What is VoxAnvil, the AI audiobook generator for authors?

A direct answer

VoxAnvil is an AI audiobook generator for authors. The product lives at voxanvil.com. You bring a manuscript, prepare how difficult phrases should be spoken, choose a narrator, and generate audio chapter by chapter. ElevenLabs performs the speech synthesis. VoxAnvil is the author project around that request: file import, chapter organization, a pronunciation dictionary, a narration review, and a distribution checklist.

That distinction matters when you are comparing tools. A voice engine reads the text you send. An author studio is where the book is split, where a name is given a spoken form, and where you keep the files you will later upload somewhere else. VoxAnvil is the second of those. It does not replace your judgment about whether a chapter sounds right, and it does not accept or reject a file for Audible, Findaway Voices, or any other retailer.

The studio is aimed at independent authors who already have a manuscript and want a narration workflow in one project. A free plan exists so you can try a short amount of chapter audio. Starter and Pro raise the monthly chapter-credit allowance and include the pronunciation dictionary. Pro adds the narration review and voice cloning. Support mail for the product is support@voxanvil.com.

Who the studio is built for

The fit is a writer who has finished, or nearly finished, a book and needs an audio edition without staffing a separate recording session for every chapter. The import path accepts PDF, DOCX, EPUB, and plain text. After the text is extracted, VoxAnvil looks for headings such as Chapter, Part, or Section. When those headings are present, they become the chapter list. When they are not, the text is grouped into chunks of about 2,500 words, which is also how the pricing page defines a chapter credit.

The same project can hold a novel, a novella, or a shorter work you are using as a test. Free accounts are limited to one active project. Starter accounts can keep five. Pro accounts are listed with unlimited projects. If you are producing a series, the project limit is the practical reason to look at Pro, not a claim about audio quality.

Podcasters have a related path. Starter and Pro include the Author Podcast Studio, an RSS feed, and a website embed. Episode formats in that studio include conversation, solo, and interview, among others. Those episodes are a separate workflow from a book’s chapter list. This page is about the audiobook project. The podcast tools are there if you also want a spoken episode built from a topic or a source you paste in.

What you can do inside a project

A project moves through manuscript, review, voice, generation, and listening. You can paste text or upload a file. The upload must contain readable text. A scan or a photo of a page with no extractable words is rejected, because the studio narrates text, not an image of a page.

Before generation, two preparation tools are the heart of what the product calls the Manuscript Intelligence Engine. The preflight scan reviews manuscript text and returns flags for numbers, abbreviations, contractions, names, punctuation, currency, dates, formatting, and other spots that a speech engine is likely to misread. Each flag includes the original phrase and a suggested spoken replacement. Pronunciation Forge, available on Starter and Pro, stores the reading you actually want. When a chapter is generated, saved rules are substituted into the text before it is sent to ElevenLabs. Longer phrases are applied first so a short rule does not cut the middle out of a longer name.

On the Pro plan, Narration QA adds a readiness report. It checks the manuscript for unclosed bracket tags, tags the studio does not recognize, repeated lines, very long sentences, dense ellipses, acronyms that have no pronunciation rule, and all-caps lines that may be over-emphasized. It can also ask for a short editorial pass on pronunciation, pacing, repeated material, speaker ambiguity, and production formatting. The report returns a readiness score from 0 to 100. Critical issues lower that score more than warnings or notes. The score is a review aid. It is not a certificate that a retailer will accept the audio.

• Import PDF, DOCX, EPUB, or TXT, or paste the manuscript.
• Split on chapter-style headings, or into roughly 2,500-word chunks.
• Save spoken forms for words, names, acronyms, and phrases.
• Generate chapter audio with ElevenLabs, then listen in the project.
• Use the distribution checklist when you prepare files for ACX or Findaway.

Credits and the prices published in the product

The pricing page defines one chapter credit as one chapter, typically up to 2,500 words. An 80,000-word novel is shown there as about 32 credits. Plans are billed as a monthly allowance of those credits. Annual billing is offered on the pricing page at a lower monthly equivalent than the monthly list price. This page quotes the monthly figures so it does not invent an annual total.

Free is $0. It includes 2 chapter credits a month, 1 active project, 10 curated narrator voices, MP3 download, and PDF, EPUB, or DOCX import. Starter is $14.99 a month. It includes 50 chapter credits a month, 5 active projects, the narrator voices, manuscript preflight for number and abbreviation fixes, the Author Podcast Studio, an RSS feed, a website embed, and audiogram poster export. Pro is $39.99 a month. It includes 200 chapter credits a month, unlimited projects, voice cloning, character casting, the full preflight suite, and the distribution workspace.

Credit packs are one-time additions, not a second subscription. The Spark Pack is 25 credits for $4.99. The Forge Pack is 100 credits for $14.99. The Anvil Pack is 500 credits for $59.99. Those are the pack prices in the product. VoxAnvil does not publish a per-minute studio rate alongside them. If a checkout amount ever differs from this article, the pricing page is the source to trust.

Voices, cloning, and the separate video step

Narration uses ElevenLabs. You pick a voice for the project, and the prepared chapter text is sent to that voice. Emotion and expression tags you insert, such as emphasis or whispering, stay in the text because the synthesis step is expected to interpret them. Pause tags are converted into pauses before the request. Sound-effect tags are removed and are not read aloud. A chapter longer than about 4,500 characters is split into smaller requests and the resulting audio is joined, because a single synthesis request is kept under the size the integration allows.

Voice cloning is a Pro feature. You provide an audio recording and attest that you are authorized to use it. The clone is created as an ElevenLabs voice and can then be selected as a narrator. Cloning someone without the right to use their voice is outside the product’s consent step. The studio also has a separate talking-avatar step rendered with HeyGen. That video step does not create the audiobook voice, and it is not a substitute for the chapter narration described here. Narration stays on ElevenLabs.

Pro also lists character casting. The studio can read a manuscript excerpt, propose speaking characters, and store a cast with a voice assignment. A single narrator voice remains the default path. Multi-voice casting is an extra production choice, not a requirement for generating a chapter.

What VoxAnvil does not decide for you

You still listen. A clean pronunciation rule can stop a name from being spelled aloud, and it cannot tell you that the performance suits the scene. The distribution workspace includes a checklist for ACX and Findaway that covers the kinds of file, loudness, and metadata questions those retailers publish. Treat it as a preparation list. Retailer rules change, and a checklist in a studio is not an acceptance letter. Confirm the current rules on the retailer’s own site before you upload.

VoxAnvil also does not claim your manuscript as training data. The privacy policy states that manuscript content is processed to provide preflight scanning and audiobook generation, that it is not sold, and that it is not used to train AI models without your explicit consent. Payment details are handled by the payment processor. Those are the boundaries published with the product.

If you want the production sequence in order, the narration walkthrough is the next page. If the problem you care about is numbers, abbreviations, and names, read the manuscript intelligence answer. If you already pay for ElevenLabs and want a careful comparison of doing the book there versus doing it in this studio, that comparison is separate so it does not pretend the voice engines are different products.

Frequently asked questions

Related reading

• How AI audiobook narration works in VoxAnvil
• How manuscript intelligence prepares numbers, abbreviations, and phonetics
• VoxAnvil pricing
• The Publishing Times Independent industry reporting on the wider book business. VoxAnvil does not publish or operate that site.