Tim O'Reilly has posted some interesting thoughts on the use of voice in the mobile interface, sparked by Google's recent voice powered search application for the iPhone. His first idea, that an easily accessible and usable mobile device will put people closer to 'the cloud' is one worth thinking of, especially when you consider how much faster accessing your online data would be when using speech recognition.
Read on in the full article.
I don't think speech recognition is ever going to be reliable enough for most people, it's just too flakey compared to actually entering data yourself. It's going to have a big novelty factor, but not much beyond that in most situations. (Of course there might be niche applications where it is useful...)
It's one of those things like handwriting recognition or artificial intelligence which features heavily in fiction and predictions of the future, but never really gets good enough to actually fly.
We people have difficulties with understanding each other. Guess that computer with a weaker understanding of the human mind and frame of reference can do? Did you ever use the wrong word in a spoken sentence? Slip of the tongue? Exactly! Pressing a button is far more secure to active or executes something hazardous for both man and machine. I.e. Telling your machine to 'get lost' might not be a good idea as Asimov demonstrated in one of his stories.
Google search application is a pretty innocent and useful application for voice recognition, and so is voice recognition for several other functions of my mobile phone, but not for driving my car i.e.
Even with entering search terms, there is absolutely no way that any speech recognition software will be able to convert speech into text reliably, and that will put a lot of people off using it in the long term.
For example "there", "their" and "they're" are all pronounced the same way but are written differently and mean totally different things. I can't see any way for a computer to know which one a human would mean.
Of course there are similar problems with SMS predictive text, but you only have to press a button to see the next alternative and you're pressing buttons anyway. If you have to press a button with speech recognition... why not type in the term yourself?
I remember my old Nokia had voice recognition (2001?) albeit very basic. I dont see it being beyond the bounds of todays technology to have voice recognition software on a mobile..
I'm sure Tzer2 is right about voice recognition being flaky. The amount of editing required (later) of the text is too much.
I can't agree that 'there' and 'they're' should sound the same though. Surely you get the y involved in 'they're'?
Whilst we're on the subject I wish someone would help Rafe to sort his 'our' from his 'are'. He's done one in one of the current news stories. So disappointing.