SoapBox Labs Speech Technology

by Anastasia Deripaska

SoapBox Labs Caters to Kids with New Speech Technology

The purpose of Artificial Intelligence (AI) gizmos like Apple’s Siri, Microsoft’s Cortana, or Alexa by Amazon is to give everyday people the feeling of being immersed in the digital world, where everything they want is connected through their devices and the Internet of Things (IoT). Tabbed as perfect for use at home, the devices tend to have one major shortcoming: When kids talk, they don’t seem to have the faintest idea what is being said.

Different Voices, Different Technology

SoapBox Labs, a tech startup in Ireland, is seeking to offer kids the same type of immersive experience that their parents get from using those AI personal assistants.
They’re doing that by developing speech recognition tech designed just for children, the same type that is already being used in a host of educational apps and devices. The company is the brainchild of CEO and Founder Patricia Scanlon, who noticed her own device struggling to understand her children’s requests because they have different speech patterns, don’t always say words correctly, and have higher pitched voices.
“An industry that has spent decades working on technology and only focusing on adults runs into a fairly significant problem when they try to apply children to it," Scanlon said in an interview with CNN.

To compensate for the differences, SoapBox’s voice engine was built from the ground up and focused on kids from age two all the way up to 12, basically the age before most of them hit puberty and start experiencing changes in their voices that guide them into adulthood.
The engine was created by feeding thousands of hours of children’s speech into a machine-learning environment; including real-life noisy environments such as classrooms, cars, and cafeterias.
It wasn’t just for kids in Ireland either, but from 192 different countries and a wide range of dialects and accents.Rather than tweaking the available voice technology already on the market, SoapBox built its voice engine from scratch — focusing on children aged two to 12. The company says it created a data set to train its AI system, made up of thousands of hours of children's speech collected in real-world noisy environments — kitchens, classrooms and cars — from kids of all ages, accents and dialects, hailing from a total of 192 countries.Amazon at least has seen the need and launched Echo Dot Kids Edition in 2020. There are already 4 billion AI digital voice assistants in use as of 2020, with that number expected to double by 2024.
SoapBox is selling a technology rather than trying to compete with the established players.
By enhancing the technology and merging the AI with screens, it can create an immersive experience for children that blends learning, information, and creates the role of “helpful adult” that can assist with homework, set timers, make phone calls and lots of other functions that could be vital to a child over the course of their day.