Face Off: Hilarious Siri Vs. Google Search

I did laugh, but the poor performance is predictable. A single session of siri will produce nonsensical answers when asking a set of unrelated questions. The comparison is designed to favour google voice search and to produce giggles. Below is just a short explanation for why siri goes so horribly wrong, and how a reverse comparison can be set up to favour siri over google voice search.

First up, don't get me wrong, for this accent google voice search should be a far superior voice recognition system.

The 'problem' is that Siri is tailored to produce a particular kind of result, its goal is to be a personal assistant. It skews the voice recognition to more probable commands to enhance the accuracy of the 'personal assistant'.

Siri builds a query using an arrangement of data that surrounds the user. Their location, address book, contacts and successive questions. It tweaks the voice-recognition results to more closely match its functionality, this leads to better accuracy when using it like a personal assistant, and poorer accuracy when trying to use it for dictation. (iOS does come bundled with dictation software that doesn't have Siri's personal assistant tailoring, that would be a better way of comparing the voice recognition technologies.) However iOS doesn't have a comparable voice-for-search service out of the box.

Google's voice recognition is designed to run google searches, for some data it produces siri like results (e.g. game scores). That's because Google search provides a set of summary results for basic searches, e.g. definitions, weather and the like, these have been piped to create the appearance that of Siri-like functionality. However, one could set up a 'reverse' video, with Siri converting successive searches for "italian", "take away" and "meat balls" into a meaningful search for an italian restaurant in the area that offers a meatball dish and takeaway. Google voice search isn't going to produce a similar clarity of results.

In the end it comes down to the user, I don't feel that successive specific searches are ideal in a voice-setting. At the moment google voice search is a product that produces searches very well, but is that useful to users who could just as easily type in the request? The OS-shortcuts make sense and are very siri-like (e.g.asking it to set a timer) Google's system will garner strength by taking on more siri-like functionality, which is a voice-scenario that is more useful to real world users.