Home / Blog / Embedded Barbie

Embedded Barbie

My daughters are well beyond Barbie age, but a recent article in the NY Times (“Barbie Wants to Get to Know Your Child”) still managed to catch my eye. Barbie, it seems, is getting smart – at least smarter than she was twenty years back when she could utter a few lines like, “Math class is tough!” Thanks to AI, speech recognition, and embedded technology, Hello Barbie – which is the name Mattel has given her – will soon be intelligent, albeit artificially so.

What’s inside Barbie is, among other things, a rechargeable battery in each thigh. In fact, they had to widen the thighs a bit to make room for the batteries. There’s also a “mini­USB charging port…tucked into the small of her back.” And, as the article’s author found when he got a look inside Mattel while the new Barbie was being developed:

“A microphone, concealed inside Barbie’s necklace, could be activated only when a user pushed and held down her belt buckle. Each time, whatever someone said to Barbie would be recorded and transmitted via Wi­Fi to the computer servers of ToyTalk [an AI company]. Speech­recognition software would then convert the audio signal into a text file, which would be analyzed. The correct response would be chosen from thousands of lines scripted by ToyTalk and Mattel writers and pushed to Hello Barbie for playback — all in less than a second.” (Source: NY Times)

Apparently, given the form factor that the toy designers were working with, most of what’s on the inside had to be custom built. Critical Link was not asked for any help. It doesn’t sounds like they’re actually doing much of any processing within Barbie herself, just doing some recording and transmitting. If they do want to make Barbie any smarter, we can create a System on Module that’s pretty darned small. (I’m trying to remember just how big Barbie is. I’m sure I stepped on more than one back in the day.)

What’s happening on the processing end is occurring on Amazon’s S3, the platform that ToyTalk uses.

“The company’s AI and speech technology are written in C++ with iOS and Android clients built on top. Its desktop authoring software is written in Python and PyQt. The server code is written in Go, and ToyTalk provides a RESTful Web API to build Web-based conversational clients.” (Source: Tech World News)

Even though Hello Barbie isn’t expected to be released until the holiday season, it’s already inspiring some pushback. Much of it is around the potential for the recording technology to be misused, given its potential for invading the privacy of the children holding conversations with their dolls. (Which, of course, they’ve done for years. It’s just that most dolls haven’t had all that much to say back.) Pretty much a 21st century problem, I’d say.

In any case, as I mentioned, I’m out of the Barbie fray. Just thought that coming up with a Siri or Cortana for Barbie-lovers is an interesting use of technology.