Greg Britton is the editorial director at Johns Hopkins University Press. He is also a part-time professor of the practice in scholarly publishing at George Washington University. As my (and Eddie Maloney’s) editor at JHU Press, it was Greg I had in mind when I wrote a piece asking “Should University Presses Use AI Narration for Absent Audiobooks?” Gracious and generous as always, Greg has agreed to explore the issue of digital narration in this space.
Q: How are you thinking about AI audiobook narration for JHU Press? What are the issues, challenges and likely future directions?
A: As someone who lives in the world of books—in their multiple formats—it’s also frustrating to me that more titles aren’t available in audio. As you point out, audiobooks can be a convenient way to read while moving through our daily lives. We can listen while we commute, exercise, cook or stand in line at the grocery store. Audiobooks also make their content accessible for neurodiverse readers and those with visual impairments.
Given these multiple affordances, why aren’t all books available in audio? The reasons have to do with systemic factors in publishing. Because of the expense of making an audiobook—studio, equipment and voice talent, plus a shopping cart and delivery infrastructure—most midsize and smaller publishers do not publish their own audiobooks. Instead, they license the rights to a third-party publisher who produces the audio file. Those audiobook publishers, in turn, sell their .mp3 files on online sites like Audible.com. Audible, by the way, is owned by Amazon. The original publisher earns a licensing fee for the content, and they split that fee with the book’s author. (Read your book contracts, people!)
Although a publisher may want all their books available in an audio format, the decision isn’t up to them. Those third-party audiobook publishers are selecting only the books they believe can be published at a profit. The decision is a market one. So, as much as we would love to deliver all our scholarly and trade books in an audio format, market forces prevent it.
You’d expect that AI would disrupt this by dramatically lowering the cost of producing those audio files, but up to now, it hasn’t. Why? The major distributors of these books—Audible, for one—have prohibited non-human-voiced books on their platforms because of the low robotic quality of early text-to-voice software. This has kept the cost high and the access to audiobooks relatively low.
It has also protected, for now, anyway, the jobs of voice actors who produce thousands of audiobooks annually. [See Katherine A. Powers’ Aug. 23 piece in The Washington Post, “AI Is Coming For Your Audiobooks. You’re Right To Be Worried.”]
It’s important to note that Amazon (through their ACX subsidiary) have created an avenue for authors, self-published and otherwise, to hire a producer to create an audiobook for them or even let them narrate their own book. I have heard of a few examples of university presses reverting audio rights to authors wishing to pursue this. It’s important to note that the expense of this work falls to the author, and files are only available for purchase on Amazon.
As text-to-voice technologies improve, we might anticipate all this changing. As you say, Apple Digital Narration provides a great example of the state of the technology. It definitely passes the Turing test for me.
It also seems like Apple is willing to host those AI voices on their iTunes store. I suspect it is only a matter of time before Audible raises their prohibition, and that might not be so bad.
Why? The vast majority of the work of scholarly presses and academic journals simply isn’t available in audio form because of the current market forces at play. Allowing AI tools to lend a hand would make this scholarly content much more accessible. Imagine taking this a step further by pairing these AI audio tools with the growing catalog of open-access books and journals. This content could be available beyond measure. Imagine the impact!
As America’s oldest university press, Johns Hopkins has taken accessibility seriously since its founding. Daniel Coit Gilman, Hopkins’s first president, thought publishing wasn’t just the necessary business of a great research university but that it was at the very heart of higher education’s mission—to spread ideas beyond the walls of the classroom and beyond the gates of the university. We are listening.