The second annual Voice Search Conference was held in San Diego last week, organized by the non-profit Applied Voice Input Output Society (AVIOS) and Bill Meisel (president, TMA Associates, editor, Speech Strategy News), and like last year, contact centers were a big part of the session content, as was mobility like last year's event. However, unified communications has emerged as a strong secondary theme at the conference. It was mentioned in numerous sessions and was the main topic in at least one of them.One of the newer topics to come up during the conference that we are discussing in the UC community is the addition of social networking to UC and mobility applications. In the speech technology vein, this means having access to Twitter feeds, Facebook, etc., using voice either to add or access information from those sites. Indeed, during the conference I watched one of my colleagues use her iPhone and Vlingo to speak her status into Facebook; something my Facebook-addicted self is sure to try soon (I just downloaded it).
Many speakers discussed what they perceive as the big user interface challenges they are seeing as new devices, such as the iPhone, come to market. One of the best talks at the show was Gary Wright of Applied Speech Resources, speaking on "A Year Using Voice Search," in which he chronicled his intentional and unintentional use of voice search to see what the pitfalls and the promises of voice search are over the course of a year. He spoke about new devices coming to market, and some of the phone-based applications used for search today, including 1-800-GOOG-411; Google's speech enabled search application, and multimodal applications, such as Vlingo.
After the year was up some of his observations were that he used directory assistance a lot more when it's free, it was great being able to get directions, phone numbers, addresses and maps, but that as a user you don't always want to be connected to a number right away. He now uses phone-based applications more, but mostly while driving, and he feels that there is a lot more information that people might want that just isn't available yet. Gary also contrasted the use of the speaking on the phone versus multimodal, and had a lot to say about the user interface. As an example, he said that it was extremely irritating to do a Google search for a business on his iPhone, have an IM window pop up during the search, and then have to deal with the IM first because the window won't go away. More frustrating was during the process that meant leaving the search window to get to IM, and then having to start over, only to find that Google reorders the search so you don't get the same business listed first the second time. He had many examples that really pointed out that we have a long way to go to iron out the UI on some of these applications and phones.
In all it was stressed that application/user interface design should be adaptive to the user and shouldn't put a cognitive load on the user. We have seen this issue in IVR application development for years, such as not giving the caller an extensive list to choose from, and this type of design issue is now surfacing in a big way with multimodal applications. Many speakers brought up the fact that the use of mobile devices are becoming much more like using the desktop, and discussed the challenges this brings.
Another side of voice search, besides the mining of voice calls in contact centers or searching on mobile phones is searching for audio in audio and video files, and searching for strings within text. To this end there were some fantastic demonstrations on using a search box to search for keywords, names or phrases in news broadcasts. It was almost instantaneous. As we start to add video and audio to our business interactions, the ability to effectively mine content is going to make those applications so much more valuable.
Hopefully next year's conference will be in the same sunny location; making it all the more attractive for UC vendors and prospects to attend.
