How to Self-Publish an Audiobook: What the Format Involves

To self-publish an audiobook, you turn your finished book into a narrated recording and release it through audiobook retailers yourself. The work has four stages: adapt the manuscript into an audio script, record the narration (your voice, a hired narrator, or approved AI), prepare the files to retailer standards, and choose between selling exclusively through one store or widely through many.

How Long Will Your Audiobook Run? The Finished Hour Explained

Audiobook length is measured in finished hours: one hour of edited, mastered audio, ready for sale. Mastering is the final technical polish that sets a recording’s loudness and consistency, and a finished hour counts only that completed audio; raw recording time is always longer, so the two should never be confused. Narrator quotes and production schedules are priced per finished hour, and on Audible the retail price band follows length as well.

Estimating your own figure takes one division. Narrators read at a steady pace, and ACX (Audiobook Creation Exchange), Amazon’s audiobook platform, estimates that 9,300 words of text produce about one finished hour of audio. An 80,000-word novel therefore runs to roughly 8.5 finished hours, because 80,000 divided by 9,300 comes out near 8.6. A 50,000-word non-fiction book lands close to 5.5 hours.

The format rewards that arithmetic. The Audio Publishers Association’s 2026 sales survey put US audiobook revenue at $2.43 billion for 2025, up 9% on the year, and publishers reported more than 750,000 active titles (a 43% jump from 2024). For an author weighing an audio version of an existing book, the audience is clearly there.

What Changes When a Book Becomes an Audio Script?

A manuscript is never read aloud exactly as printed; it becomes an audio script first. Anything that only works visually has to be cut, described, or restructured before recording starts, and a few spoken elements are added that the print edition never needed.

The subtractions come first. Hyperlinks, image captions, and “see chapter three” cross-references mean nothing to someone listening in a car. Tables, charts, maps, and illustrations either get rewritten as descriptive sentences or dropped, and footnotes are folded into the surrounding narrative or left out; the Alliance of Independent Authors (ALLi) guide to audiobooks walks through this adaptation work in detail. It follows that books built around visuals resist the format hardest: cookbooks, travel guides organised around maps, heavily illustrated titles, and reference books that depend on tables all lose something essential in audio.

The additions are the credits. An audiobook opens with a short spoken announcement (the title, author, and narrator, stated before chapter one) and closes with a matching credit that tells the listener the book has ended. Both are recorded separately from the chapters; ACX, for one, requires the credits as their own files.

Should You Narrate It Yourself, Hire a Professional, or Use AI?

Three narration routes exist for a self-published audiobook: record it in your own voice, pay a professional narrator, or (on the platforms that permit it) generate a synthetic-voice edition from your ebook. Each route trades money against time and control differently, so compare them before buying gear or signing contracts.

Hiring a professional narrator

Hiring a professional narrator is priced per finished hour. ACX’s own guidance puts retail-ready production at roughly $200 per finished hour for narration, plus about the same again for editing, mixing, and mastering (together called post-production). For comparison, the SAG-AFTRA union minimum on ACX is $250 per finished hour. At a combined rate of about $400 per finished hour, an 8.5-hour novel budgets near $3,400; lighter post-production and newer narrators pull that figure down.

On ACX’s marketplace, payment follows one of two models. Pay-for-production means a per-finished-hour fee, paid when you approve the final audio, and you keep all your royalties (your share of each sale’s income). A royalty share means the narrator records for nothing upfront and instead takes half of your royalties for the duration of the distribution agreement; on ACX this option is tied to exclusive distribution, which the distribution section below explains. Away from the platform, narrators and studios negotiate their own rates and terms.

Narrating it yourself

Self-narration swaps cash for time. Professional narrators typically spend about two hours in the booth plus another two to three hours of editing for every finished hour; first-timers should budget more for both, plus re-recording. The room matters more than the microphone: a quiet space softened with rugs, curtains, or panels beats an expensive mic in an echoey one. A workable chain is a decent microphone, an audio interface (the small box that connects the mic to your computer), wired closed-back headphones (which keep sound from leaking into the mic), and recording software. ACX’s gear guide recommends the low-cost Reaper licence or free Audacity, and advises against GarageBand for audiobook work.

AI narration, where platforms allow it

Synthetic narration is platform-gated rather than a general option. ACX’s audio submission requirements (covered in the next section) prohibit text-to-speech and AI narration in indie submissions unless Audible has authorised it. Amazon’s separate KDP Virtual Voice beta instead generates an audiobook from an eligible KDP ebook by invitation, currently for the US marketplace, with a choice of more than 50 voices, list prices between $3.99 and $14.99, and a 40% royalty. Apple Books digital narration does a similar job free of charge through Apple’s approved distribution partners and labels the result “Narrated by Apple Books”. Kobo Writing Life accepts AI-narrated audiobooks when the narrator is listed in the book’s store details (the metadata) as a synthesized voice. Listeners remain lukewarm, however: the same APA survey found AI-narrated titles made up just 0.03% of 2025 audiobook revenue. And if AI touched the book’s text as well as its narration, KDP has separate rules for declaring AI use.

What Audio Standards Must Your Files Meet?

Audiobook retailers accept or reject files against published, measurable standards (average loudness, background noise, file format), not against a vague sense of what sounds professional. ACX’s audio submission requirements are the most specific public list and make a practical mastering target, but treat them as one platform’s worked example: every store publishes its own version. Google Play Books, for instance, also accepts AAC, FLAC, and WAV files, with a lower 128 kbps floor for mono MP3s.

