How to Make an ElevenLabs Audiobook With Studio

An ElevenLabs audiobook is made in Studio, the long-form editor where a manuscript is imported, split into chapters, assigned narrator and character voices, generated paragraph by paragraph and exported as MP3 or WAV. Narration costs about one credit per character, so a book’s length decides the plan. This guide covers each Studio step, the credit math and the export settings.

Last updated: October 2026. Plans and credit costs checked against ElevenLabs’ official pricing page.

What is ElevenLabs Studio for audiobooks?

ElevenLabs Studio is the long-form audio editor where whole books, podcasts and narration projects are built. Instead of pasting text into a single box, a manuscript becomes a project with chapters and paragraphs, each with its own voice and settings. ElevenLabs markets it as an editor combining its best audio models.

“The best AI audio models in one powerful agentic editor”

Source: ElevenLabs Studio

Studio uses the same voices and credits as standard text to speech, so everything in the ElevenLabs review about voice quality applies here. What Studio adds is structure: chapters, per-paragraph control, a pronunciation dictionary and book-level export.

In short, Studio turns text to speech into a tool to make an audiobook from start to finish.

What you need before starting an audiobook

An audiobook project needs four things in place before any credits are spent: a clean manuscript, the rights to narrate it, a plan with enough credits and a commercial license, and a voice that suits the genre. Fixing problems at this stage is free; fixing them after generation costs credits.

Five preparations save the most time and credits:

• Clean manuscript: Removing headers, page numbers and footnote markers prevents the narrator from reading layout debris aloud in the middle of chapters.
• Narration rights: Authors own their text, but public domain checks or publisher permission are needed for anyone else’s book before narration begins.
• Paid plan for selling: ElevenLabs’ Terms limit free use to non-commercial purposes, and the pricing page lists a commercial license from the Starter plan upward.
• Pronunciation list: Collecting character names, places and invented terms upfront makes it quick to build a project dictionary before the first chapter is generated.
• Numbers written as words: Spelling out figures and dates avoids misreadings, since automatic text normalization can guess wrong on ambiguous numbers in long texts.

Script preparation follows the same principles as any voiceover, covered in the ElevenLabs text to speech settings guide.

Prepare the text and rights first, because both are expensive to fix later.

How do you import a manuscript into Studio?

Importing a manuscript means uploading the book file into a new Studio audio project. ElevenLabs’ docs recommend EPUB, and say Studio splits a book into chapters automatically when each chapter heading is formatted as Heading 1. PDF, DOCX, TXT, HTML and URL imports are also supported.

EPUB works best because it carries clean structure. A DOCX exported from a word processor is a good second choice if chapter titles use a consistent heading style. PDFs often bring line breaks and page artifacts with them, so they need more cleanup inside Studio. Starting from the cleanest available source saves review time on every chapter.

Export the manuscript as EPUB with Heading 1 chapter titles for the cleanest import.

Setting up chapters in Studio

Chapters are the main structural unit of a Studio project, and they control how the book is generated, reviewed and exported. Studio detects them automatically on import when the manuscript contains them. The chapters sidebar then allows adding, renaming, removing and reordering chapters by dragging them into place.

Good chapter structure pays off at export, where each chapter can become its own file. Front matter such as a dedication or an opening credits line often works best as a short separate chapter, and very long chapters are easier to review when split at natural scene breaks.

Check the chapter list before generating anything, because the structure carries through to the final files.

Which voice should narrate the audiobook?

The narrator voice is the single most important choice in an audiobook, because listeners hear it for hours. Studio accepts voices from five sources: Instant and Professional voice clones, the Voice Library, and voices created with Voice Design. Each source suits a different kind of project and budget.

Five voice sources work in Studio:

• Voice Library voices: Ready-made voices from a library of thousands, filterable by accent and style, which suit authors who do not want to use their own voice.
• Instant Voice Clones: Quick clones of the author’s own voice from a short sample, available from the Starter plan, giving a personal sound with minimal setup.
• Professional Voice Clones: Higher-fidelity verified clones trained on far more audio, available from Creator, better suited to hours of continuous listening.
• Voice Design voices: New voices generated from a text description, useful for a narrator with a specific age, accent or tone that the library lacks.
• Model pairing: The same voice can sound different on each model, so testing a narration sample on two models before committing the book is worthwhile.

The ElevenLabs Voice Library guide explains filtering, and the Instant vs Professional cloning comparison helps authors who want to narrate in their own voice.

Listen to a full page in the candidate voice before choosing, not just a single sentence.

Assigning character voices for dialogue

Character voices are additional voices assigned to specific lines, so dialogue sounds like a cast instead of one narrator. In Studio, selecting text and choosing a voice from the sidebar drop-down assigns it. Each voice gets a colored marker, and its text is highlighted in the same color across the project.

A Replace voice across project option swaps every paragraph using one voice for another, useful when a character voice is changed late. Override settings applies stability, similarity or speed changes to a selected passage only. For a character with no suitable library voice, the ElevenLabs Voice Design guide shows how to create one from a prompt.

