In early November 2023, Amazon’s Kindle Digital Publishing (KDP) surprised the publishing industry by announcing that a beta program for creating audiobooks from eBooks was in the works called “Virtual Voice”. The new service would use AI (Artificial Intelligence) to “read” your KDP eBooks and quickly generate audiobooks which would then be published to Amazon’s Audible store. Left unsaid was that Amazon/KDP would have some sort of exclusivity program to list books produced by their services only in their stores and no other channels.
As with other initiatives from Amazon, their Virtual Voice program announcement has had a mixed reception, with opinions ranging from “revolutionary and game-changing” to “its-the-end-of-the-publishing-world” extremes. There’s no disputing that Amazon has had a huge impact on every part of the book industry, from retailers to publishers to readers. Since Amazon began selling books online in 1995, many bookstores and publishers have been put out of business (or have significantly scaled back their services). Meanwhile, since 2007 when KDP began, many authors who would’ve likely never been published can quickly and easily publish their own books in a variety of forms from eBooks to paperback to hardcovers. And what does it cost to publish with KDP? $0.00 per book.
For myself, the announcement of Virtual Voice couldn’t have come at a better time, as I was looking to publish several more audiobooks in 2024. Ideally I would publish all my books as audiobooks, but that would be a big undertaking. Two years ago, I used FindAway Voices to publish two audiobooks (“The Author of Life” and “The Seed of Haman”), though the process was cumbersome and expensive. Even with the royalty-sharing program in which the expenses are split between the author and narrator, the cost per book was at least $450 — and more if your book is longer. And that doesn’t include the time spent on producing audiobooks that could be better used elsewhere — like writing more books.
To traditionally produce an audiobook, you first have to find a narrator whose voice fits with your manuscript, interview them, hope they’ll want to narrate your book, and then contract their services. Depending on their schedule, you might get the audiobook proofs back for review within a few weeks up to several months. After receiving the proofs, you need to manually review the audio content, send back the corrections, wait for those to be re-produced, then finally approve the finished audio content when everything is correct. At minimum, in my experience this process will take at least 1-2 months. You can narrate and produce your own audiobooks, but then you would need to buy the equipment, record your material, and then find the time to professionally produce the material.
While the traditional audiobook process may be acceptable for one or two books (not to mention affordable), it isn’t for many more. In my case, I have over thirty books in my library — and more on the way. I simply can’t afford the overhead nor the time to produce more than a few audiobooks a year — at least not traditionally. As with my previous self-publishing experience, the new KDP program came along just in time, so my plans to publish more audiobooks was put on hold until Virtual Voice was ready.
By mid-December, I received the invitation to join the Virtual Voice beta and I dove in headfirst, with plans to generate as many audiobooks from my existing materials as possible. I was a bit skeptical at first, so I started with some of the smaller books and put it to the test. After playing around with their Virtual Voice Studio editor a bit and listening to the AI narrators/voices, it didn’t take long to publish the first audiobook and then I was off and running. At the start of the beta, there were four voices to choose from but now there are eight. And the cost for publishing an audiobook with KDP’s new Virtual Voice service? $0.00!
On the first day of the beta (after about 2-3 hours), I had fifteen audiobooks published and on their way to the Audible store. After resolving several issues with KDP, by mid-January all my eBooks had been published as audiobooks (33 in all). The main issues I encountered were how KDP processes the eBook’s Table of Contents, along with a quirk that was likely unique to one of my books. The bug was with a book I had written about the Bible character Job (pronounced like “robe” not “rob”). The KDP editor kept erroring when trying to replace the pronunciation, likely because it occurred over 1,000 times in the book. KDP ended up having to fix the code on their side to resolve the issue.
I found the quality of the AI narrators to be acceptable though not perfect, and it’s certainly better than Siri’s or Google’s voice used in their Maps mobile app. I expect the quality to improve over time as the technology matures and more authors use the Virtual Voice service. Machine Learning “learns” and improves as more data (and a variety of data) is added, and I would expect that Amazon/KDP will continue improving the voices and the narration over the upcoming months/years. Also, I would expect that other languages will soon be supported too.
While the AI narrators in Virtual Voice did sound a bit “artificial” at times, in my case the tradeoffs and benefits were certainly worth it. For myself, I’d rather have my books reach a larger audience at a low publishing cost immediately rather than wait several months or years (if at all). Also, with a niche readership, it doesn’t make sense to spend hundreds of dollars per book and never recoup the production costs. If I decide to traditionally publish another audiobook, I can always de-list it from KDP and then re-distribute the new version made through another service.
All in all, I’ve had a great experience with KDP’s Virtual Voice program and am looking forward to seeing how it grows and evolves as more authors join it.
For a more in-depth analysis on Virtual Voice and its potential impact on the audiobook world, I found this article most helpful:
Being Refactored 