Blog
Distribution11 minutes read

ElevenLabs Audiobook Publishing: How It Works, What It Costs, and What It Leaves to You

Published
AK

Alek Karp

Co-founder

Share this article


Mailbox on a dark ground with a brass flag raised and a waveform envelope in the slot, with the ElevenLabs wordmark
Summary

ElevenLabs audiobook publishing compared: free ElevenReader vs Studio files, Spotify's digital voice rules, and the work left to you.

Introduction

In February 2025, Spotify started accepting audiobooks made with ElevenLabs. That made ElevenLabs audiobook publishing a real DIY route for indie authors with a finished manuscript.

This guide covers the two publishing paths and the production steps inside the editor. Then it walks the Spotify route and prices the work, including the jobs the tool hands back to you.

Full disclosure: Lex is an audiobook studio that uses professional engines, ElevenLabs among them. The engine gives you a voice, and the audiobook is the production around it.

What Is ElevenLabs Audiobook Publishing?

ElevenLabs audiobook publishing has two routes. Per ElevenLabs' documentation, ElevenReader Publishing is free: readers stream your book on demand in the ElevenReader app, and you keep your rights. Its payment terms (January 2026) pay 60% of net revenue on direct sales, plus $0.20 per hour streamed by premium listeners, capped at $2.50 per listener per title. The paid route is ElevenLabs Studio's Audiobooks workflow: upload, cast, edit, then export MP3 or WAV files for Spotify and other platforms.

The free route has a catch. Per Lex's AI audiobook route comparison, you cannot pick the voice in advance or edit the read on ElevenReader Publishing. There is also no audio to export, so the book lives only inside the app.

Free is a price, paid in files. Studio gives you files to sell wherever AI narration is accepted.

ElevenReader PublishingElevenLabs Studio (Audiobooks)
CostZero costMonthly credit plan
Voice controlNo voice choice in advance, no editingYou cast, preview, and regenerate every voice
ExportNo audio exportMP3 or WAV, full book or by chapter
Where it sellsStreamed on demand in the ElevenReader appElevenReader, partner platforms such as Spotify, and stores you upload to

Both routes are measured in finished hours. A finished hour is edited, ready-to-publish audio, roughly 9,300 words, so a 75,000-word novel is about 8 finished hours.

The market makes those hours worth producing. The Audio Publishers Association's APA 2026 audiobook sales survey reports US audiobook sales revenue "reaching $2.43 billion in 2025, 9% over the previous year."

How to Produce and Publish an Audiobook with ElevenLabs, Step by Step

The ElevenLabs workflow runs in five production steps, then export or publish. Lex's notes follow each stage.

Upload Your Manuscript and Pick Base Settings

Per ElevenLabs' documentation, you upload an EPUB or PDF, then choose single cast (one narrator) or multi cast (a narrator plus distinct character voices).

Automatic parsing extracts the cover and detects copyright notices, dedications, prologues, chapters, and epilogues. Checkboxes let you include or exclude each section. The docs warn parsing may not always get sections right, so review the structure first.

Cast Characters and Choose Voices

Character Casting detects speakers and proposes a voice for each, previewed on real dialogue. ElevenLabs' documentation caps it at 150 detected voices per book and calls detection a starting point that can miss minor or unnamed characters.

The library holds over 10,000 voices across 90+ languages, filtered by language, accent, gender, and age. In a 10-hour book, the hard part is drift between takes.

Lex's cold-start probe renders one sentence three times and measures the pitch gap. Under 3% is locked, 3–5% is workable, and over 10% means audible seams. The worst voice we probed drifted 16%, and the best held at 1.3%.

A multi-cast book needs the probe on every role, so test AI voice stability before you commit a cast.

Set Pronunciation Rules

ElevenLabs auto-suggests terms that may need a rule. You write alias rules or phoneme rules, and the documentation says phoneme rules work only on English models.

Preview each rule in context, because the first matching rule wins. Changes apply only to newly generated audio, so every fix means a regeneration that costs credits.

