HOUND Voice Search

Presented in June last year, Hound you have upgraded a few hours ago on iOS and Android entering the much-desired stable phase. For those of you who don’t know what we’re talking about, remember that Hound comes from colossus SoundHound, one of the most famous applications dedicated to listening and recognition, Shazam style music. The main objective of SoundHound Inc. is to create a system that allows to communicate naturally with PC, using normal sentences and without including any apparent specific command. This project has the name of Hound and, almost 10 years after its origin, appeared in the form of beta on Android and iOS in late spring 2015.

Unlike competitors such as Cortana, Siri and Google Now, Hound uses a voice recognition system much more Advanced, able to integrate in the same process as the semantic analysis phase with that of recognition of words spoken by the user, a process very similar to what we all maintain when we speak and we listen to a person. This allows the application to “listen” to the user, eliminating the waiting time since Hound can begin searches already at the moment when he understood the theme of phrase, narrowing gradually more and more fields are provided to new criteria. This process is referred to as speech to meaning and is opposed to the classic speech to text classically adopted so far.

But not only a powerful speech recognition system, but also a rich pool of partners include ACEs up his sleeve to Hound, which can access well 50 search domains against the 22 available to Siri. This makes Hound able to search increasingly complex thanks to the enormous amount of information that can be accessed. The footage that we propose below shows an example of all this potential into action: the user says a long sentence in which seeks “a 4 or 5 star hotel in Seattle for three nights from Friday between 50 and 250 $ per night,” adding later “with free WiFi and gym which are not of the B & B“. The system first offers an answer to the first question and then narrows the search fields on the same data previously provided, adding the new information provided by the user.

Now, quite a few updates and developments, the software came out of beta though it is still confined to the us market only. Unfortunate though it is clearly understandable as the English language is the only supported at the moment. It only remains to wait for further developments to understand if there is any possibility of a multilingual support.