ACX’s headline numbers, in plain language:

• File format: MP3 at 192 kbps or higher, constant bit rate (CBR), at 44.1 kHz; all three are export settings you choose in your recording software. CBR means the quality setting stays fixed for the whole file, which the store’s encoding systems require.
• Average loudness: between -23 dB and -18 dB RMS. RMS is the average level across a file, and decibel values here are negative, so the nearer to zero, the louder. Keeping every chapter in this band saves listeners from riding the volume control.
• Peaks: no higher than -3 dB, which leaves headroom against distortion.
• Noise floor: below -60 dB RMS, a measure of how loud your “silence” is between sentences.
• Room tone: 1 to 5 seconds at the start and end of each file. Room tone is the natural sound of your quiet room, not digitally inserted dead silence.
• File structure: one chapter or section per file, no file longer than 120 minutes, and every file in the same channel format (all mono or all stereo, meaning one-channel or two-channel sound). Opening and closing credits are uploaded as their own files, along with a retail audio sample of five minutes or less; the first five minutes of the book is the recommended choice.

ACX’s free Audio Lab tool checks a finished file against these numbers before you submit, so failures surface at your desk rather than in the review queue.

The audio edition also needs its own cover, and it is square. ACX asks for at least 2,400 × 2,400 pixels in RGB colour (the on-screen colour system; the same RGB-versus-CMYK distinction that applies to print covers), while Spotify for Authors recommends 3,000 × 3,000 pixels. A rectangular ebook cover cannot simply be resized; the artwork has to be recomposed as a true square.

Exclusive or Wide: Where to Self-Publish an Audiobook in 2026

Audiobook distribution comes down to how widely you sell: through one retail group exclusively, or “wide” across many stores and library services. The exclusivity trade, a higher royalty in exchange for selling in one place, is specific to Audible’s ACX; the other channels below are non-exclusive and can be combined. An aggregator (a distributor that sends one upload out to many channels for a share of income) covers the stores that take no direct uploads. Here are the main options, stated neutrally.

ACX feeds Audible, Amazon, and Apple Books. Under the royalty model that opened to all creators on May 26, 2026, exclusive distribution pays a 50% royalty and non-exclusive pays 30%, each with an added share of Audible membership-listening revenue; the legacy 40% and 25% rates retire by the end of 2026. ACX agreements run an initial seven-year term. A title can switch from exclusive to non-exclusive once, after 90 days on sale; a royalty-share title is the exception and stays exclusive for the term. While a title is ACX-exclusive it sells only through those three storefronts, so library channels are reached through wide distribution instead.

Spotify for Authors takes direct uploads free of charge, non-exclusively, and leaves ownership with you. À-la-carte sales (single-copy purchases, as opposed to subscription listening) pay 50% of the suggested retail price, and subscription listening is paid pro-rata (in proportion to time listened) from a shared revenue pool.

Voices by INaudio is the wide-distribution aggregator formerly known as Findaway Voices. Spotify wound that service down on August 1, 2025, and Findaway’s original founders relaunched it as the independent INaudio; ALLi’s coverage of the change explains what moved where. One upload reaches dozens of retail and library channels.

Kobo Writing Life takes direct uploads non-exclusively and pays 45% of list price on à-la-carte sales priced at $2.99 or above (35% below that), and 32% on Kobo Plus subscription purchases.

Google Play Books takes direct audiobook uploads for accounts in supported countries, against the file specs linked in the previous section.

Producing an audiobook is a sequence of decisions more than a single task: measure the length in finished hours, adapt the text into a script, choose who narrates and at what cost, master to the published specs, and only then commit to a distribution model. Taken in that order, each step sizes the next one.

Frequently Asked Questions

How much does it cost to self-publish an audiobook?

Professional production typically lands in the low-to-mid thousands for a full-length book: ACX’s guidance estimates roughly $200 per finished hour for narration plus about the same for post-production, so an 8.5-finished-hour novel works out near $3,400 at those rates. Self-narration cuts the cash cost to equipment and software; royalty-share deals remove the upfront fee but commit half your royalties for the contract term.

Can you self-publish an audiobook on Audible?

Yes: Audible takes self-published audiobooks through its ACX platform. You claim your book, produce or upload audio that meets the submission requirements, and choose exclusive (50% royalty) or non-exclusive (30%) distribution under the model introduced in May 2026. Titles must be human-narrated unless Audible has authorised otherwise.

What software do you need to record and edit an audiobook?

Free Audacity handles recording and editing, ACX’s gear guide also recommends the low-cost Reaper licence and notes Adobe Audition for experienced editors, and the same guide advises against GarageBand for audiobook work. Whatever you record in, ACX deliveries must export as 192 kbps or higher constant-bit-rate MP3 files at 44.1 kHz.

Does an audiobook need its own ISBN?

An audiobook counts as its own edition, so it never shares the print or ebook ISBN (the 13-digit book identifier). Publish only through ACX and no ISBN is required, because Amazon assigns an ASIN, its own product number; distribute wide and you supply your own. We covered the identifier question separately in Do You Need a Separate ISBN for Your Audiobook?

What does an audiobook production company do?

A production company handles the whole audiobook pipeline for a fee: casting and directing the narrator, editing, mastering, quality control, and delivering retail-ready files. Some also prepare the square cover and the store metadata. Scope varies by contract, so confirm in writing which deliverables you receive, and which rights to the finished files come with them.