Producing an AI audiobook: how to do it, what it costs and tips

For many readers, the audiobook is the obvious way to enjoy a book – on the bus, during a walk or at the kitchen sink. Yet it has long been out of reach for most self-publishers, because a traditional studio recording costs a lot of money and takes a long time. With AI voices, that has changed. Today you can produce an audiobook from your own manuscript in a few hours and have it out on the streaming services the same day.

In this guide we go through the whole journey: what an AI audiobook is, how it differs from a studio recording, how you prepare the manuscript, how production works step by step, what it costs and how you get the best possible result.

What is an AI audiobook?

An AI audiobook is an audiobook where the text is read aloud by a synthetic voice instead of by a human narrator. The AI models of recent years emphasise, pause and intonate the text in a way that is very close to a human reading. The voices understand the context of the sentence and can handle dialogue, questions and shifts in emotion.

So it is not the same thing as the monotone screen reader many associate with speech synthesis. A modern AI voice reads an entire passage at a time, takes into account what has been said earlier and adapts tempo and emphasis to the content.

Listeners are also more positive than many think. In the Swedish Publishers' Association's Book Barometer 2026, a majority (54%) of the audiobook listeners who have themselves listened to AI-narrated audiobooks reply that they are positive towards synthetic voices and can imagine continuing to listen to them. Read more in the post Swedes increasingly positive towards AI-narrated audiobooks.

AI narration or studio recording?

There are two ways to make an audiobook: hiring a human narrator in a studio or letting an AI voice read the text. Both have their place, and the differences are primarily about time, cost and control.

Time

A studio recording begins with finding the right narrator and booking a studio. Then the book has to be read in, often with several retakes, and after that edited and mastered. This normally takes weeks and sometimes months. An AI audiobook is produced in under an hour for a full-length book, and the whole process including pronunciation review is normally done within a few hours.

Cost

In a studio production you pay for the narrator, studio time, sound engineer and post-processing. With AI narration those costs disappear, and you pay a fraction of the price. See the calculation examples further down.

Control

With AI you choose the voice yourself and listen to a sample of your own text before you pay. You don't have to guess how the book will sound. The voice also maintains the same quality and tone from the first to the last chapter.

When is a studio recording a better fit?

A skilled human narrator can give an interpretation and a personality that is hard to replace, and some authors want to read in their book themselves. If you already have a studio-recorded audiobook you can distribute it through GoRead instead of producing a new one – more on that further down.

Why should you publish your book as an audiobook?

You reach the listeners. The streaming services have many users who mainly consume books in audio format. Without an audiobook you don't reach them.
Low cost. You pay a fraction of what a studio recording with a narrator, studio and sound engineer usually costs.
Fast. The audiobook is ready within a few hours instead of weeks or months.
You're in control. You choose the voice yourself and can listen to a sample of your own text before deciding.
No lower or upper limit on length. A half-hour short story works just as well as a fifteen-hour novel.
More languages. Combine the audiobook with a translation and reach listeners in several countries.
You own the result. You buy and own the finished audio file and keep the copyright to your text.

Prepare the manuscript before you start

An AI voice reads exactly what is in the manuscript. The time you put into preparation is therefore heard directly in the finished audiobook.

Correct file format and clear headings

The manuscript is uploaded as a Word file (.docx). Set chapter headings in Heading 2 and subheadings in Heading 3 and 4. Then the audiobook gets the right chapter division. Simpler changes, such as changing heading level or merging and splitting chapters, you can make directly in the service after uploading.

Proofread

A spelling error that the eye skips over in print becomes a mispronounced word in an audiobook. When you upload the manuscript we do a free spell check that you can choose to review. Correct what you want and upload the manuscript again before you order the production.

Consider what is not read aloud

Images, tables, footnotes and captions are not read aloud in the audiobook. If they contain information the listener needs, you can write it into the body text. A table can for example become a short passage that summarises the figures.

Write for the ear

Try reading a chapter aloud to yourself. Very long sentences, parentheses within parentheses and references such as "see the image above" work worse in audio than in text. Punctuation controls the voice: full stop, comma and dash determine where it pauses and how it emphasises. Two line breaks in the manuscript mark a scene change or time jump.

