Five years ago, "publish an audiobook" meant ACX. Now it means three platforms minimum if you want full coverage, and each one publishes a different spec sheet with different numbers. The natural assumption is that you need three different masters. You do not. The platforms disagree on policy, not on what a clean audiobook master looks like.
Here is exactly what each platform requires in 2026, where the specs differ, and the single master that satisfies all of them.
The Three Platforms in 2026
- Audible / ACX: The original platform. Exclusive or non-exclusive distribution through Amazon's ACX portal. Automated QC. Royalty share or per-finished-hour models.
- Findaway Voices (Spotify): Spotify-owned wide-distribution platform. Distributes to 40+ retailers including Apple Books, Google Play, libraries, Storytel. Human QC. Per-unit royalty.
- Spotify Audiobooks: Direct distribution to Spotify's in-app audiobook catalog. Accessed either via Findaway Voices distribution or directly by qualifying publishers. Automated QC similar to ACX.
Most independent narrators and small publishers ship to all three, either directly or via Findaway's downstream distribution. The question is what master to make.
Side-by-Side: The Numbers
| Parameter | ACX | Findaway Voices | Spotify Audiobooks |
|---|---|---|---|
| Loudness | -23 to -18 dB RMS | -23 to -18 dB RMS (recommended) | -24 to -14 dB RMS |
| True Peak Ceiling | -3 dBTP | -3 dBTP | Not specified |
| Noise Floor | -60 dBFS or lower | -60 dBFS or lower | Under -60 dB |
| Sample Rate | 44.1 kHz | 44.1 kHz | 44.1 kHz |
| Bit Depth | 16-bit | 16 or 24-bit | 16-bit |
| File Format | MP3 192 kbps CBR | WAV preferred, MP3 accepted | MP3 preferred, WAV or FLAC |
| Channels | Mono or stereo | Stereo preferred | Mono or stereo, not mixed |
| Head Silence | 0.5 to 1 sec | 1 sec recommended | 0.5 to 1 sec |
| Tail Silence | 1 to 5 sec | 2 to 5 sec recommended | 1 to 5 sec |
| Chapter Structure | One file per chapter, max 120 min | One file per chapter | One file per chapter, max 120 min |
| QC Type | Automated | Human | Automated |
Specs sourced from ACX audio submission requirements, Findaway Voices delivery documentation, and the Spotify for Authors Metadata & Assets Guide (last updated 11/2024).
Where the Specs Actually Differ
Loudness Target Window
ACX and Findaway publish the same -23 to -18 dB RMS window. Spotify Audiobooks is the loosest of the three: the official guide specifies RMS between -24 and -14 dB, wider on both ends. Any master inside the ACX window automatically sits inside Spotify's. A master at -20 LUFS integrated, which lands around -20 dB RMS on dense spoken word, passes all three with margin.
True Peak Ceiling
ACX and Findaway specify a -3 dB peak ceiling. Spotify Audiobooks publishes no true peak limit at all. The choice stays simple: master to -3 dBTP and every platform is covered, with headroom left for the MP3 encoders that raise inter-sample peaks downstream.
File Format and Bit Depth
All three platforms take MP3, but they differ on everything else: ACX uploads are MP3 192 kbps CBR only, Findaway strongly prefers WAV, Spotify prefers MP3 and also takes WAV and FLAC. The cleanest workflow is to master in 44.1 kHz / 16-bit WAV and encode per platform as the last step. Where a platform takes MP3, run the QC check on the encoded file, not the WAV master. Do not master to MP3.
QC Type
ACX and Spotify run automated checks. A file that misses the numbers gets bounced by a machine inside 24 hours. Findaway uses human QC, which means broader tolerance on numbers but stricter checking on perceptual things: mouth clicks, breath consistency, room tone matching across chapters, performance issues. A file that scrapes through ACX's loudness check by 0.5 dB will likely also pass Findaway, but Findaway is the platform most likely to flag a chapter that sounds inconsistent with the rest of the book.
