Voice-based search is all the rage these days in the video world, as Microsoft, Google, and Samsung are all rushing to offer new ways for consumers to navigate through billions of videos online, without having to use awkward remote controls. There’s just one problem: Up until now, voice search has pretty much sucked. In many cases, natural language processing isn’t as robust as it should be, meaning you have to speak in some horrible, stilted way, just to get it to understand you. Video discovery specialist Veveo is seeking to change that, with voice search that actually just works.
Veveo has its background in helping users to quickly find the videos that they want to see: It provides predictive search and personalization technology that is used by a number of major pay TV providers on their set-top boxes. It’s also available on connected TV devices and mobile platforms. Altogether, more than 45 million TVs around the US use the technology, with customers like Comcast, Cablevision, Rogers, and DirecTV. Take into account mobile and connected TV applications, and Veveo tech touches more than 100 million different devices.
But for the most part, that technology just helped users search and navigate video options with traditional remote controls or — at best — touch-based keypads on smartphones and other mobile devices. It sped things up, by predicting what users would want to watch based upon a knowledge graph of previous searches. Veveo CMO Sam Vasisht told me in an interview that the majority of searches using its technology were completed in three clicks. But think about what you could do without any clicks — that’s the idea behind adding voice search into the mix.
To do so, Veveo has integrated voice and natural language capabilities for video search, discovery, and navigation through its SmartRelevance Conversational Platform. It works pretty much like Siri or other voice capabilities, except that it’s a lot smarter. Users can ask questions in actual English, not using the kind of stilted keywords that are necessary with talking to Microsoft Kinect, for instance. With that technology in place, Veveo’s customers — the big cable TV players and device manufacturers of the world — will be able to leverage voice in a way that they haven’t previously been able to.
To show off the technology, Veveo has built out a proof-of-concept app as a demonstration of the platform’s capabilities. I tested out the app and while not perfect, it’s a heck of a lot better than any other voice-enabled search that I’ve ever seen. The app allows users to search a variety of live, on-demand, and online video services, so whether they’re looking for live TV like sports matches, or searching for a particular movie on Netflix, Amazon, iTunes, or another online service, it can do so.
The platform has pretty robust capabilities for looking at metadata to find particular actors, directors, or genre when searching for a movie. And the platform even allows users to filter down through results. So someone could do a search for “James Bond,” the “Sean Connery” and then Netflix, to find only Sean Connery Bond films available through the streaming service. Or they can search for NBA games, and then highlight just those which are on that night and include a user’s favorite team.
Another key advantage is that the app will learn a users’ preferences over time, so that they don’t have to specify their interests each time they search. Of course, which services are actually available in third-party apps will depend entirely on what its cable or device manufacturer customers want to enable. But I look forward to seeing it deployed, as it could make voice search actually viable in a way that it isn’t today.