start and says how far it has got: Making… 40% · about 12 min left · 3 of 9 chapters, Waiting (in the queue) or Paused.
  • The player shows a small progress ring at the top while the book you are listening to is still being made. Point at it for details. On a wide window, the book's pane also shows Making chapter 3 of 9, Waiting to be made or Paused, with a Resume button when it is paused.

Listening while it is made

With the on-device voice, a book is made part by part: a chapter's opening, or one of its subchapters, at a time. The book appears in your library as soon as its first part is ready. You can play it straight away, and each new part joins the chapter list (and what is playing) as it is finished. Chapters not made yet are listed greyed out, marked Not made yet.

When a book is finished it glows softly in the library until you first open it.

Stopping and resuming

  • Stop in the banner, or Stop making in the book's menu, stops making the book. Everything made so far is kept and can be played.
  • Resume making in the book's menu, or Resume in the wide player, carries on from where it stopped, with the same settings.
  • If the app is closed while a book is being made, making carries on the next time you open the app.

What you get

With the on-device voice

A real audiobook, stored in your library folder: one audio file per part, each with its timed lyrics and the cover. On most devices the parts are .m4a files. Where the device can't make those, such as Linux without the GStreamer AAC encoder, they are .wav files, which are larger.

You can:

With the system voice

The device's own text-to-speech voice reads the book live each time you listen, so no audio file is made. The book opens in its own read-aloud screen. See Books read aloud by the system voice.

Removing a made book

Remove in the book's menu deletes the audio made for it. The document it was made from is never deleted.

Tips

  • Check Chapters found before you start. A wrong chapter list is easier to fix now than after hours of making.
  • For a PDF, look through Pages for headers, footers and labels the app missed, and box them.
  • Making a long book takes a while. Plug in a phone or laptop; see Making books faster.
  • Use Preview voice before committing to a voice.
Editing a book
Voices