Voice Recognition 3.0

Can we just pass right over voice recognition 2.0? I am getting really tired and bored after 25+ years of incrementalism. For sure, the incrementalism has paid off – just look at cars and iPhones and siri and a host of call center applications (which are equally annoying and progressive).

So here is my version of voice recognition 3.0. It would have these characteristics:

Goodbye Typing: it is so good that typing basically becomes a secondary application. Most emails and text messages etc would be generated by voice not be typing.
Grammar, stutter, and uh checks: it is so good that it comes to recognize uh’s and simple makes sure that I want to delete them and then does so automatically. Ditto bad grammar.
Personal vocabularies – it is so good that it builds a personal vocabulary for me over time and remembers it. For example, why can’t I expect my voice recognition system to learn that I live in Serenbe? And will use it in my communications a lot? Why does the best system today choke and think I am saying “Saran Be” or “Seren Bee”. But that is easy. Tough is that it should know that my brother lives in West Townsend, and that the movie I saw last Saturday was called The Hunger Games (and all I should need to say is “the movie I saw last Saturday night” or “the people I met at lunch today” – in other words it should draw regularly and liberally from my calendar and emails as I allow it to do so).
Search: I am tired of talking about this one. We should be able to search using natural language – just buy saying what is on my mind. “I vaguely recall a great article in one of the national magazines, I guess it was published in the last 24 months, and it spoke about genomic mapping getting cheaper and better. Can you find me that article?