50% off on all Annual Plans.Get the offer
Narration Box AI Voice Generator Logo[NARRATION BOX]

How to Pass ACX Audio Requirements (Without a $5,000 Recording Studio)

Written by

Chetan Malhotra

Published

Last updated

ACX audio requirements guide for US and UK authors using AI voice cloning to publish Audible audiobooks without a recording studio

If you want your audiobook on Audible through ACX, every file has to pass a set of technical checks. Most rejections have nothing to do with the performance. They come from levels that are too quiet, peaks that are too hot, or a room that is too noisy.

You don't need a $5,000 studio to pass. You need a quiet space, a decent microphone, a careful recording routine and a simple mastering chain. This guide covers the numbers ACX checks, why home recordings fail, and how to get your files right before you upload.

One rule comes first: ACX requires human narration unless otherwise authorized. You can narrate the book yourself or hire a narrator, but AI or text-to-speech narration can't go through a standard ACX upload. If you want an AI-narrated edition, the last section explains where it can go instead.

TL;DR

  • ACX checks RMS, peak level, noise floor and file format on every file. Most rejections are technical , not about storytelling.
  • A quiet, soft room matters more than an expensive microphone.
  • Record at a steady distance and gain, then master every chapter with the same settings.
  • Check each file with the free ACX compliance checker before you upload.
  • ACX needs a human narrator. AI-narrated audiobooks belong on stores that accept them, such as Spotify for Authors, Google Play Books and Kobo Writing Life.

What ACX checks in every file

These are the core requirements from ACX's audio submission rules:

  • RMS (average loudness): between -23 dB and -18 dB
  • Peak level: no higher than -3 dB
  • Noise floor: below -60 dB
  • Format: MP3 at 192 kbps or higher, constant bit rate, 44.1 kHz
  • Room tone: a short stretch of silence at the start and end of each file, no more than 5 seconds
  • File length: one chapter or section per file, and no file longer than 120 minutes
  • Credits: opening and closing credits uploaded as their own files
  • Retail sample: up to five minutes, with no explicit content

If your noise floor sits at -55 dB because of an air conditioner, the file fails. If your RMS lands at -25 dB because you were cautious with gain, it fails too. For a deeper explanation of each number, see ACX audio specs explained .

Why home recordings get rejected

The same problems come up again and again:

  • Room noise: fridges, fans, traffic and computer fans push the noise floor above -60 dB.
  • Echo: bare walls and hard floors make the voice sound distant, and no plugin fully removes it.
  • Uneven levels: moving closer to or further from the mic changes loudness from line to line.
  • Over-processing: heavy noise reduction or compression to hit the numbers makes the voice sound thin or watery.
  • Missing room tone or credits: easy to forget and an easy reason for a file to come back.

Every rejection means editing, re-exporting and re-uploading. On a 20-chapter book, a few hours of setup work up front saves far more time later.

A home setup that passes

You can get a clean noise floor without building a booth.

  1. Pick the quietest room you have. A walk-in closet full of clothes is a classic choice. Soft furnishings absorb echo.
  2. Turn off what hums. Air conditioning, fans, fridges in the next room and notifications on your phone.
  3. Use a decent microphone. A good USB microphone or an XLR microphone with a small audio interface is enough. A pop filter helps with plosives.
  4. Fix your distance. Stay a consistent hand-width or two from the mic. Mark your position if it helps.
  5. Set gain once and leave it. Aim for speech peaks well below 0 dB while recording, then leave the gain alone for the whole book.
  6. Record room tone. Capture 30 seconds of silence in your room at the start of each session. You'll use it for edits and file boundaries.

Before you record the whole book, record one chapter and run it through the ACX compliance checker . If the raw noise floor is already near -60 dB, your room is good enough.

Editing and mastering to ACX specs

Keep the chain simple and use the same settings on every chapter:

  1. Edit first. Remove mistakes, long breaths and mouth clicks. Fill gaps with your recorded room tone, not digital silence.
  2. Light noise reduction only if needed. Use a gentle setting based on a sample of your room tone.
  3. Gentle compression. Evens out loud and soft passages without squashing the voice.
  4. Normalize or apply gain so RMS lands between -23 dB and -18 dB.
  5. Limit peaks to -3 dB or lower.
  6. Add room tone at the start and end of each file.
  7. Export as MP3, 192 kbps constant bit rate, 44.1 kHz, the same mono or stereo setting for every file.

Then check every exported file, not just the first one. Chapters recorded on different days often land at different levels. Our guide to audiobook mastering goes deeper on RMS, LUFS and noise floor.

Uploading to ACX

Once your files pass:

  • Upload your cover and confirm your title details
  • Upload opening credits, each chapter, and closing credits in order
  • Choose your retail sample
  • Submit for review

ACX reviews the files before the book goes live. If a file comes back, the message usually names the problem. Our guides to what happens during ACX review and what to do if ACX rejects your audiobook walk through the common fixes.

Before you choose a distribution option, read how Audible royalties work and weigh Audible exclusive against going wide .

What about AI narration?

ACX's rules say audiobooks must be narrated by a human unless otherwise authorized, and they prohibit unauthorized text-to-speech, AI and automated recordings. That includes a clone of your own voice made with a third-party tool.

Audible does sell some AI-narrated titles, but only through Amazon's own programs: KDP Virtual Voice, an invite-only beta where Amazon generates the audio, and ACX Voice Replica, a limited beta for human narrators.

If you want an AI-narrated edition, publish it on stores that accept AI narration you produce yourself, such as Spotify for Authors, Google Play Books and Kobo Writing Life, with the disclosure each store asks for. Our AI audiobook distribution map lists each store's rules and labels.

Narration Box's Audiobook Creator turns an EPUB, PDF or Word manuscript into chapters you can narrate, edit and export for those stores. You can also use voice cloning to narrate in your own voice.

Frequently asked questions

What are the audio requirements for ACX?

RMS between -23 dB and -18 dB, peaks no higher than -3 dB, a noise floor below -60 dB, and MP3 files at 192 kbps or higher, constant bit rate, 44.1 kHz. Each chapter goes in its own file, no longer than 120 minutes.

Can I record an ACX audiobook at home?

Yes. A quiet room with soft furnishings, a decent microphone, consistent mic distance and careful mastering are enough for many authors to pass.

Should I record at 44.1 kHz or 48 kHz?

ACX wants 44.1 kHz MP3 files. You can record at 48 kHz, but export the final files at 44.1 kHz.

How big is one hour of audio at 192 kbps?

About 85 to 90 MB per hour as an MP3.

Can I use AI narration on ACX?

Not through a standard upload. ACX requires human narration unless otherwise authorized, and that includes a clone of your own voice made with a third-party tool. AI-narrated editions can go to Spotify, Google Play Books and Kobo with a disclosure.

Can I upload audio I recorded myself?

Yes. ACX accepts self-produced, human-narrated audio as long as every file meets the technical requirements.

Check your files before you upload

You don't need a $5,000 studio. You need a quiet room, a consistent routine and files you have checked. Run every chapter through the free ACX compliance checker before you submit.

Check out similar posts

Get Started with Narration Box Today

Choose from our flexible pricing plans designed for creators of all sizes. Start your free trial and experience the power of AI voice generation.