If you’re a lover of Apple Books, there’s a new feature that might interest or horrify you: Apple has announced the launch of a new digital narration project to create audiobooks using AI. Since the audiobook market is exploding in popularity but only a fraction of books are actually converted to audio, Apple’s new technology offers a way to remove the “cost and complexity” associated with creating human voice recordings.
The first publicly available results of the project are already on the Apple Book Store. If you open the Books app and search for “AI narration” you’ll find several dozen fiction and romance titles labeled as “Narrated by Apple Books.” Tap that label and you’ll be informed whether the book will be read to you by “Madison,” the digital soprano used for these genres, or “Jackson,” the baritone equivalent. Both are said to be based on a real human narrator, but the audio has been generated using AI.
On Apple’s website, you can also listen to samples from “Helena” and “Mitchell,” who are the digital voices for the non-fiction and self-development genres, but these haven’t yet been used for any publicly available audiobooks.
Apple assures us that the voices are the result of “advanced speech synthesis technology with important work by teams of linguists, quality control specialists, and audio engineers,” but it’s clear that Madison and Jackson’s most appealing quality is their efficiency. Recording a human narration takes a matter of weeks, potentially costs thousands of dollars, and requires the presence of an experienced voice artist or celebrity (or, increasingly, the author), which is why so many books remain text-only. The AI approach vastly increases the cost-effectiveness of the process.
But is it actually effective? I listened to the first few pages of Shelter From The Storm by Kristen Ethridge, and was struck by how inhuman it sounded. “Madison” doesn’t quite sound like a person, and can’t convincingly adapt her speech patterns to the content of the text. There also isn’t the intangible human emotional quality that helps the listener to stay engaged. I continually drifted off, filtering the sound out as background noise.
The technology is no doubt improving and may be able to replace human voice actors within a few years, but it’s not there yet. It’s clear that automating the narration process will save a lot of money in the long tail of quickly produced, low-readership books, but it’s also obvious why Apple is being cautious about rolling this out for more high-profile titles, and about highlighting the move in general. As far as we’re aware, there has not been a standard-issue press release for the project, and Apple seems to be directing its publicity at authors and publishers more than media sites or the general public.
And those authors might not be so keen on the idea. As the Guardian notes, it’s a strategy that has to contend with criticism from plenty of stakeholders other than readers. An audiobook producer and a literary agent contacted by the paper each spoke of the value provided by high-quality narration and expressed a degree of skepticism about the project, but more serious reservations may come from political or legal entities. With this new service, Apple appears to be moving into the creation rather than simply the selling of audiobooks, which will open up a new front for accusations of anticompetitive behavior.
