Opinion: AI voices for audiobooks is a strange new trend


The AI voice can record a book in a fraction of the time as an actual human, at a fraction of the cost, which drops the price from more than US$10,000 (RM42,465) to US$2,000 (RM8,493) or less, Publishers Weekly reported. — Image by Racool_studio on Freepik

Siri, read me a story!

One of the joys of being read aloud to is listening to the voice of the person reading. If it's someone you know and love, the experience is filled with warmth and affection.

And if it's someone you don't know — say, a professional actor reading an audiobook — the voice still matters. Narrating an audiobook is a skill, and not an easy one; it takes about three years to master the art, says the Library of Congress.

"A narrator can ruin, make, or even enhance a book," Katherine A. Powers told me not long ago. Powers has reviewed audiobooks for the Washington Post for nearly 30 years and has listened to thousands of audiobooks.

"The best narrators, with a few exceptions, are actors," she said. "Amateur narrators... tend to have a dead, airless, one might even say 'robotic,' way of reading, or sloppy enunciation, or inept pacing, or something else distracting or awful."

Screeeech. Let's back up to that word "robotic." Robotic as in Siri, perhaps? As in Alexa? As in those mechanical voices that newspaper websites employ to read articles aloud and thus save you four minutes of your valuable time?

Yeah. Robotic like that.

Those automated voices are fine for something short, such as road directions or a newspaper article. But imagine listening for an entire book. Of course you now can, thanks to a development touted by Google and others - artificial intelligence (AI) voices for audiobooks.

Why? Money, of course. And time.

AI narration is inexpensive, allowing self-published writers to cheaply create an audiobook of their novel. The AI voice can record a book in a fraction of the time as an actual human, at a fraction of the cost, which drops the price from more than US$10,000 (RM42,465) to US$2,000 (RM8,493) or less, Publishers Weekly reported.

Cheap is good, if you're footing the bill yourself. But what about the listener?

"Based on the little I've listened to, AI text-to-voice often misses the right emphases, the right note," Powers said. "Not always, but enough to feel like there's sand in your salad. You begin listening for that - for the wrongness, rather than to the book."

Kimberly Wetherell is a member of the Professional Audiobook Narrators Association and, not surprisingly, she doesn't think AI voices are a good idea - for voice actors, audiobook listeners, or anyone. Her group is sending around a petition in protest.

Pushing AI narrators puts "cost over quality, at the expense of the listening reader's experience," she said.

Joseph Papke, supervisor of Minnesota Radio Talking Book (which records reading matter for sight-impaired people), said the most important characteristics in readers are correct pronunciation, followed by "diction, pacing and intonation.... It's a challenging test: only about a third of takers pass."

While he wouldn't rule out using AI voices someday, "I am of the opinion that this sort of technology should never replace capable human beings.... The 'less expensive' angle sounds like a dangerous race to the bottom."

A friend who does volunteer reading put it succinctly. "Above all," she said, "remember you are telling a story."

That might be the key. I don't want Siri to tell me a story, and I don't want Siri to know she's telling me a story. Am I old-fashioned? Refusing to accept our new robotic future? (I don't have a Roomba, either.) – Star Tribune/Tribune News Service

Follow us on our official WhatsApp channel for breaking news alerts and key updates!
   

Next In Tech News

EU forces Apple to also allow alternative app stores on iPads
TikTok blocks 37 million suspicious product listings from online shop
Google Podcasts, one of the most popular podcast apps, to end in June
Review: ‘Tales of Kenzera: Zau’ translates the journey of grief into a video game
Atos creditors reach deal to rescue debt-laden group, La Tribune says
In an online world, a new generation of protesters chooses anonymity
After two winsome Ori games, a pivot into dark fantasy
Teenager in China dies of heart attack after teacher forces her to exercise, insists illness is ‘fake’, delays first aid, enrages mainland social media
NoSpace is Gen Z’s answer to MySpace
What if customers were rewarded for tipping their meal delivery drivers?

Others Also Read