Keep the cast small, since every extra voice is one more thing to keep consistent across chapters.

How do you fix pronunciation across a whole book?

Book-wide pronunciation is handled by the Pronunciations Editor, which saves alias and phoneme rules to a dictionary connected to the project. Once a rule exists, every chapter uses it, so a character name is pronounced the same way in chapter one and chapter thirty. When several rules match a word, the first one applies.

Alias rules replace a word with a respelling the voice reads correctly, which works on any model. ElevenLabs’ docs say phoneme rules work only with Eleven Flash v2, so alias rules are the practical default for most narration models. Building the dictionary from the pronunciation list before generating avoids paying to regenerate mispronounced passages later.

Build the dictionary first, then generate, so fixes apply everywhere automatically.

Generating, reviewing and regenerating narration

Generation is the step where Studio turns paragraphs into audio and spends credits. Each paragraph is generated on demand, and Studio offers a limited number of free regenerations for a selection as long as the text, voice and settings stay unchanged. Reviewing chapter by chapter keeps mistakes cheap to fix.

Five Studio tools make the review loop efficient:

• Generate: Creates audio for a selected paragraph or chapter and spends credits based on the characters converted, so generating one chapter first limits risk.
• Regenerate: Produces a new take of a weak line, with a few free attempts while the text, voice and settings remain exactly the same.
• Generation history: Stores earlier versions so a previous take can be replayed, downloaded or restored if a newer regeneration sounds worse than the original.
• Lock paragraph: Protects finished narration from accidental changes, which matters on long books where edits elsewhere could otherwise trigger regeneration.
• Override settings: Adjusts voice settings for one passage only, such as lower stability for an emotional scene, without changing the narrator’s baseline elsewhere.

Generate one chapter, review it in full, fix it, lock it, then move on.

How much does an ElevenLabs audiobook cost?

An ElevenLabs audiobook costs roughly one credit per character of manuscript on Multilingual v2, and between half a credit and one credit per character on Flash, per the ElevenLabs pricing page. The credit total therefore tracks the book’s character count, plus whatever is spent on regenerations.

The table uses example manuscript sizes and about 1,000 characters per minute of speech, an approximation based on the pricing page’s estimate of about 10 minutes per 10,000 credits:

Source: ElevenLabs pricing page (elevenlabs.io/pricing), checked October 2026. Manuscript sizes are examples; listening time assumes about 1,000 characters per minute.

A mid-length book fits comfortably inside Pro, with room for regenerations, while Creator suits shorter guides and novellas.

Unused credits on paid plans roll over for up to two months, capped at twice the usual monthly quota, so a smaller plan can bank credits across months before a long book.

Count the manuscript’s characters first; that number is the budget.

Which plan fits an audiobook project?

The right plan for an audiobook depends on book length, how many books are produced, and whether the audiobook will be sold. The pricing page lists a commercial license from Starter, so any audiobook for sale needs at least Starter, and longer books push the practical minimum to Creator or Pro.

Source: ElevenLabs pricing page (elevenlabs.io/pricing) and Terms of Use, checked October 2026.

Starter is enough for testing a sample chapter commercially, but its credits run out long before a full-length book is finished.

Plan-by-plan cost math, including annual billing, is in the ElevenLabs pricing guide, and the license rules are explained in the ElevenLabs commercial use guide.

Creator suits short books, Pro suits most full-length titles.

Exporting the finished audiobook

Exporting is the step that renders chapters or the whole project into downloadable audio. Studio exports MP3 or WAV, either as a single file or as a ZIP with one file per chapter. Any paragraphs not yet generated are converted during export at the normal credit cost, and Studio shows that cost before confirming.

Five export settings matter for a finished audiobook:

• File format: MP3 suits most listening and distribution platforms, while WAV keeps an uncompressed master for further editing or mastering elsewhere.
• Per-chapter files: A ZIP with one file per chapter matches how most audiobook platforms expect uploads and makes later corrections to a single chapter easier.
• Metadata: Title, author, ISBN and description fields are written into the downloaded files, saving manual tagging after export.
• Volume normalization: An optional setting that evens out loudness across chapters, so listeners do not need to adjust the volume between files.
• Publishing exports: Studio offers export flows for InAudio and Spotify, including a preview chapter choice, while voices are labeled as Digital Voice by default.

ElevenLabs’ docs add that a project using an Instant Voice Clone must have that clone verified before publishing exports. Each retailer also sets its own rules for AI narration, so checking the target platform’s requirements before export avoids rejected uploads.

Export per chapter with metadata and normalization switched on.

From manuscript to audiobook: the full workflow

The full ElevenLabs audiobook workflow is a seven-step sequence that moves from a clean manuscript to exported chapter files. Following it in order keeps credit spending predictable, because text, structure, voices and pronunciation are settled before generation starts. It mirrors the Studio steps covered above.

