How Long Does It Take to Make an Audiobook? A Timeline by Route

Introduction

Before you set a budget or a launch date, one question comes first: how long does it take to make an audiobook? The answer depends on the route you pick. A human production takes weeks to months, and directed AI production takes days.

The Audio Publishers Association's APA audiobook sales survey puts 2025 audiobook sales at $2.43 billion, up 9% over the previous year. In the same release, publishers reported over 750,000 active titles, a 43% increase from 2024. The Audio Publishers Association's APA listener research finds that 58% of US adults, an estimated 157 million people, have ever listened to an audiobook.

This guide is for indie authors, including authors with a backlist and authors who want the book in their own voice. You will learn the finished-hour math, what happens at each stage, how the routes compare, and what moves the schedule. It is written from the producer's desk, where the time actually goes.

How Long Does It Take to Make an Audiobook?

By Lex's production figures, a human-narrated audiobook takes weeks to months from manuscript to finished files. On those figures, a narrator spends 2 to 4 hours in the booth per finished hour, and editing and mastering start after that. Directed AI production returns a first chapter in about a day and the full book in days.

Store review then adds time on each retailer's own schedule. Whichever route you take, the main input is the length of the book, measured in finished hours. So that is where the math starts.

What Is a Finished Hour?

A finished hour is one hour of edited, ready-to-publish audio, roughly 9,300 words. Narrators and studios price and schedule in this unit, so every quote and calendar starts from it. Booth time runs 2 to 4 hours per finished hour, because the raw session holds every stop and retake.

The National Center for Voice and Speech puts the average US speech rate at about 150 words per minute. That pace works out to about 9,000 words an hour. It describes general speech rather than narration, so treat the match with 9,300 as approximate.

How Long Is an Audiobook From Your Word Count?

The formula is one line: word count ÷ 9,300 = finished hours. Multiply the result by 2 to 4 for the booth hours a human narrator needs.

Word count	Finished hours (÷ 9,300)	Human booth hours (2–4 per finished hour)
50,000	≈5.4	≈11–22
75,000	≈8	≈16–32
100,000	≈10.8	≈22–43
120,000	≈13	≈26–52
150,000	≈16	≈32–65

The rule is a planning number.

Pacing and scene breaks move the runtime. Plan with the rule, then price from the files, as our guide to pricing by finished length explains.

How Long Does Each Stage Take, by Route?

Every audiobook passes through the same four stages: preparation, recording, editing, and store review. The route decides who does the work at each stage and how long it takes. If you are working out how to create an audio book, choose the route first.

Authors pick from four routes. You can hire a human narrator or studio, record your own audiobook, run a raw AI engine, or use directed AI production. On a raw engine you become the producer, while with directed AI the voice is AI, but the performance is human.

Stage 1: Preparation and Casting

Preparation starts with the manuscript. Build a pronunciation list for character names and invented terms. Mark chapter and scene breaks, and decide which front and back matter gets read aloud.

On the human route, casting means auditions and then a booking. The narrator's calendar drives the schedule from there. If you plan to narrate yourself, the casting question is whether your voice suits the book.

With directed AI, casting is a voice decision. You choose from voice models, or you clone your own voice from about one minute of reading, an uploaded recording, or a YouTube link. Weeks of auditions compress into one choice, and the author's voice reaches the book without a booth.

Stage 2: How Long Does It Take to Record an Audiobook?

Recording is the slow, expensive part of a human production, and it scales with length. At 2 to 4 booth hours per finished hour, a 75,000-word novel (about 8 finished hours) needs about 16 to 32 hours in the booth.

Those hours spread over days of sessions. The National Center for Voice and Speech explains how extended voice use causes fatigue, noting that long stretches of talking may result in vocal fatigue. A narrator's voice has to sound the same in every chapter, so sessions get paced.

Recording a book yourself adds retakes and learning time on top of the booth ratio. The schedule then depends on the evenings you can give it. Directed AI production has no booth: a producer marks affect, pace, and pauses on every line, then retakes anything that misses.

Stage 3: How Long Does It Take to Edit an Audiobook?

Editing covers four jobs. The editor cuts breaths and retakes, then proofs the audio against the manuscript for dropped or misread lines. After that comes mastering to retailer loudness and peak specs, then QC.

Every Lex file is checked against the audiobook file specs stores check. Head and tail silence runs 1 to 5 seconds per file.

