Audible Just Drew the Production Tiers

On October 9 at the Frankfurt Book Fair, Audible's chief content officer Rachel Ghiazza told Publishers Weekly that the company's goal is that "every book is an audiobook in every language". PW reports that Audible expects to use AI tools across the entire production process, and that productions will range from "high-end multicast titles such as Harry Potter" to niche or long-form works that "maybe don't need the same production quality". Two days earlier, at the same fair, Spotify announced audiobooks in more than 180 markets by the end of the year and told Publishers Weekly that AI narration could bring titles to audio in languages where a human narrator and producers cost too much. The two platforms that used Frankfurt to go global described the same machine. Audible went one step further and named its tiers.

The tier a book lands in is set by what it is expected to earn, not by what the book is. For an independent author that has always meant the bottom tier by default, because the top one cost studio money. That is the number that has changed, and it is the only part of this week an author can act on.

The expansion is concrete. Audible will launch localised services in Indonesia, South Korea, Denmark, Norway and Finland in 2027, on top of the eleven marketplaces it announced at the London Book Fair in March. Read & Listen, which pairs an ebook with its audiobook in one app, is coming to Canada, France, Spain, Italy, Brazil, Japan and India, starting with English and German titles and adding local languages in 2027. Audible for Business is rolling out in the US. The company offers more than 1.5 million titles today. "Audio is the fastest-growing format in publishing, but the potential is so much higher," Ghiazza told PW.

Alongside the announcements, Audible presented The Future of Listening, a study it commissioned from NielsenIQ, polling 18,000 listeners in eleven countries: Australia, Brazil, Canada, France, Germany, Italy, Japan, Mexico, Spain, the UK and the US. The figures below are Audible's, from its release and PW's report; the company has not published the methodology.

Most countries were more open to AI narration for fiction than for nonfiction; Germany and Japan were the exceptions. Ghiazza called the approach "an and strategy", Publishing Perspectives reported: human performers remain, and production savings are meant to fund more titles. "Quality and transparency are critically important," she said. Her message to publishers still hesitating was shorter: "The time isn't coming to do this, the time is now."

Then the sentence that matters. PW's Ed Nawotka reports that Audible "expects to use AI tools across the entire production process", including an internal tool that "breaks down the book, it defines the character, it gives you the background" and handles scheduling, and that "productions will range from high-end multicast titles such as Harry Potter to niche or long-form works that 'maybe don't need the same production quality'". A car ride may call for an immersive listen, gardening for something lighter. Voracious audiences such as romance listeners "just want volume", Ghiazza said, and that range of production styles will now "grow globally".

Spotify's announcement takes audiobooks from 22 markets to more than 180 by the end of 2026, a potential audience of 750 million people. The rollout began October 7 in Colombia, Italy, Mexico and Spain; Argentina, Brazil, Norway and Poland follow on October 15 and Portugal on November 12. Each market launches with an English catalogue of more than 350,000 titles; Duncan Bruce, Spotify's director of audiobook partnerships and licensing, told PW the Spanish catalogue has more than 45,000 titles, Polish more than 20,000, Italian more than 14,000 and Portuguese more than 6,000, and that the catalogue will eventually span more than 120 languages. All publisher deals are non-exclusive. Audiobooks+, the add-on for extra listening hours, passed one million subscribers and $100 million in annual recurring revenue in just over a year.

PW reports that Bruce acknowledges audiobooks remain scarce in many languages, and that Spotify sees AI narration as a possible way to speed some titles from print to audio, "especially in languages where the market may be too small to justify the expense of a human narrator and producers". In his words: "AI done responsibly can give authors and publishers the opportunity to launch an audiobook in local languages, which they may have never done or had the financial ability to do."

Storytel said a version of the same thing on October 8. Its chief content and publishing officer, Helena Gustafsson, told PW the streamer has offered AI-generated narrator voices since 2025 and now automates translation so that "we can get an eight-book series out in just six months", because on a streaming service "if readers have to wait too long for the next book, they'll move on to something else".

