The company describes the technology as “speech-to-meaning,” and it first debuted with Hound, the company’s digital voice assistant, back in March. Our own tests showed Hound was able to pull up results much faster than Apple’s Siri, although it seemed to perform less accurately than Google Now.
Users just need to open the app and say the phrase “OK, Hound.” From there, you speak conversationally, and the Houndify backend technology should be able to understand the context and meaning of your queries. You can even make follow-up queries related to the first one.
“Everyone else uses a voice-to-text engine,” SoundHound general manager Katie McMahon told us in an interview last week. “The next step is for the machine to layer in an understanding of what you said.”
The problem with those voice-to-text services is the delay involved. Not only is the service processing what you’re looking for, but it’s also spending time converting your query to text. Since Houndify does this as you speak, the search process is much quicker.
“Hound is not only a little bit faster but it’s a step ahead of Alexa, OK Google, and Cortana,” McMahon argues, as you can do follow-up queries that build on previous ones.
Here’s a few examples of queries you can do with Hound:
“There’s no requirement for the user to learn 40 different skill-set trigger phases,” McMahon says. “We encourage our users to make their search phrases as if they were speaking to a friend.”
The app is now available for download through the App Store or Google Play.
Why this matters: While SoundHound is focusing on its own products first, the Houndify platform is being marketed to anyone willing to license it. Samsung is one of the first licensees; the company is incorporating it into its new Artik platform.
McMahon speculated that Samsung might also use the technology in its home appliances, allowing us to tell our washer to start a new load, or find out how much time is left before the dishwasher cycle completes.