While searching for stories on Open Source (for our new ZDNet blog) I came upon a conference in New Delhi where 200 engineers from around Asia shared experiences on speech synthesis and recognition.

It was the speech of Dayanidhi Maran (right), India's IT Minister, that first attracted my attention. He wanted applications in all India's languages, not just the "majors" like Hindi, Tamil, and English. And he wanted this to be open source.

But it was the conference itself I found most fascinating.

I have been thinking of voice processing mainly as a way to control Always-On applications. If grampa knows he can control his environment by saying a few magic words from his chair, that's a better user interface for him than making him click a mouse at a screen -- he's more likely to use it, and enjoy the application.

But as Maran noted, speech recognition and synthesis also enable better human-human communication. Given the large number of languages in Asia (it's far more complex than most westerners credit) that's a huge benefit. And it's spurring a lot of interest, a lot of development, the kind of work that could be adapted easily to an Always-On world. (Hey, Bengali grampas need virtual love, too.)

Frankly anything that pushes voice work along is good in my book. I'm absolutely convinced that this is the right interface for Always-On applications.