1. Prepare the manuscript: Clean the text, write numbers and symbols as words, and format chapter headings as Heading 1 in an EPUB file.
2. Create an audio project: In Studio, start a new audio project and upload the manuscript so chapters are detected automatically.
3. Check the chapters: Review the chapters sidebar, then add, rename, remove or reorder chapters where detection missed something.
4. Choose and assign voices: Pick a narrator voice for the whole project, then assign different voices to character dialogue where needed.
5. Build a pronunciation dictionary: Add alias or phoneme rules for names and terms so every chapter pronounces them the same way.
6. Generate and review chapter by chapter: Generate one chapter, listen in full, regenerate weak lines, then lock finished paragraphs.
7. Export the audiobook: Export chapters or the full project as MP3 or WAV, add title and author metadata, and enable volume normalization.

Authors who also want a companion podcast can reuse the same voices, following the guide on making a podcast with AI tools.

Settle text, voices and pronunciation first, then generate one chapter at a time.

Bottom line on making an ElevenLabs audiobook

An ElevenLabs audiobook is realistic for independent authors: Studio handles chapters, multiple voices, a pronunciation dictionary and per-chapter export in one editor. Cost scales with manuscript length at about one credit per character, so Creator suits short books and Pro suits most full-length titles. Preparation decides how much regeneration is needed.

“generating audiobooks and assisting individuals who have lost their voices due to medical conditions”

Source: Wikipedia, Audio deepfake

An AI audiobook is one of the most established uses of synthetic speech, and Studio is built around audiobook narration. For authors who prefer human review, ElevenLabs also lists a Productions service offering human-edited audiobooks on the platform.

“Studio makes a solo audiobook practical, as long as the manuscript is clean before generation.”

Source: GrowthStackKit, ElevenLabs Review

Prepare carefully, narrate chapter by chapter, and size the plan to the manuscript’s character count.

Frequently asked questions

The 12 most-asked questions about making an ElevenLabs audiobook.

Can I make an audiobook with ElevenLabs for free?

A short test is possible on the Free plan, which includes 10,000 credits per month and 3 Studio projects. ElevenLabs’ Terms limit free use to non-commercial purposes, so an audiobook for sale needs a paid plan. The pricing page lists a commercial license from Starter.

How many credits does an audiobook need?

Roughly one credit per character on Multilingual v2, and between half a credit and one credit per character on Flash, according to the pricing page. A 300,000-character manuscript therefore needs about 300,000 credits, plus extra for regenerating weak lines.

Which file format should I import into Studio?

ElevenLabs’ docs recommend EPUB. When each chapter heading is formatted as Heading 1, Studio splits the book into chapters automatically. PDF, DOCX, TXT, HTML and URL imports also work, though PDFs often need more cleanup.

Can I use different voices for different characters?

Yes. In Studio, selecting text and choosing a voice from the sidebar assigns it to that passage. Each voice is color-coded, and a Replace voice across project option swaps one voice for another everywhere it appears in the book.

Can I narrate an audiobook in my own voice?

Yes, with a voice clone. Instant Voice Cloning starts on the Starter plan and needs about 1 to 2 minutes of audio. Professional Voice Cloning starts on Creator, needs at least 30 minutes of audio and includes verification, and suits long listening better.

How do I fix mispronounced names in Studio?

Studio’s Pronunciations Editor saves alias and phoneme rules to a dictionary connected to the project, so a fix applies in every chapter. ElevenLabs’ docs say phoneme rules work only with Eleven Flash v2, so alias respellings are the practical default for most models.

What audio formats does Studio export?

Studio exports MP3 or WAV, either as one file for the whole project or as a ZIP with one file per chapter. Export settings also add title, author, ISBN and description metadata to the files and include an optional volume normalization setting.

Which ElevenLabs plan is best for audiobooks?

It depends on length. Creator’s 121,000 monthly credits suit short books of around 100,000 characters, while Pro’s 600,000 credits cover a 300,000-character novel with room for regenerations. Starter’s 30,000 credits suit only sample chapters.

Does regenerating a paragraph cost credits?

Studio allows a limited number of free regenerations for a selection as long as the text, voice and voice settings stay unchanged. Changing any of those and generating again spends credits, which is why settling voices and pronunciation first saves money.

Can I sell an audiobook made with ElevenLabs?

ElevenLabs’ Terms state that users retain all rights in their output, and the pricing page lists a commercial license from the Starter plan. Each retailer sets its own rules for AI-narrated audiobooks, so the target platform’s requirements should be checked before publishing.

Can Studio publish directly to audiobook platforms?

Studio includes export flows for InAudio and Spotify, with a choice of preview chapter and narrator credits, and voices are labeled as Digital Voice by default. Other retailers need a standard MP3 or WAV export uploaded through their own systems.

How long does an audiobook take to produce in Studio?

ElevenLabs does not publish generation times for Studio projects. In practice, the time goes into manuscript cleanup, voice selection, building the pronunciation dictionary and listening back to each chapter, which takes far longer than the generation itself.