Song-recognition app company SoundHound launched Hound back in June as a Siri competitor, a verbal assistant able to process complex questions. They also quietly started letting users test out the beta version of Houndify, the voice activation and natural language processing technology that Hound uses, for companies to try out with their own products. Now after six months of tinkering, SoundHound is launching Houndify as a full platform.
Want Siri-style voice control in your app? Houndify’s API lets you do that, so you don’t have to come up with your own tech. When Hound launched, SoundHound founder and CEO Keyvan Mohajer said the company had been experimenting with voice recognition for almost a decade. Using Houndify to power Hound was an intentional move to show confidence in the technology, says Mohajer, and prove that they would only release a product that they’d use themselves.
He claims that Houndify’s all-in-one capability–“if you want a coffee machine that answers questions about the weather or the stock market”–stands out by allowing product makers to combine voice recognition, voice-to-text, and basic command and control to turn on and off via voice. In order to get the weather or stock tips, developers can plug into Houndify’s platform aspects. SoundHound got several companies to donate their API data to Houndify, including big names like Expedia and AccuWeather.
This makes the Houndify platform a bit of a marketplace. It is organized by “domains,” programs that provide users with relevant information or actions related to their queries–so Expedia would be the first domain under “travel.” Developers can build their own domains in the hopes that folks will choose theirs, netting the domain developer money should they decide to charge.
But Houndify needs to attract developers first to fill its domain marketplace. To entice them, SoundHound is releasing software development kits (SDKs) for a wide array of platforms: iOS, Android, Windows, Unix, Java, and Raspberry Pi, with more expected.
“Our mission can be summarized in two words: Houndify Everything. Voice interface will coexist with touch and keyboard, but more and more things won’t have a touch screen, and you can only talk to them,” says Mohajer.
Houndify’s price model is based on usage: you pay for the data requests your users make in the app that you’ve adapted. There’s a free tier that will mostly benefit tinkering developers whose testers are making very few requests, while the cheapest tier is $9 a month.
Once you bake Houndify into your app or service, it’s easy to update new actions and data into it. By switching on more domains, users can get access to new commands and data instantly.
“When we started to make Houndify, we came up with the following problem statement: We wanted a platform that is easy to learn and easy to use by developers who are not necessarily scientists in [the voice control and natural language processing] field,” says Mohajer, who adds that he’s hired new college grads with no experience in speech recognition or AI, but who started contributing to the platform within days.