Kids’ play and entertainment has changed tremendously in the smartphone era. Children use apps, watch short video clips, and interact with their “toys” in ways that weren’t even possible a decade ago. And instead of just watching cartoons, kids want cartoons that talk back.
Oren Jacob, CEO of ToyTalk, is focused on developing that kind of next-generation interactive app. The company’s apps, such as SpeakaZoo and The Winston Show, present cartoon characters that can speak and react to what kids say, having a conversation of sorts with children.
“The expectations of how interactive the entertainment is are much richer now than they were when I was a kid,” explains Jacob, who worked in filmmaking for two decades at Pixar before founding ToyTalk.
The secret of the ToyTalk apps is a speech recognition program, called PullString, which works in the background to analyze children’s responses and select an appropriate response for the cartoon character. Everything the characters say is scripted in advance, but the conversation flows to respond to a child’s comments.
“It’s kind of like writing a screenplay where we do lines one, three and five and the child answers back lines two, four and six,” Jacob says. “I’m not in control of the final narrative; that’s as much in the child’s control as it is in mine.”
Jacob, who worked on popular kids movies including Toy Story and Finding Nemo at Pixar, still sees a place for huge Hollywood blockbusters aimed at kids even as the younger generation gravitates towards more interactive fare.
“There’s absolutely a place for linear story telling and a place for interactive entertainment,” Jacob says. “Those things are related, but they’re not the same. At least as we’ve discovered, coming from a film background as many of us at ToyTalk have, what it means to craft in this space interactively is different.”
Though ToyTalk’s speech recognition features are unique, the children’s app category is a vast and growing market. Sales just of “educational” game apps for kids reached $1.7 billion worldwide in 2013, heading to $2.4 billion by 2018, according to research from Ambient Insight.
Kids have to get parental permission to start using one of the ToyTalk apps. That’s in part because ToyTalk has to record what its child users are saying and send over the Internet to computers that run the speech recognition software (the same way Apple’s Siri and Google’s Google Now work).
Parents then have the added benefit of being able to listen to recordings of their kids talking to — and telling stories to — ToyTalk’s cartoon characters. Parents can delete the recordings or share them via email with Grandma and Grandpa. Some nominate their kids’ funniest chats as ToyTalk “editor’s picks,” highlighted on the company’s website for all to hear.
The company’s playful attitude extends even to the serious parts of its website. A team listing includes childhood snapshots of all the employees.
A new app coming next month — which, unlike the current offering, users will have to pay to download — will offer some more recognizable cartoons, such as the Loch Ness Monster, for kids to chat with. And conversations are ongoing with big entertainment companies that want to license ToyTalk’s technology for use with their famous — and copyrighted — children’s characters, Jacob says.
While scientists and technology companies have been working for decades to perfect computer recognition of adult speech, ToyTalk is at the forefront of newer efforts to analyze children.
“The field of children’s speech recognition is nascent — it didn’t even exist two years ago,” Jacob says.