Editing time scales with finished hours and with the number of correction rounds. On a human production, fixes mean rebooking the narrator, so ask for line retakes, because full-chapter redos cost time.

Stage 4: Store Review and Going Live

ACX is Audible's self-serve marketplace for human-narrated audiobooks. ACX self-serve still expects human narration. ACX does not accept externally produced AI narration.

That matters on any AI route. Audible on the Publisher pack goes under Lex's publisher agreement: a direct Lex upload. Our guide to where AI-narrated audiobooks sell covers the rest of the stores.

Review timing belongs to each retailer, and no single window is a guarantee. The store list and disclosure rules are on our distribution page.

Going wide also means libraries. OverDrive reported 315.9 million audiobook loans in 2025, up 13%, across its global Libby and Sora network.

How Do the Routes Compare?

Here is each route against the stages above. The table uses only the figures covered in this guide.

Route	Who does the work	Booth time	Turnaround	Cost basis	Audible path
Human narrator or studio	Narrator records; editor proofs and masters	2–4 hours per finished hour	Weeks to months	$200–$400 per finished hour narration, plus $75–$150 editing, proofing, and mastering	ACX self-serve
Self-recording	You narrate, edit, and master	Your own sessions, plus retakes and learning time	Depends on your evenings and retakes	Your equipment and your time	ACX self-serve (human narration)
Raw AI engine (general-purpose text-to-speech)	You generate, check, and fix	None	Your hours of generation and QC	Engine fees plus your time	ACX does not accept externally produced AI narration
Directed AI production (Lex)	A producer directs every line and masters to spec	None	First chapter in about a day, full book in days	$49 Voiceover or $99 Studio, one-time, 10 finished hours included	Publisher pack: direct Lex upload under Lex's publisher agreement

Directed AI sits between a flat robot read and paid booth time. You can see what each tier includes in the Lex production packages.

What Changes the Timeline?

The same 75,000-word novel can take days or months. The gap comes from a handful of variables.

Each one below comes from the producer's desk, along with the practical questions authors ask once they see the calendar.

Book Length, Genre, and Cast

Length scales booth and edit time in a straight line through finished hours. A 150,000-word book is about 16 finished hours, roughly double a 75,000-word novel. On a human route, booth time and cost roughly double with it.

Character-heavy fiction needs distinct voices and more pronunciation work. With directed AI, characters are cast across voice models without extra booth sessions. A full cast is available on Publisher.

What Is the Most Common Scheduling Mistake?

The most common mistake is treating the day the files are finished as the launch date. Store review runs on each retailer's timeline, and a spec miss sends the book back.

Build the retail package before upload: silence, sample, credits, cover, and metadata.

Hear a sample before you commit to a route or a date. Lex's first-chapter demo is finished within 24 hours.

How Much Does It Cost to Produce an Audiobook?

In Lex's production cost breakdown, human narration costs $200 to $400 per finished hour, plus $75 to $150 per hour for editing, proofing, and mastering. Add a few hundred dollars for QC and retail file prep. By the same breakdown, a human-narrated 75,000-word novel costs $2,400 to $5,850 produced. Time and cost move together here, because booth hours drive both.

Lex prices by pack. Voiceover is $49 and Studio is $99, one-time, each with 10 finished hours included and extra hours at $5/hr and $10/hr. Publisher is a custom quote per book, with extra hours at $20/hr.

Voiceover and Studio take no royalties, so you keep 100% of every sale and own the finished masters. The full math is in our breakdown of what audiobook production costs.

What Equipment Do I Need to Record Audiobooks?

To record your own audiobook, start with a quiet room and a microphone. You also need editing software that can master to −23 to −18 dB loudness with clean silence pads. The bigger time cost is learning to master to spec.

A voice clone asks for much less. It needs about one minute of clean reading in a browser, or a recording you already have. Authors who want the book in their own voice can skip the home-studio build.

Summary

Start with the math. Divide your word count by 9,300 to get finished hours, then plan every stage from that number.

• A human production needs 2 to 4 booth hours per finished hour and takes weeks to months overall.
• Directed AI production returns a first chapter in about a day and the full book in days.
• Store review runs on each retailer's timeline, so set the launch date after the files pass spec.
• ACX self-serve does not accept externally produced AI narration; Audible on Publisher is a direct Lex upload.

Next, count your words and pick a route. Prepare the pronunciation list early, and leave a launch buffer for store review. Whichever route you choose, own the files and judge the result by what the listener hears.