Monday, February 27, 2023

Apple AI Audiobook Narrators by L.A. Kelley


Audiobooks are nerve-wracking to produce by yourself. Sure, you can offer a shared royalty route with the narrator and hope to find a good one. But a narrator search can be difficult, frustrating, and time-consuming. Not to mention, the entire production process is  daunting and expensive. Where to begin?

According to Apple, not with a human. Apple started production of digital audiobooks using artificial intelligence, specifically two AI narrators named Madison, a female soprano, and Jackson, a baritone. The platform uses advanced speech synthesis technology combined with input from linguists, quality control specialists, and audio engineers. Currently Madison and Jackson are used for fiction and romance only. In the near future, Apple will have two additional digital voices, Helena and Mitchell, for nonfiction and self-development audiobooks.

 If you’d like to hear a sample of Madison click here.

 According to Apple, these are the benefits:

·        The audiobooks will be easy to produce and delivered via preferred partners. The original ebook must be created in either Draft2Digital or Ingram CoreSource.

·        Audiobooks have wholesale price limits. (The website isn’t clear on what they’ll be.)

·        Distribution will be solely via Apple Books and to public/academic libraries.

·        Publisher/author retains audiobook rights, and there are no restrictions on producing and distributing other versions of the audiobook.

How do you make an Apple audiobook?

The ebook must be created with either Draft2Digital or Ingram CoreSouce. Then the author selects the title. Apple has a review process and acceptance isn’t guaranteed. The general requirements are as follows:

·        Ebook must be available on Apple Books.

·        Author must own the audio production rights.

·        Primary category must be romance or fiction. The only subcategories currently accepted are literary, historical, or women’s fiction.

·        Book must be in English.

How good are AI narrators?

Frankly, Madison is fine. A lousy narrator can ruin a good audiobook and I stop listening when any set my teeth on edge. Madison has a pleasant voice with good tone, if a trifle unemotional. If you listen to her, you understand why Apple limits the AI to audiobooks without an exciting chase scene or hilarious denouement. An AI can’t make handle extreme ranges of emotion. Other things AIs can’t do:

·        A multitude of characters in the same book. Their voice is typical of a national newscaster; nothing to determine a regional accents. Nor can an AI handle a stutter, quirky word pronunciations, foreign words and phrases, or fictional words. (Sorry, Samuel Clemens, J. R. R. Tolkien, or me. Our work doesn't qualify.)

·        Difference in ages. All the characters sound roughly the same age.

·        Difference in genders. Male and female characters speak with a similar tone.

 So what’s the big problem with AI?

Even if an AI sounds okay, I have issues. A top-drawer narrator makes an audiobook memorable, and AI has a long way to go to reach that level. Audiobooks are also an art, and frankly, I’m not crazy about the idea of a soulless digital character taking a job from a human who spent years honing a craft. Not to mention, a human narrator adds emotional nuance to every page, something an AI can’t do. AI narrators are still few and far between and until I have no choice, I’ll opt for a human every time.


L. A. Kelley writes sci-fy adventures with humor, romance and a touch of sass. Her intelligence isn't artificial.




 

 

 

6 comments:

Keri Kruspe said...

I love how you outline the audio pool yet, much less us an AI narrator. I appreciate what you outlined here because I'm not opposed to use other AI tools like Midjourney and Sudowrite. Which I'm considering writing about in next month's article....

Jessica E. Subject said...

I don't like YouTube videos narrated by AI. I turn the volume to mute and read the subtitles. So, I can't imagine listening to an entire book read by AI.

Nancy Gideon said...

I've had audio books done by a publisher and have had to go through the selection of voices and editing process a number of times. Some of the errors of hilarious to listen to - but not as it applies to your listening audience. They tend to put me to sleep because I have nothing to look at while I listen. On the up side, they sold VERY well.

Maureen said...

I don't listen to many audiobooks, but I do use the computer narrator to review my freelance work and fiction drafts. I can't imagine listening to a book for enjoyment with that voice, lol.

Diane Burton said...

Great topic. Thanks for the pros and cons. AI voices can be very irritating.

Nightingale said...

Like Maureen, I listen to the computer narrator in the final read-through. I'd much prefer a human narrator, but I don't think I'd turn my book over to an AI narrator for an audiobook. I had a bad narrator on two audiobooks, and it ruined the entire book.