It's the latest development in a long and rocky relationship between our voices and technology. Macintosh computers have had some kind of elementary voice recognition software built in since 1993; Windows users had to wait another ten years or so, as while the functionality was included with Windows XP, to use it you had to jump through hoops and install a special SDK that wasn't officially supported. But neither of them were highly developed enough for obvious applications such as dictation; the Mac version did originally allow you to operate menus by holding down a key, then loudly and deliberately saying "Open" or "Copy", but it was such a resource hog that this was abandoned, and only reappeared about 4 years ago with the "Speakable Items" facility in OS 10.3.

Fact is, though, it wasn't really missed by the vast majority of people. We're used to interacting with our computers by pointing an clicking in blissful silence, and while it's briefly amusing to watch them respond to orders, it's only ever going to feel natural if they respond accurately to a casual murmur, regardless of any background noise. Barking loudly at an erratic machine that's unable to discern between accents isn't only stressful and unproductive, it has the capacity to turn the office environment into a bizarre cross between the noisy carriage on a commuter train and an army training ground. Of course, voice recognition technology is developing all the time - driven not least by US defence departments, who appear to place more faith in a computer's ability to activate weapon release systems via a voice command than a pilot's ability to press a button marked "fire".

So at some point we'll undoubtedly be freed from the tyranny of the keyboard and mouse as an input system (with services like SpinVox currently showing the way forward); while those who'd resent turning their music down to write an email probably couldn't care less, there'll be several thousand people with carpal tunnel or RSI who'll be punching the air.