When we talk of smart devices driven by voice technology, most people think of Amazon's Alexa, Apple's Siri or Google Home. But thinking deeper, we come across a lot more use cases. What about voice messages on WhatsApp or voice search on Chrome? Every time we call a helpline number based on spoken responses, we are using the same technology. And this technology of making devices, apps or interfaces work by understanding the human voice has one major problem – it’s designed primarily with adults in mind.
Kid-specific voice recognition software
Irish tech start-up SoapBox Labs realised that kids needed a separate voice rcognition system. The company’s founder Patricia Scanlon noticed that children’s voices sound too different from adults, even to a machine. Kids have a higher pitch, and their language usage and speaking pattern also vary, making a robot read it as a different accent, or even worse, another language. So Scanlon and team set about researching and building a voice recognition software, specifically for children. Their technology was well-received and is already being used by manufacturers, from search and education apps to toys.
How did Soapbox build their software?
SoapBox decided to build the voice engine from scratch, focussing on children aged 2 to 12. The company collected children’s voice samples and created a data set to train its AI system based on these commands in kid voices. They had thousands of hours of children's speech collected in real-world noisy environments, kitchens, classrooms and cars, across the age group, in varying accents and dialects, spoken by children from a total of 192 countries.
How is Soapbox using their software?
SoapBox, is selling a technology rather than a consumer product. But given the immense demand for kid-related smart products, the company has raised more than $12 million in funding since its founding in 2013. SoapBox has been doing even better since 2020. They now have more than 50 clients from around the world. The company says it falls into two categories: ‘educate’ and ‘play’, which app manufacturers adapt according to usage.