Spotify for Authors currently accepts digitally narrated audiobooks for direct distribution on Spotify, but it does not send those files to partner retailers. That distinction is easy to miss when a marketplace simply says it accepts AI narration.
There are two different permissions in play. One lets a store generate narration from your ebook inside its system. The other lets you upload a finished synthetic-voice master made elsewhere, which gives you far more control over performance and reuse.
Treat store-specific AI delivery rules as part of production planning, not as paperwork saved for release week. A clean master can still be unusable at the retailer you wanted, while a store-generated version may not be reusable elsewhere.
This creates a split inside the same publishing dashboard. A human-narrated or author-narrated upload can use wider partner distribution, while a digital-voice upload needs a separate submission method for every non-Spotify destination.
Kobo also accepts an AI-narrated master through Writing Life. It asks you to identify the narrator contributor as Synthesized Voice, Female Synthesized Voice, or Male Synthesized Voice, turning disclosure into structured metadata rather than a line hidden in the description. Kobo does not demand exclusivity.
That metadata choice has commercial consequences. Kobo places audiobooks in both its individual-purchase catalog and token-based subscription program, and direct upload does not currently send the title into library distribution. If the audiobook tab is absent from an account, Kobo says access is still rolling out and can be requested.
The program does not accept arbitrary MP3 uploads. Google builds the audiobook from an eligible EPUB tied to an ebook offered on Google Play, so a finished external master does not enter through the same process.
Portability comes with two conditions that many distribution roundups omit. The auto-narrated edition must remain available on Google Play in every country where it is sold elsewhere, and its Google Play list price cannot exceed the price on those other stores. Using Google's free generation process therefore creates an ongoing availability and price-parity obligation.
Apple Books digital narration is also a production program, not a general promise that any synthetic file will pass review. Nominations go through preferred partners, the ebook must already be on Apple Books, the publisher must hold audio rights, and the current program requires English. Accepted titles appear as Narrated by Apple Books.
Apple's broader audiobook service still uses distributors for finished productions, but its digital-narration promise applies specifically to titles made through Apple's program. That difference prevents authors from treating the digital narration page as blanket permission for any AI file.
Apple says processing and quality checks take one to two months. The resulting edition distributes through Apple Books, while the author or publisher retains audiobook rights and may create another version elsewhere. That separation matters if you want one store-generated edition and a separately cast master for Spotify or Kobo.
An eligible KDP title displays an option to add an Audible audiobook with virtual voice. Amazon generates the audio from the ebook, provides editing controls, and publishes the result through Amazon and Audible. A finished AI file from an unrelated service cannot simply be substituted into that workflow.
The ebook must be live and eligible, have compatible formatting, and not already have an Audible audiobook. Rights holders cannot request access merely by owning an external AI master.
This limitation sits beside the changed economics of AI narration, but it answers a different publishing problem. Low generation cost matters only after the intended storefront accepts the production method, disclosure method, language, rights position, and file provenance.
File provenance should remain attached to each production record. Record who generated the edition, whether its files can leave the platform, which metadata identifies synthetic narration, and whether another retailer may accept that same asset. Those details determine whether one production can be reused widely or several separate editions are required.
There are two different permissions in play. One lets a store generate narration from your ebook inside its system. The other lets you upload a finished synthetic-voice master made elsewhere, which gives you far more control over performance and reuse.
Treat store-specific AI delivery rules as part of production planning, not as paperwork saved for release week. A clean master can still be unusable at the retailer you wanted, while a store-generated version may not be reusable elsewhere.
Upload permission separates Spotify from Kobo
Spotify offers the clearest direct option for a self-published AI audiobook. Its author platform accepts digital-voice files and publishes them on Spotify, yet the usual option to pass an uploaded title to Voices by INaudio does not apply to digitally narrated books. The upload is nonexclusive, so you remain free to deliver an eligible copy elsewhere yourself.This creates a split inside the same publishing dashboard. A human-narrated or author-narrated upload can use wider partner distribution, while a digital-voice upload needs a separate submission method for every non-Spotify destination.
Kobo also accepts an AI-narrated master through Writing Life. It asks you to identify the narrator contributor as Synthesized Voice, Female Synthesized Voice, or Male Synthesized Voice, turning disclosure into structured metadata rather than a line hidden in the description. Kobo does not demand exclusivity.
That metadata choice has commercial consequences. Kobo places audiobooks in both its individual-purchase catalog and token-based subscription program, and direct upload does not currently send the title into library distribution. If the audiobook tab is absent from an account, Kobo says access is still rolling out and can be requested.
Google and Apple generate narration inside their systems
Google Play Books uses a different model. You supply an eligible EPUB, confirm that you own the audio rights, and create the auto-narrated edition inside its Partner Center. After publishing on Google Play, you can download the audio files and sell them through another platform that accepts them.The program does not accept arbitrary MP3 uploads. Google builds the audiobook from an eligible EPUB tied to an ebook offered on Google Play, so a finished external master does not enter through the same process.
Portability comes with two conditions that many distribution roundups omit. The auto-narrated edition must remain available on Google Play in every country where it is sold elsewhere, and its Google Play list price cannot exceed the price on those other stores. Using Google's free generation process therefore creates an ongoing availability and price-parity obligation.
Apple Books digital narration is also a production program, not a general promise that any synthetic file will pass review. Nominations go through preferred partners, the ebook must already be on Apple Books, the publisher must hold audio rights, and the current program requires English. Accepted titles appear as Narrated by Apple Books.
Apple's broader audiobook service still uses distributors for finished productions, but its digital-narration promise applies specifically to titles made through Apple's program. That difference prevents authors from treating the digital narration page as blanket permission for any AI file.
Apple says processing and quality checks take one to two months. The resulting edition distributes through Apple Books, while the author or publisher retains audiobook rights and may create another version elsewhere. That separation matters if you want one store-generated edition and a separately cast master for Spotify or Kobo.
Audible permits synthetic narration only through authorization
Audible's policies make provenance even more important. Standard ACX submissions must be human-narrated unless ACX specifically authorizes another method, so technical compliance alone does not make an independently generated AI master eligible. Amazon's permitted synthetic process runs through its invite-only KDP Virtual Voice beta for qualifying ebooks.An eligible KDP title displays an option to add an Audible audiobook with virtual voice. Amazon generates the audio from the ebook, provides editing controls, and publishes the result through Amazon and Audible. A finished AI file from an unrelated service cannot simply be substituted into that workflow.
The ebook must be live and eligible, have compatible formatting, and not already have an Audible audiobook. Rights holders cannot request access merely by owning an external AI master.
This limitation sits beside the changed economics of AI narration, but it answers a different publishing problem. Low generation cost matters only after the intended storefront accepts the production method, disclosure method, language, rights position, and file provenance.
File provenance should remain attached to each production record. Record who generated the edition, whether its files can leave the platform, which metadata identifies synthetic narration, and whether another retailer may accept that same asset. Those details determine whether one production can be reused widely or several separate editions are required.