None of this is unreasonable as catalogue economics. A platform producing toward every book in every language has to decide what each title gets, and the only variable it can see at that scale is expected return. A franchise earns a cast, a director and a score because it will pay them back. A long-tail title in a language with a few thousand audiobooks gets the production that costs least per finished hour, because that is how the catalogue grows fastest. "Romance listeners just want volume" is a statement about throughput, and it is probably an accurate description of how a catalogue team sees the genre.

Notice who decides, though. Under this logic a book's production is assigned by its sales forecast. A dialogue-heavy romantasy with six speaking parts is, as a text, a multicast book. As a catalogue entry from an author without a track record, it is a volume title, and it gets the volume production. The manuscript never enters the decision.

Both platforms described the plain automated read as how the long tail gets made, in 120 languages, on non-exclusive terms, with the platforms' own surveys saying half of listeners will try it. Whatever one thinks of that, it settles what the plain read is: the floor. It is no longer a way to stand out on a shelf, because by the end of 2026 it is the shelf.

What sits above it is already visible. Publishers have priced the full-cast edition as the flagship: Hachette is re-recording nine Witcher novels with a cast, music and sound design, and Audible cast its own Dracula separately in five languages. Audible's player is being rebuilt around the character, with a card for each performer behind each voice. And in the largest blinded test of its kind, a multi-cast production beat a single human narrator reading the same book on favourability, perceived quality and engagement. The top tier is where the attention is. The floor is where the volume is. Audible has now said in public which titles it expects to put where.

The platforms sort books by forecast. An author can sort by the manuscript. Read your own book for dialogue density and speakers per scene: not every title needs a cast, and a quiet literary memoir may be better served by one voice. Most dialogue-heavy genre fiction, the kind the volume tiers at Frankfurt were describing, is a multicast book on the page.

The reason that choice was never really available is the cost of the top tier. A studio multicast production with music and sound design runs $5,000 to $50,000 and months of scheduling, which is why the multicast tier has belonged to franchises. On Midsummerr a manuscript goes in and a cast, scored, mixed production comes out at $5 per thousand words, about $450 for a 90,000-word novel, in hours, with every line editable and the rights staying with the author. The same holds for a translated edition: Spotify switches on Brazil on October 15 and Portugal on November 12 with a Portuguese catalogue of about 6,000 titles, and a translation you already own can be produced in Portuguese, Spanish, Italian or Polish as a full cast rather than as the floor. Dracula, the novel Audible cast five times over, is free to stream here produced this way, if you want to hear the tier before you decide on it.

Where a produced edition is sold is a separate question from what it is. Audible does not accept third-party AI narration through ACX, and Spotify's own creation tool produces a single voice for US self-published English titles. This post is about the edition, not the storefront.

Three things cut against the argument above, and they belong in it. The survey is Audible's: commissioned by the company, presented by the company, methodology not published, and as Publishing Perspectives noted, a stated openness to AI narration is not purchasing behaviour. Audible's tiering is also a defensible way to run a 1.5-million-title catalogue, and "an and strategy" that keeps human performers for the titles that warrant them is a reasonable position, not a retreat. And we sell the top tier, so read our conclusion with that in mind.

The number in Audible's own data that argues for the human studio read is the narrator one: 65% of listeners seek out more books by a narrator they like. A performer with a following is a marketing asset, and no production method synthesises a following. A cast gives a book several voices to follow, but it does not give it a famous one. For a franchise, the studio tier buys both. For everyone else, the question Frankfurt leaves open is whether the production or the platform picks the tier, and for the first time the author has a say.

• The full-cast edition just became the premium edition: Hachette's Witcher programme and Audible's five-language Dracula.
• Audible's AI features are built around the cast: the player catching up with the format.
• Listeners don't prefer AI voices, they prefer a cast: the blinded test and the APA's falling willingness number.
• Dramatized audiobooks hold listeners to the end: the completion data.
• Portuguese audiobook production guide: the market Spotify switches on this month and next.