Lex reviews pronunciation on every book from a short glossary with phonetic hints (Kehlani, keh-LAH-nee). No IPA needed.

Choose Original Audio or Dynamic Narration

ElevenLabs' documentation describes two narration modes. Original audio is generated in advance and fixed for every listener, with full control over timing and music. Dynamic narration lets each listener pick the voice at playback, and music, sound effects, and external audio are left out.

This choice decides what you own. Dynamic narration lives inside the player, with no file to sell elsewhere, so choose original audio for Spotify or a wide release.

Edit, Score, and Review the Read

The editor generates audio paragraph by paragraph. Replaying generated audio uses no credits, but any change to text, voice, or model means regenerating, which does. A timeline layers music and sound effects under the narration, picked from ElevenLabs' libraries or generated.

Treat the score as seasoning. In Lex's rules for scoring an AI audiobook, music plays for well under half the runtime and sits roughly 20 dB under the narration. Each cue is anchored to a word, so the score follows the scene.

ElevenLabs' documentation says "editorial and quality checks are essential for a polished result." The tool supplies the timeline, and the listening is yours.

Export Files That Pass Store Review

ElevenLabs exports the full project or single chapters as MP3 or WAV, as one file or a chapter ZIP, plus subtitles and an AAF timeline. Ungenerated sections are completed automatically on export, using credits.

Those exports are raw, and store specs are your job. For direct Spotify for Authors uploads, the reference is Spotify's November 2024 metadata and assets guide, which Spotify's upload help page still links as its full guidelines. Spotify's guide specifies that "RMS must be between -24dB to -14dB." It also asks for:

  • MP3 files at 192 kbps or higher, 44.1 kHz and 16-bit
  • A noise floor under -60 dB
  • Head silence of 0.5–1 second and tail silence of 1–5 seconds
  • A retail sample of 5 minutes or less
  • Files no longer than 2 hours, one chapter per file
  • An audio ISBN separate from the print and ebook editions

Stores check this at the end.

Where Can You Publish an ElevenLabs Audiobook?

An exported ElevenLabs audiobook has three destinations: ElevenReader, Spotify (direct or through Findaway Voices, now Voices by INaudio), and other stores you upload to yourself. Each has its own review and disclosure rules.

ElevenReader and Partner Platforms

ElevenLabs' documentation describes a Publish tab with an ElevenReader submission flow. You build an author profile, add metadata and distribution details, accept payouts and agreements, then wait for review. The same tab lists partner platforms such as Spotify and InAudio.

Plan to disclose AI narration in each store's own wording. Lex's distribution table shows Google uses "Synthesized voice," Kobo uses "Synthesised voice," and Apple marks "AI Generated" by role.

ACX self-serve still expects human narration and does not accept externally produced AI narration. Audible on the Publisher pack goes under Lex's publisher agreement: a direct Lex upload, not a separate ACX project.

Apple reviews third-party AI audio title by title and may decline. Check the stores that accept AI narration before you upload anywhere.

Submitting to Spotify via Findaway Voices

In its February 2025 ElevenLabs announcement, Spotify said "authors simply download the file package and upload to Findaway Voices by Spotify." The announcement described a standard review, then release on Spotify and other retailers that accept digital-voice titles from Findaway Voices. For this route, it said Spotify prepends the description with "This audiobook is narrated by a digital voice."

Spotify's INaudio help page says the distributor has since been renamed: "Voices by INaudio was previously called Findaway Voices by Spotify."

You can also upload direct through Spotify for Authors. Under the Spotify digital voice narration policy, you select the digital voice narration option, and Spotify says it adds "a short sentence" to your description noting digital narration.

Direct upload carries no partner fee. Under Spotify's distribution terms, Spotify pays a royalty of 50% of the suggested retail price you set on à la carte sales. Spotify's terms pay Premium listening from a shared pool, by share of hours listened.

How Much Does It Cost to Publish an Audiobook with ElevenLabs?

ElevenReader Publishing costs nothing, and you get no files. ElevenLabs Studio runs on monthly credit plans, and credits go to custom-audio previews, generation, regeneration, and export of ungenerated sections.