The Single Master That Passes All Three
Master to this:
- Integrated loudness: -20 LUFS (±0.5 LU)
- True peak ceiling: -3 dBTP
- Noise floor: -60 dBFS or lower at every silence inside the file, not just the head
- Sample rate / depth: 44.1 kHz / 16-bit WAV
- Channels: stereo (dual mono is fine if you record mono; duplicate the channel)
- Head silence: 0.75 seconds
- Tail silence: 3 seconds
- One WAV file per chapter
That render goes to all three platforms as-is. No re-export, no re-encode, no per-platform variant. The conversions each platform needs (ACX's MP3, Spotify's streaming codec, Findaway's retailer-specific formats) happen on their side.
Common Rejection Patterns by Platform
| Platform | Most Common Rejection | What It Means |
|---|---|---|
| ACX | Noise floor above -60 dBFS mid-file | HVAC or breath in a silence segment. See our ACX guide |
| ACX | True peak above -3 dBTP | Limiter not in true-peak mode. See true peak guide |
| ACX | RMS outside -23 to -18 dB | Single chapter out of range, not the average |
| Findaway | Chapter-to-chapter inconsistency | Room tone or level drift across recording days |
| Findaway | Mouth noise / clicks | Editing pass missed transients human QC catches |
| Findaway | Metadata mismatch | Chapter titles, track numbers, ISBN inconsistencies |
| Spotify Audiobooks | Mixed mono and stereo files | The guide requires all files mono or all stereo, not a combination |
| Spotify Audiobooks | Credits and sample issues | Missing opening credits, or a retail sample over 5 minutes / containing music |
The Findaway Wide-Distribution Question
Findaway Voices distributes to retailers (Apple Books, Google Play, Kobo, Storytel) and library systems (OverDrive, hoopla) with their own internal specs. Findaway's job is to take your master and conform it to each retailer's requirements, which is why their preferred input is 44.1 kHz / 16-bit (or higher) WAV. They re-encode downstream. If you deliver MP3, they cannot recover the quality before sending to retailers who expect lossless input.
This is the main reason every modern audiobook workflow should master to WAV regardless of which platform is the primary destination. Even if ACX is the only platform you ship to today, a WAV master is the input every other platform will eventually ask for.
One Master, Two Common Variants
There are two reasons to deviate from the universal master:
- Spotify-direct delivery at hotter loudness. The Spotify for Authors guide allows RMS up to -14 dB, noticeably hotter than ACX's ceiling. Mastering hotter for a Spotify-only title is allowed, but playback normalization erases most of the perceived advantage and it costs you the ACX-compatible master. Rarely worth the second render.
- High-end retail (Apple Books premium). If you are delivering to a premium retail program that accepts 24-bit / 48 kHz, render a higher-res master alongside the universal one. The universal master still goes to ACX, Findaway and standard Spotify.
For everyone else: one master, three platforms, no re-render.
Pre-Delivery Checklist
- Every chapter measured individually: -20 LUFS ±0.5 LU
- True peak verified per chapter at -3 dBTP or lower
- Noise floor checked at multiple points inside each file, not just the head
- 44.1 kHz / 16-bit WAV confirmed in export dialog (CBR/VBR does not apply to WAV)
- Head silence 0.5-1 sec, tail silence 1-5 sec, consistent across all chapters
- Room tone consistent between chapters (see our room tone guide)
- One file per chapter, named consistently, under 120 minutes each
- Metadata complete and consistent (chapter title, track number, ISBN)
Why This Gets Hard at Scale
For a single book on a single platform, dialing in the spec and running manual checks is fine. For a catalog going to all three platforms, every chapter has to pass three different QC processes (two automated, one human) and a miss on any of them means a redelivery weeks after you thought the project was done.
Running every chapter through an automated check against the universal master spec, before any of the three platforms see it, removes the platform-specific failure modes. If the report says pass, the file goes to all three without modification. If it finds an issue, you fix one chapter against one spec instead of three.