How to make an audiobook as a self-published author
Audiobooks are the format indie authors most often postpone — the production feels like a second job stacked on top of the first. It doesn't have to be. There are exactly three ways to get a book narrated, each with a different trade of money, time, and control, and the decision framework fits on a page. What you can't skip is the part that comes before any of them: an audiobook is an unforgiving format, and every inconsistency in your manuscript gets read aloud, slowly, into a listener's ears.
Path one: narrate it yourself
Self-narration costs the least money and the most time. You'll need a quiet space (closets full of hanging clothes are the classic budget booth), a decent microphone, recording software, and — this is the part people underestimate — the stamina to perform for many hours at a consistent energy level. Narration is a performance skill, not a reading skill. Listen to five minutes of a professional narrator and five minutes of an untrained author reading their own work and the gap is immediate: pacing, character differentiation, breath control, and the ability to sound the same on Tuesday as you did on Friday.
Self-narration makes the most sense for nonfiction (where the author's voice is part of the product), memoir, and authors who already have voice or performance experience. For fiction with a large cast, be honest with yourself about whether you can carry it. And budget real time for editing and mastering — the raw recording is roughly half the work. Every mouth click, page turn, and flubbed line has to come out, and the audio has to be processed to meet retailer specs. Many self-narrators record their own audio and then pay a professional to edit and master it, which is a sensible middle path.
Path two: hire a professional narrator
This is the standard route for indie fiction. Narrators price their work per finished hour — an industry convention meaning one hour of final, edited audio, which takes considerably longer than an hour to produce. Rates vary widely with the narrator's experience, union status, and whether editing and mastering are included, so get current quotes rather than trusting any number you read in an article. A rough rule of thumb for scoping: a typical novel runs somewhere in the range of nine to ten thousand words per finished hour, so a 90,000-word book lands near ten finished hours. Multiply that by a quoted per-finished-hour rate and you have your production budget.
There are also royalty-share arrangements on some platforms, where the narrator takes a cut of sales instead of (or in addition to) an upfront fee. These lower the cash barrier but cost more over the life of a book that sells, and the best-known narrators rarely take pure royalty-share deals on unproven titles. Whichever structure you choose: audition multiple narrators with the same passage (pick one with dialogue between two characters plus some narration), check their finished work on retail stores, and agree in writing on revision rounds, pronunciations, and delivery format before anyone records a chapter.
- Audition with a passage that has dialogue, at least two characters, and narrative prose.
- Send a pronunciation guide up front — every invented name, place, and term, spelled phonetically.
- Agree on revision rounds in writing. 'Unlimited revisions' is not a thing; two rounds of corrections is common.
- Confirm whether the quote includes editing, mastering, and proofing, or raw narration only.
Path three: AI narration — cheaper, faster, and complicated
AI narration has improved enormously and it's now a real option, especially for long backlists and nonfiction where a serviceable read beats no audiobook at all. It's dramatically cheaper and faster than human production. But the retail side is genuinely complicated, and this is where authors get burned: acceptance policies differ by store, differ by how the AI audio was produced, and have been changing. Some retailers accept AI-narrated audio produced through their own in-house programs but restrict third-party AI narration; others are more open; requirements about disclosing AI narration to listeners also vary. As of mid-2026 there is no single answer, and anything specific we printed here could be stale by the time you read it — verify the current policy of every store you plan to distribute to before you produce, not after.
The craft consideration is separate from the policy one: AI narration still struggles most with exactly the things fiction leans on — character voices, emotional escalation, comic timing, and invented-word pronunciation. Listeners forgive a flat read on a business book far more readily than on a romance or an epic fantasy. A defensible strategy many indies use: human narration for frontlist fiction where the listening experience drives reviews, AI narration to get deep backlist or supplementary material into audio at all.
The quality bar, rights, and where to distribute
Whichever path you take, retailers enforce technical standards before an audiobook goes live: consistent volume across the whole book, a quiet noise floor, no clipping or distortion, clean room tone (silence that sounds like the recording space, not digital dead air), opening and closing credits, and files organized to match the book's structure. Submissions that fail get bounced back, so if you're producing yourself, read your target store's technical spec before you record a word — fixing levels in mastering is easy, fixing a noisy room after the fact is not.
Two business basics. Rights: confirm you actually hold the audio rights — self-publishers who've never signed a contract almost certainly do, but if any version of the book ever went through a publisher or a boxed-set deal, check the paperwork. Your contract with a narrator should also spell out that you own the finished recording. Distribution: the recurring decision is exclusive versus wide — exclusivity with a single retailer typically pays a higher royalty rate on that store in exchange for not selling anywhere else, while wide distribution trades a lower per-store rate for presence in libraries and every other shop. Neither is universally right; it depends on where your ebook readers already are.
One last thing that costs nothing and saves the most: the manuscript you hand to production should be final. Narration is the one format where 'I'll fix it in the next upload' barely exists — a continuity error or a renamed character means re-recording, re-editing, and re-mastering the affected files, and paying for it. Run your full consistency pass before the narrator sees page one. This is honestly where Scribegrove fits in an audiobook workflow — it doesn't produce audio, but a whole-book Story Doctor scan flags contradictions, timeline slips, and continuity breaks with chapter-and-scene anchors before they get expensive, and you can export a clean DOCX or PDF for your narrator when it's done.
Frequently asked
How long does a book take to narrate?
Scope it in finished hours: a typical novel runs roughly nine to ten thousand words per finished hour of audio, so a 90,000-word book is around ten finished hours. Production time is a multiple of that — recording, editing, mastering, and proof-listening commonly take several hours of work per finished hour, which is exactly what a per-finished-hour rate is pricing.
Can I use AI narration and publish everywhere?
Not automatically. Store acceptance of AI-narrated audio varies by retailer and by how the audio was produced — some accept it only through their own programs, some restrict third-party AI narration, and disclosure requirements differ. Policies have been changing, so verify each store's current rules before production. Producing first and checking after is the expensive order of operations.
This guide is general information for authors, not legal advice. Platform and store policies change — verify the current terms wherever you publish.