Monday, 12 November 2007

ERE I AM, J.H.

While investigating the curious "Operating System" outburst from my computer last night (turns out PhotoShop's 'Save for Web' shorcut also "speaks" selected text - there go my Ghost in the Machine theories) and I started tinkering with some of the speech features on my Macintosh.

I recently installed the latest Mac operating system - 'Leopard' - and among a veritable gaggle of new features it has one new "voice" in the Speech options. I switched to this voice a while back, and it was this new voice (Alex to his friends) that muttered those fateful words to me last night.

So I thought about having a good listen to Alex speak, as I've always liked fooling around with the ole speech synthesis and I was impressed by the new sound. While it's still obviously synthetic, they are getting there in terms of natural-ness. Here's a sample I recorded of Alex reading 'Jabberwocky'. Alex has trouble with some of the phonetics, but still does a decent job.

Jabberwocky (reading) - Alex the Macintosh voice [1.4MB mp3 file]

For a while now the speech synthesis has changed it's cadences in accordance with punctuation and this feature remains and seems to be built upon. However, the thing that startled me (and you can hear it in the above sample) is that Alex is taking breaths. That sound before the odd line and word is the computer taking a breath. Like a real person. Some are long breaths, some are short. They are not always there. Someone, or a team of people have written parameters for my computer to breathe while speaking. Wow.

Unfortunately, one thing that has been removed from the Speech function (and has been gone for a while) is the ability for the computer to reel off ludicrously large numbers - I loved this feature! You used to be able to fill a document with random numbers and the computer would read it and calculate novemdecillions, septendecillions, quintillions and the like. Now it only reads them as a string of numbers. It's a shame as I would've liked to record this and use it... Somewhere.

Maybe in the next revision.