How to produce an AI audiobook with GoRead – step by step

Upload your manuscript. Log in with BankID and upload the manuscript as a Word file (.docx). The e-book and the interior of the printed book are created automatically and free of charge, even if you only intend to publish the audiobook.
Fill in details. Enter title, author and language. We suggest subject categories and a back-cover text that you can edit. If you have used AI to write the book, this must be stated.
Choose a cover. The audiobook gets a square cover. Upload your own (at least 1000 × 1000 pixels) or choose among three AI-generated suggestions created based on the book's content. If you only have a portrait cover we create the square one for you.
Listen to samples of the voices. During the Audiobook step you can have our AI voices read any chapter of your book in minute-long clips. It costs nothing, so take your time.
Order the production. Once you have chosen a voice you see the price for your specific book before you pay.
We review the pronunciation. Before production starts we go through unusual words, names, places, abbreviations, numbers, dates and words in foreign languages, and check homographs – words that are spelled the same but pronounced differently depending on context.
The audiobook is produced. Production is normally done within a few hours. You receive an email when the audiobook is ready and can listen through it in the service.
Publish. Under the Streaming tab you publish the audiobook to Storytel, BookBeat, Nextory and Audible. It normally becomes available to listeners within a few hours. You can also set a future publication date.

How to choose the right voice

The voice is the most important choice you make. It determines how the listener experiences the narrative voice, the tempo and the mood of the book.

Start from the story. A voice that suits a calm feelgood novel does not always suit a thriller or a non-fiction book.
Listen to samples of different kinds of text. Choose a dialogue-heavy chapter and a more descriptive passage. The voice should work in both.
Listen like your reader. Put on the clip while you do something else, as listeners do. Is the voice pleasant to have in your ears for several hours?
Compare at least three voices. The sample listening is free, and the differences are heard most clearly when you compare the same passage.

How to get the pronunciation right

What most often gives away a carelessly made AI audiobook is the pronunciation: a place name emphasised wrongly, a year read as an ordinary number or an abbreviation that is spelled out. That is why a pronunciation review is included in every production at GoRead.

After you have ordered, we go through the manuscript and look for what an AI model may struggle with: unusual words, names of people and places, abbreviations, numbers, dates and words in foreign languages. We also check homographs, that is words that are spelled the same but pronounced differently depending on meaning. Based on this, a special audiobook manuscript is created and sent to the AI model.

The review is done manually. It normally goes quickly, but can take a few hours in the evenings and at night.

What does it cost to produce an AI audiobook?

The price is a start fee of SEK 249 plus SEK 300 per hour of finished audiobook. One hour of audio corresponds to roughly 45,000–50,000 characters including spaces. Prices include VAT, and the distribution itself to the streaming services is free. Some calculation examples:

Short story, about 50,000 characters (approximately 1 hour): about SEK 550.
Shorter novel, about 150,000 characters (approximately 3 hours): about SEK 1,150.
Full-length novel, about 350,000 characters (approximately 7 hours): about SEK 2,350.
Longer novel, about 500,000 characters (approximately 10 hours): about SEK 3,250.

You can see the number of characters in your manuscript in Word under Review and Word Count. The exact price is calculated automatically and shown before you pay.

This is included in the price:

sample listening of your own text with all voices, before you order
pronunciation review of names, numbers, abbreviations and unusual words
finished audiobook in MP3 and WAV, which you own
distribution to Storytel, BookBeat, Nextory and Audible
ISBN for the audiobook

If against expectation you are not satisfied with the quality of the audiobook, we have a money-back guarantee.

How long does it take?

Upload, details and cover: quick if the manuscript is prepared.
Sample listening: as long as you want. It's free.
Pronunciation review: normally quick, but up to a few hours in the evenings and at night.
Production: under an hour for a full-length book, considerably shorter for a short story.
Publication: the streaming services normally make the book available within a few hours.

This means you can upload a finished manuscript in the morning and have the audiobook out with the listeners the same day.

Where does your audiobook end up and how do you get paid?

GoRead has agreements with Storytel, BookBeat, Nextory and Audible. Your audiobook becomes available to listeners in the countries where each service has customers. Audiobooks are distributed to the streaming services and not to online bookstores, where sales of audiobooks are very small.

You get 70% of the sales revenue, with monthly payouts within 60 days after each month-end. Every Monday the statistics are updated so that you can follow listening per title, channel and format.

