Friday 18 November 2011

Speaking of Technology

So my wife now has an upgraded phone, the iPhone 3GS (my old one) and I have my new iPhone 4S. So we could move to iOS5 together. One of the unique features of iOS5 on the iPhone 4S is Siri. Siri is not just voice recognition; it is contextual language understanding. Voice recognition on most other devices relies on you speaking particular words or sometimes phrases, which are recognised, matched and cause actions to be carried out. With Siri, you don't really have to think what the phrase is that you have to use; you simply say what you want. And you can usually rely on Siri using previous voice commands when interpreting subsequent ones. For example, if you ask something about weather for one location, you do not have to mention weather again when you want that information for another location; you simply say "and in London?".

Perhaps the most impressive aspect of Siri was demonstrated the first time I ever tried it. After removing it from the box, and the SIM card was inserted, I instantly loaded my mail, address book, and calendar information via MobileMe. I then used Siri for the first time by saying "Call my wife". I expected it to ask who my wife is, at least the first time. But instead my wife's phone rang! The only way I can think that this was possible was from multiple lookups to my Address Book. My own entry in it gives my spouse's first name. There are two records in my address book that have that first name, only one of which is my wife. A match between surname of that person and me would have correctly led Siri to the correct person. The mobile number in that person's address book record was then selected from the three numbers listed for her. It's therefore not only the speech recognition of Siri that is impressive, but the actions taken according to the meaning of the words recognised.

I have seen many speech recognition systems, some very good. Siri is an order of magnitude further forward. It is certainly not perfect and does make mistakes. But it is still the best speech system I have used. Many of the actions you can ask Siri to take are quicker to do that way than without Siri, such as setting a reminder call. That is ultimately why I (and many others) will use it - not because of the quality of the technology but because it is more efficient.