Hey there, lovely reader!
I've been studying and experimenting with artificial intelligence for months now as I think about when, how, and if I want to use it. What AI can do is amazing, shocking, and mediocre all at the same time. Undoubtedly it will get better.
Is AI exciting? Yes. Is it terrifying? Yes. Is it exploitative? It can be.
(Fun fact: Back in the late 1980s, I worked for a technology company which pioneered the use of artificial intelligence. It is not a new concept to me. I shared a cubicle with people who talked about it all day long.)
Today we have a look at what AI can do as it relates to audiobooks, a promised bonus, and a request. For those of you are interested, I have a progress report on that “impossible” puzzle I mentioned last week.
Ai, ai, ai, audio
When I was in New York a few months ago, my husband and I were on the top deck of a double decker sightseeing bus. It was a sunny day and we were enjoying the sights as we were driven through the streets of the city.?
As we went along we each wore a pair of earbuds, through which a commentary relayed to us details of the sights as we passed. Mid-journey the commentary changed. I turned to my husband and told him, “This is AI generated. It's not a real person.” ?
Now, my husband is a Silicon Valley tech person. All day he studies highly technical computer things and talks about them (to other people, thankfully.) He had no idea he was listening to a robot. But I noticed it immediately. ?
There are tells. Once they have been pointed out, you can't unhear them. But it may take a while for you to notice.??
A while back, I cloned a narrator's voice. It was dead easy if you know the tool to use. The match was *perfect*. For a few words the AI generated copy was indistinguishable from the human one.
But, and herein lies the rub: In my opinion, AI audio doesn't work over the long haul. I have no plans to employ it. Here's why:
AI narration will replicate one voice perfectly. But it won't speak in voices/accents without lots of human intervention to chop up files and bits of them together making the process not much less intensive than it currently is.
There is no change in speed, cadence, rhythm. Everything is spoken at one speed and inflection. AI will not recognize certain words and pronounce them incorrectly, and it will occasionally put emphasis on the wrong syllable. These issues pull a listener out of the story, something a good writer strives (and sometimes strains) to avoid.
But there's more.
I have no evidence for this except my personal experience, but I suspect that the very slight robotic nature of an AI voice loses the brain's attention over time. It might be able to capture someone's focus for a few minutes like the commentary on the sightseeing bus, but over hours of listening to an audiobook, I theorize the brain loses the ability to engage. I have spoken to audiobook listeners who think AI narration is very good (and it is – as a technical achievement,) but I question its use over hours and at scale for many people. ?
Lastly, and perhaps my most vociferous objection to AI generated narration is the impact on the humans whom it replaces. When I started to write books, I thought I would make a modest living for myself. I did not anticipate that one of the biggest rewards would be the ability to employ people – to help others support themselves and their families through the work I was able to give them. ?
The notion that my brain concocts stories, that people enjoy, that generate income, which I can pay forward by giving work to other people gives me an enormous amount of pleasure. I call it the “Alison economy” and if you've ever bought one of my books you are part of it. At a stroke, AI audio puts highly trained and skilled narrators out of work and replaces them with an inferior product. And that way, I think we all lose.
I'm not against AI narration per se. It is an amazing technical achievement and think there are some uses for which it would be beneficial. The sales of my audiobooks in Norway for example would not justify the cost of a Norwegian voice actor and maybe in time, AI will be able to produce a decent enough version. (That time is not now, I've tried.) ??
But I have no plans to replace real, experienced human actors to narrate my audiobooks. Quality is always my north star, even if I have to wait longer, charge more, or work harder. For me, humans narrating to other humans like they have for millenia is the only way to go. ⭐
If you would like to hear real humans narrate the Annabelle and Inspector Graham books, start here.
Follow me, please
If you read this newsletter infrequently or would like an extra nudge when I release new books, did you know you can follow me on Amazon? If you do that, you'll:
- Get updates on new book releases and special offers.
- Receive personalized recommendations: book recommendations based on your interests and this author's genre.
You can follow me there by clicking on the button (see image above) that sits just under the book cover. (At least on my screen it does, it might be somewhere else on your device so look around for it.) Good job! ➡️
As you can see I didn't give up. I did abandon doing the edges for now. I'll do those at the end. The key was to start with some colour and work to fit pieces only where two pieces are already joined together creating a “corner.” It's not the easiest puzzle, but I've done harder. It is the only puzzle that I can't fit in the middle section of my table – the left and right edges spill over into the sorting sections, but I'll make it work. ?
I'll be back next week with more updates. ?️
Happy reading!
