Apple’s AI narration for fiction.

Published:

https://itunespartner.apple.com/books/articles/apple-books-digital-narration-3479

The voices sound surprisingly great! (Note that the page may be updated at any time, with new samples. But I guess the voices can only get better from here, no?)

The drawbacks are that

  • their processing time is extremely slow (1-2 months)
  • once the audiobook is up for sale, it cannot be taken down for 6 months
  • the copyright owners cannot determine the price (but they can determine the royalty amount they would like to get)
  • the copyright owners cannot listen to the audio prior to the audiobook going live. (!!!)
  • It doesn’t seem that the owners can download the file for sale elsewhere (Google Play Books allows this, so long as the audiobook is also sold at Google Play Books)

Still, I’ve submitted one of the short stories, Milk, to the audiobook creation process.


Presently, Google Play Books does not have AI voices for fiction. But the voices that Google does have can be edited at a fairly granular level (ex: pronunciations for fantasy names). I wonder how Apple will deal with pronunciations for words that do not exist… although, one of the info statements does say that if a book cannot be turned into a decent audiobook (and it sounds like there is some human intervention on Apple’s end), it won’t get published. But that “Nope, can’t be turned into an audiobook” decision may take 1-2 months!

Nevertheless, quite exciting, these voices. Yes. Indeed. What matters the most, at this stage, is the voices. They’re quite nice. Really. I am excited about what will happen in the next few years.