This Smart Speaker Will Read Stories to Your Kids With a Deepfaked Copy of Your Own Voice

This Smart Speaker Will Read Stories to Your Kids With a Deepfaked Copy of Your Own Voice

You stop judging all the shortcuts and workarounds that make parenting easier once you actually become one, like Takara Tomy’s new AI-powered smart speaker that can imitate a parent’s voice at bedtime and read those children’s books and fairy tales most parents are sick of reading over and over.

For a few years now, we’ve been aware of the threat that deepfake technology poses, particularly when used to imitate someone in a position of power or influence. In 2019, scammers successfully used a deepfake voice copy of an energy firm’s CEO to fool an employee into transferring $243,000 USD ($337,333) to them, and in 2021 the stakes were raised even higher when an AI-cloned voice was used as part of a $US35 ($49) million heist. There’s good reason to be mildly concerned, but Takara Tomy is here to remind us that any technology can be used for nefarious purposes, and there’s still some good that can come from a device that’s able to imitate a parent’s voice.

Coemo is an adorable smart speaker that looks like a caricatured speech bubble with a glowing face. Instead of answering questions about the weather or incorrectly setting kitchen timers when asked, Coemo is designed for entertainment purposes only, and exclusively children’s entertainment. It comes with 60 different stories and songs from around the world, including Grimms’ fairytales, nursery rhymes, Japanese folktales, and songs.

The smart speaker’s other neat trick, and the whole reason we discussed deepfakes earlier, is that it doesn’t just let parents change the voice used to read all of the content it comes pre-loaded with. It also lets them create AI copies of their own voices, so Coemo can serve as their bedside avatar. Using an accompanying iOS app (an Android version is promised soon), parents or caregivers go through a process to train the AI model, which involves reading through a specific script for about 15 minutes.

The resulting deepfaked voice won’t be a perfect copy, but is supposed to a close match that emulates the way the parent speaks, including their unique intonations and inflections. The AI model that the smart speaker creates presumably can’t be used as part of a money-stealing scam, but Takara Tomy claims it will be close enough to help a child relax next to the familiar voice of a parent or a loved one, as multiple voice profiles can be created and selected as needed.

Powered by three AA batteries, Coemo will be available for pre-order starting on June 14 for $US105 ($146), and in addition to the stories and songs, it includes sound effects and even relaxation routines designed to help kids fall asleep easier. And through the mobile app, Takara Tomy plans to introduce additional purchasable content at a later date tailored to specific age groups, allowing the device to adapt to a child’s tastes as they grow.

The Cheapest NBN 50 Plans

It’s the most popular NBN speed in Australia for a reason. Here are the cheapest plans available.

At Gizmodo, we independently select and write about stuff we love and think you'll like too. We have affiliate and advertising partnerships, which means we may collect a share of sales or other compensation from the links on this page. BTW – prices are accurate and items in stock at the time of posting.