In the summer I try to write at least one post on the lighter side, so this month I want to share with you a short chapter in my longer journey to create audio versions of my novels and short stories.
To turn a 400 page novel like On The Edge into an audiobook by using an experienced human narrator could easily cost $10K or more, which I simply don’t have right now to invest for that purpose.
A couple of years ago I used an early version of GooglePlay to create a computer-read audio version of the book. It’s OK, maybe passable, but certainly not great. On the other hand, at a much lower price point than a human-narrated version, it does offer the listener a lot of audio time for not much cost.
Great strides have happened recently in the area of AI-generated voice text, particularly by ElevenLabs (www.elevenlabs.io). Their AI narrators now do a pretty good job of changing voice inflection and pacing, depending on the content. Again, not perfect, but much better, and more life-like.
And recently they’ve added the ability to track dialogues across an entire manuscript, assigning a different, consistent voice to each
character, along with the central narrator.
So now imagine something that would be astronomically expensive with humans: an audiobook with each character speaking in his or her own voice throughout the entire book. Sort of like actors around a mic on a sound stage producing an old school radio drama. And, yes, one can now even add sound effects!
On The Edge has over ninety speaking characters, some of whom are demons. To create a multi-voice audiobook will require curating and selecting each of those voices, with several trial runs to pick the very best ones for the main characters. Then generating and editing the audio version, which should include deleting a lot of “he said” and “she said” tags from the visual version.
In short, it should be fun, but a lot of work. On the other hand, such a product would be virtually impossible to create using human readers.
So, the question is, might anyone be interested in an audiobook with individual character voices? Will the voices enhance or detract from the experience?
I thought I’d try a smaller experiment, both to get the hang of the process, and to evaluate the finished product.
So I took a short story I wrote twenty years ago, “The Vote,” and turned it into an eighteen minute MP3 audio file. The subject matter is still relevant: Absolute vs. Relative Truth. And of the four characters, two are Australian and two are American.
If you’re interested in what AI-voice generation can now do, here is the link to your free gift of this audio short story: The Vote Audio Link.
I hope you enjoy it. And, of course, your feedback in the Comments, or via email (parker@parkerhudson.com) will be really helpful, as I contemplate moving from four to ninety characters!
If you might also enjoy learning more about this subject, ElevenLabs’ website is not the most intuitive I’ve ever encountered, and I strongly recommend a great course by Author Leanna Morgan that will jumpstart your engagement.
Finally, here’s a related tip for your family’s security. As you will see at the ElevenLabs website, it’s now very easy to clone anyone’s voice. So your family should have a secret verbal password known only to your members. Then if anyone receives a desperate phone call seeming to be from a child, parent or grandparent, and it sounds just like them, laying out an emergency situation for going somewhere or sending money to solve a problem, the recipient can ask the caller for the family password to prove if the situation is real, or instead a scam/hoax.
Enjoy the short story. And let me know what you think about multi-voiced audiobooks!
Comments