Artificial Intelligence

All AI-powered voice platforms aren't created equal, here's why

While Siri, Alexa, and Cortana are the big players in voice recognition, smaller companies may have an edge. Here's how Voysis is helping businesses by providing a more customized experience.


Recent advances in voice recognition technology, in the form of virtual assistants like Siri, Alexa, and Cortana, are revolutionizing the way that humans interact with machines. But while Google, Amazon, and Microsoft have produced some of the most popular general speech recognition software available, it doesn't mean that smaller companies can't compete.A new voice AI platform called Voysis, whose distinguished architect, Ian Hodson, was Google's former head of global text to speech efforts—responsible for powering things like Google Assistant, Google Maps, and Android apps—has come onto the scene. Voysis gives enterprises an "option for a customizable and complete voice AI platform, giving them the ability to connect with their users in the most intuitive manner possible, online, on mobile, and on devices," said Hodson.

Using Alexa, for instance, which is trained on general speech, makes it "difficult to get those tailored in a very good way to another entity's product line or intentions," said Hodson. "A furniture store might have 10,000 items in your inventory, but it's a very specific set of items and with specific pronunciations, descriptions and spellings."

While a general assistant could attempt to figure out the meaning, it isn't going to be trained on the speech that matches this specific inventory, Hodson said. Also, the voice platform will "portray Google's or Amazon's personality, rather than their own branding which takes away from the feeling that you're speaking to that particular vendor or product," he added.

Instead, Voysis offers businesses the chance to train the AI on a company's specific data, or examples of the kinds of dialogue it wants it to have with its customers. Voysis also digs into the types of things specific customers are likely to say to their virtual assistant.

With the huge advances in voice recognition technology, Hodson said he sees fewer challenges to customer acceptance than there were 15 years ago. "Now that voice control and voice systems are out there and well deployed, people are very familiar with them," he said. "I think we're entering a market which is not mature, it's still early days, but it's already achieved acceptance with customers and with public."

Also, by using specific, vetted vocabulary content for domain training data, Hodson said he thinks that Voysis can circumvent some of the bias that can seep into machine learning algorithms that train on massive public data sets.

SEE: Why IBM's speech recognition breakthrough matters for AI and IoT

This kind of tailored voice platform is important for rising technologies such as the development of autonomous vehicles. It can help companies developing such cars consider the exact buttons or levers a driver might need to control, or the ways a driver might interact with a specific vehicle.

Although a voice-powered car is still years away, Voysis is already partnering with auto manufacturers on these vehicles, said Eric Bisceglia, vice president go-to-market for Voysis. "Our voice instruction and search capabilities can power a number of the specific automotive use cases, ranging from voice commands for driver assistance to GPS and even entertainment," he said.

voiceinput.jpg
Image: AH86, Getty Images/iStockphoto

"Soon your car will become closer in role to your home or office," said Bisceglia, "and just as voice assistants have become popular, they'll be the powerhouse of the car of the future. Gone will be steering wheels and shifters and instead we'll simply say: 'Bring me to the grocery store,' 'Park me at McDonalds,' or 'Take me to the beach.'"

In terms of the overall market, a company named MindMeld that develops a similar natural language technology was recently acquired by Cisco, said Bisceglia, which illustrates that these kinds of smaller brands are attractive to bigger companies.

"Our mission is to be good at whatever our customers need us to be good at," said Hodson. "There's differences in all of them that I think make them better for different use cases and different problems."

Also see...

About Hope Reese

Hope Reese is a Staff Writer for TechRepublic. She covers the intersection of technology and society, examining the people and ideas that transform how we live today.

Editor's Picks

Free Newsletters, In your Inbox