NEWS: Out of Stealth Mode, In China and Korea, Targeting U.S. Market Next: AI-Powered Speech Recognition for Kids
A software company that developed educational apps and toys for kids has launched B2B technology that enables any smart devices geared for children to recognize their speech patterns.
The company has also made it a priority to make sure privacy controls are plentiful and its technology is compliant with what’s known as the Children’s Online Privacy Protection Act. (This imposes certain requirements on operators of websites or online services directed to children under the age of 13.)
Kadho, with HQ in University Research Park in Irvine, launched its first product – a toy called Mochu – in 2016 in China. Mochu was designed to teach English to children in a way that adapts to each child’s pace. Mochu was developed in partnership with IBM. Kadho integrated its software with IBM’s artificial intelligence (AI) platform, known as Watson.
Kadho also created an app, Kadho English, which teaches English to children in China with interactive, video-based content. This app has since been expanded to include nine other languages, including French and Italian.
Over the last year-and-a-half, Kadho worked in stealth mode to respond to the parents buying its products for their children. These tech-savvy parents, who also use voice-recognition tools like Apple’s Siri and Amazon’s Alexa, complained to Kadho that existing speech-recognition technology did not go far enough to understand kids’ speech patterns, Kadho CEO Kaveh Azartash told OC Startups Now.
Kadho had amassed about a million users across the world, especially in Asia, Azartash said. So, it possessed a large reservoir of speech data based exclusively on kids. Using this data, it built an automatic speech recognition (ASR) specifically for kids – known as KidsSense.AI – to enable them to be recognized by various devices.
“We collect our data on phonetically-rich and -balanced data sets,” Azartash said. “So, it can be much faster and, with less data, can achieve performance scales that are better than (what’s already out there).”
Kadho acquired its first customers for this new tech, including LG (formerly known as Lucky-GoldStar) a multi-national company with several subsidiaries based in Seoul. It makes products, including electronics, chemicals, and telecom. It’s trying to expand into the kids market, Azartash said. Kadho also obtained several China-based robotics companies as customers, he added. Parts of Kadho’s team work out of Seoul and China.
This year, Kadho is targeting the U.S., going after leaders in the kids space, like Disney and Mattel.
Kadho saw revenue of approximately $800,000 for 2017, Azartash said. The company might open a Series Around this summer. To date, it’s raised about $2.4 million. Approximately $1 million came from angel investors, Azartash said. The rest came from a pre-Series A round from Chinese and Silicon Valley-based investors. They valued Kadho at $18 million upon closing.
KidsSense.AI offers both a cloud-based and offline product. Azartash said that KidsSense.AI does not collect any data from the children who are using it.
Competition includes the Echo Dot, introduced recently by Amazon. This piggybacks on Amazon’s original Alexa speaker, targeting children specifically.
Azartash said he sees the Echo Dot as “the biggest validation” for Kadho’s new tech.
“There is a need in the market,” he said. “From the time (children) start talking, until they mature linguistically at 11 or 12, a ‘one-fit’ technology doesn’t fit all.”
Azartash and Kadho co-founder Dhonam Pemba are both neuroscientists with doctorates from UCI.