How to create an audiobook (with the AI agent or manually) 📖

Audiobook (formerly Stories) turns a written manuscript into a narrated audiobook with an AI voice. You can then add other languages and export the audio, including a mastered version ready for distribution.

Click Audiobook in the sidebar under Projects. There are two ways to work:

You can switch between the two at any time with Switch to Manual / Switch to Agent.

1. Add your manuscript to the box. Supported formats: .pdf, .png, .jpg, .docx, .txt and .zip (images and scanned pages are read automatically).

2. In the message box, describe what you want, for example "Turn this manuscript into an audiobook with a warm female narrator".

3. Optionally click Choose Narrator to pick the voice, and pick the speech model from the menu next to the send button (MARS-X Instruct (Beta) is selected by default).

4. Click the send arrow. Each request becomes a conversation listed under the box, so you can come back to it, search it or file it in a folder.

1. Click Switch to Manual. Upload your manuscript under File (.txt, .docx or .pdf, no character limit) or paste a Link.

2. Select the Source Language (the language the book is written in).

3. Choose the Narrator Voice: a ready-made voice from the library, one of your own custom voices, or a voice shared in your workspace. See What is the Voice Library?

4. On the right, under Settings, name the project and check the Speech Model (MARS8.1-Beta by default). Under Project Setup you can pick a folder, add a description and apply keywords for names and terms.

5. Click Begin Creating Story.

Turn it on if you plan to publish the audiobook with a retailer. CAMB.AI collects the book's metadata and runs publishing quality checks so the audiobook can be packaged to retail spec. You'll fill in:

The editor splits your text into lines, one sentence per line. At first, every line is muted (no audio yet).

1. Read through the script. Check each line is split sensibly and that punctuation stays on the same line, to avoid odd pauses.

2. Generate the audio for all lines with Re-generate all stale/muted dialogues on the floating toolbar. A notification appears when it's done.

3. Press play to listen. Use the arrows on the seek bar to jump to the next or previous line; the counter (for example 51/112) shows the current line and the total.

4. Edit any line and re-generate it. Edited lines become stale until you do.

You can also split, merge, add or delete lines, change a line's speed (anything other than 1x is shown on the line), select several lines at once to re-generate them or change their speaker, and add gaps between lines, for example around chapter titles. The default gap between lines is 0.4 seconds. See The dialogue toolbar: every button explained.

If you change the narrator's voice in the speaker settings, re-generate the lines to hear the new voice.

Click + next to the language tabs, choose the language and pick Generate Translation (text only, so you can review it before generating audio) or Generate Translation And Dub (translation and narration in one go).

1. Click Generate Audio Tracks, then Export.

2. When processing is done, click Download. You can download the narration audio, dialogue-only audio (for books with several speakers) and the text as a .txt file.

For a polished, platform-ready file, see How to download mastered output.