Why I won't be replacing Terry with an AI
In which I talk about Amazon's recent announcement of AI narrators for audiobooks
Amazon seems to be having a bit of an on-again-off-again relationship with AI.
It loves hosting AI through its AWS hosting service, probably because it loves money. It loves using AI to power things like its omnipresent ever-listening Alexa, probably because it loves even more money. It doesn’t like it choking up the Kindle self-published ecosystem with AI-generated junk, because that costs it money, and so you now have to cross-your-heart-and-hope-to-die promise Amazon that you didn’t use any AI when you upload a new book to KDP.
(Actually, you just have to tell them if you used AI or not, which is kind of a different thing…)
This month though, Amazon decided they also liked AI for something else. Amazon wants AI to narrate audiobooks.
Honestly… I’m in two minds about this.
The good bit
A lot of books aren’t available as audiobooks and a lot of authors, particularly indie authors, don’t have the resources to get their books professionally narrated or record their own audiobooks.
In this context, AI narration is a good thing - it’s going to open up opportunities for authors and make more content available for people who rely on audiobooks and narration to access content.
More opportunities and more content… that has to be good, right?
The bad bit(s)
Firstly, there’s the obvious risk that AI narration is going to put good narrators out of business. Publishers who maybe don’t care quite so much about the quality of the narration are going to stop paying human beings to read books and let the machines do it instead.
Secondly, if you’ve heard any AI-generated speech recently, you’ll know how emotionless and empty it sounds. TikTok is full of mass-produced videos with synthetic speech content and… they’re awful. There are tools out there that allow you to inject emotions into the text, prompting the AI on where to use what tone, but this is a time-consuming process and still doesn’t produce a great result.
We’re terrible though, and so can’t have nice things.
Analytics however will back publishers who want to take an AI-driven approach. Those videos on TikTok you hate because of the emotionless, robotic audio? They’re as viral as the Outbreak monkey with a freshly minted passport. They get clicks, and clicks are what it’s all about.
Did you know that a lot of audiobooks are listened to by people at 1.5X speed and above? I don’t know why, but a lot of people prefer it that way.
It’s maybe because audiobooks are read at 150-160 words per minute, but we read at around 300 words per minute. That’s not my theory, I found that here - https://www.insidehighered.com/blogs/technology-and-learning/my-half-baked-hypothesis-about-audiobooks-and-reading-speed.
(My theory is that some people are just maniacs).
Whatever the reason, the crisp and efficient output of an AI is going to do just as good a job, if not better, than a human narrator for people who listen at speed. Those AI-narrated books are going to sell and they are not going to pay out royalties to those pesky humans.
There may be no way to stop the tide of robot-read books.
Why I’ll stick with Terry
When it came to producing the audiobook of “Welcome to Neverbury”, I was lucky. Very lucky. With minimal coercion and only mild hypnosis, the extremely talented Terry Cooper “volunteered” his services as the narrator for the audiobook.
I said “yes” immediately, as I know how incredibly talented Terry is as an audio performer. I was not disappointed.
Terry understands the stories in Welcome to Neverbury. He gets it, he feels it, and that resonates in every word. He’s also an exceptionally talented actor with a huge vocal range, so every character in Neverbury is blessed with their own distinctive voice.
There was a point during recording that Terry actually started to leave me messages in the voice of Mabel from the story “Open House”. If you know that character, then you’ll know how deeply disturbing that will have been!
Terry was also an amazing collaborator in terms of giving the stories a final polish; pointing out sentences that didn’t work and tiny errors that had slipped past my proofreader.
I know that an AI could not have done what Terry did with this book.
For me, the point of an audiobook is not just to turn text into speech. It’s a performance, a piece of art in its own right, interpreting and hopefully elevating the original. AI isn’t going to do this. AI is probably never going to do this, at least not in the mass-produced way that Amazon seems to intend.
No, when it comes to audio versions of Neverbury… I’ll stick with Terry.
If you’d like to give the audiobook of Welcome to Neverbury a try, you can pick it up from Audible: https://www.audible.co.uk/pd/Welcome-to-Neverbury-Audiobook/B0CDNK2D3K?action_code=ASSGB149080119000H&share_location=pdp
Aww ya big softy. (I'm available for all kinds of work, birthdays, weddings EXCEPT Bar Mitzvahs. I can't pronounce the Yiddish words correctly.