The bigger cost is your hours. A 10-hour book is about 93,000 words generated paragraph by paragraph: hundreds of generations to hear and redo. Here is the math against a human narrator and Lex production pricing.

OptionPriceExtra hoursWho does the production
Human narrator$200–$400 per finished hour plus $75–$150 editing ($2,400–$5,850 for a 75,000-word novel)Per finished hourNarrator and editor
ElevenLabs StudioMonthly credit planMore creditsYou
Lex Voiceover$49 one-time, 10 finished hours$5+/hrLex producers
Lex Studio$99 one-time, 10 finished hours, fully scored$10+/hrLex producers and audio engineers
Lex PublisherCustom quote, 10 finished hours$20+/hrLex, plus done-for-you distribution

Human narrator figures come from Lex's audiobook production cost breakdown.

An 8-hour novel fits inside the 10 hours every Lex pack includes. Lex takes no royalties, and on Voiceover and Studio you keep 100% of every sale. On Publisher's wide stores, the distribution partner keeps 20% of net, except on Audible and Spotify.

Will Listeners Finish an ElevenLabs Audiobook?

The Audio Publishers Association's 2026 surveys found willingness to try AI-narrated audiobooks fell from 70% in 2025 to 61% in 2026 among audiobook listeners. Only 16% have heard an AI-voiced audiobook at all.

A finished AI audiobook has to win over that hesitant audience. This producer's checklist covers the work:

  • Review the parsed structure before generating anything
  • Check every cast assignment, including minor and unnamed speakers
  • Probe each voice for drift before you commit it to a role
  • Preview every pronunciation rule in context
  • Keep the score under the voice
  • Run store QC against each retailer's specs before upload
  • Disclose AI narration plainly, in each store's wording

Everything on that list is production. The ElevenLabs audiobook alternative hands it off: Lex casts, directs, scores, and masters, then delivers files that pass retailer QC.

Frequently Asked Questions

How Much Does It Cost to Make an Audiobook with ElevenLabs?

ElevenReader Publishing is free to stream, but it gives you no audio export. Editing and exporting files in ElevenLabs Studio requires a monthly credit plan.

What File Formats Does ElevenLabs Accept for Manuscripts?

ElevenLabs' documentation lists EPUB and PDF.

What Is the Difference Between Original Audio and Dynamic Narration?

Original audio is fixed, so every listener hears the same voice and music. Dynamic narration lets each listener choose the voice at playback, without music or sound effects.

Can I Publish an ElevenLabs Audiobook on Multiple Platforms?

Yes, exported files can go to ElevenReader, Spotify, and other stores that accept AI narration with disclosure. ACX self-serve does not accept externally produced AI narration.

How Will Spotify Label an AI-Narrated Audiobook?

For titles sent through Findaway Voices (now Voices by INaudio), Spotify's February 2025 announcement said the description is prepended with "This audiobook is narrated by a digital voice." For direct upload, you select the digital voice narration option, and Spotify says it adds "a short sentence" to the description noting digital narration.

How Many Voices Can One ElevenLabs Audiobook Use?

Character Casting detects up to 150 voices per book. It draws on a library of over 10,000 voices in 90+ languages.

Are Cloned Voices Private?

At Lex, a cloned voice is a private model used only for its book, not shared, sold, or reused for other clients. For clones made on ElevenLabs, read ElevenLabs' own terms.

Summary

ElevenLabs audiobook publishing gives you two doors. ElevenReader Publishing is free streaming with no files. ElevenLabs Studio gives you files, and with them the producer's job.

Spotify accepts ElevenLabs titles through Voices by INaudio or direct upload, with a digital-voice disclosure either way. Budget your hours as well as your credits.

Next, decide whether you want to be the producer. If you do, work the checklist above.

If you don't, compare done-for-you audiobook distribution. You get mastered files with a commercial license included, and Lex takes no royalties.

Hear your first chapter free

Send the manuscript. We cast, direct, and score the opening and send it back in about a day. No card, yours to keep.

More to read