We make no tax deduction on the payout. You declare the income yourself, and we submit a statement of earnings to the Swedish Tax Agency. If your company is VAT-registered, 6% VAT is paid out in addition to the amount.

Who owns the audiobook?

You do. You keep the copyright to your text, and you buy and own the produced audio file. GoRead is not a publisher and takes no rights to your book. If you later want to publish the book with a publisher, you can download your files and unpublish the book with us.

AI-generated material is labelled according to applicable rules, and GoRead's voices have the suffix [AI] in their name. The listener thus sees that the book is AI-narrated.

Audiobook in more languages

With GoRead you can translate your book with AI and then produce the audiobook in the target language. Audiobooks can be produced in Swedish, Danish, Finnish, Icelandic, Polish, Dutch, German, Bulgarian, British English and American English.

The translation is done in several steps. First the text is translated, then names, places and other local terms are handled separately, after that uncertain words and expressions are gone through, and finally a linguistic review is made. The title, cover and all details about the book are translated at the same time. The translation costs SEK 249 in start fee plus a fee per character, and you see the price before you order. A translation normally takes about an hour. Read more in the post Translations are a success.

Do you already have a studio-recorded audiobook?

If you already have an audiobook that is recorded in a studio and edited by a professional sound editor, you can upload it as MP3 or WAV and distribute it through GoRead. You do that under the Audiobook step, after you have uploaded the manuscript, filled in details and chosen a cover. It costs SEK 999. Read more in the post Get your audiobook out on the streaming services.

Common mistakes – and how to avoid them

Skipping the proofreading. Spelling errors are heard. Review the spell check before you order.
Choosing a voice based on a single clip. Listen to samples of several chapters and several voices.
Forgetting the heading levels. Without chapter headings in Heading 2 the chapter division will be wrong.
Leaving important information in tables and footnotes. They are not read aloud. Move the important parts into the body text.
Publishing without listening through. The audiobook can be listened to in the service before it is published. Listen at least to the beginning, a dialogue chapter and the end.
Publishing only one format. The e-book is created free of charge in the same process. Publish it at the same time and you also reach those who prefer to read.

When the audiobook is out

Marketing is not included in GoRead's service, so it's up to you to tell the world the book exists. A few simple things make a big difference:

Share a direct link to the book at each streaming service, so the listener lands in the right place with one click.
Talk about the book in your own channels the same day it is released, and again a few weeks later.
Ask the first listeners to rate the book in the app.
Follow the statistics every Monday and see which channels generate listening.
Do you have more books? Publish them too. Listeners who like one book often look for the next one by the same author.

Frequently asked questions about AI audiobooks

Does an AI audiobook really sound natural?

Yes, the quality has become very high. The voices intonate and emphasise the text with natural expression, and our pronunciation review ensures that names, numbers and unusual words come out right. The best way to judge is to listen to a sample of your own text, which is free.

How long does it take to produce an AI audiobook?

The production itself takes under an hour for a full-length book. The pronunciation review before production normally goes quickly, but can take a few hours in the evenings and at night since it is done manually. Expect the audiobook to be ready within a few hours.

How long can the book be?

There is no minimum or maximum length for an audiobook. The price depends on how many hours of finished audio the book becomes.

Which file formats do I get?

The audiobook is produced in MP3 and WAV. You own the files and can download them.

Do I also have to publish an e-book and printed book?

No, you choose yourself which formats you want to publish. The e-book and the interior are always created free of charge in the process, but you decide whether they should be published.

Can I publish the audiobook under a pseudonym?

Yes. However, the combination of author name and title may not be identical to or easily confused with an already published work.

Is it visible that the audiobook is AI-narrated?

Yes. We label AI-generated material according to applicable rules, and GoRead's voices have the suffix [AI] in their name.

Are AI models trained on my manuscript?

No. The language models GoRead uses are not trained on your manuscript.

Can I change the audiobook after it is published?

Published books cannot currently be changed in a simple way. If you find an error you can unpublish the book and publish a new, corrected version. Therefore listen through before you publish.

What happens if I'm not satisfied?

Then our money-back guarantee applies. Get in touch with us and we'll sort it out.

Get started with your AI audiobook

Upload your manuscript at GoRead and listen to a sample of your own text with our AI voices – completely free of charge. Do you also want to publish the book in printed format? Read our guide on publishing a book with print on demand. You'll find more answers in our frequently asked questions.