Voice recognition will never kill the keyboard was: Re:Evolution of the Apple Mouse

One thing I believe you missed mentioning is accents. Some of the voice recognition games on the Nintendo DS have this problem, where it can understand most UK accents but some can trip it up. (Personally, I know I can't understand a word anyone says in a broad Yorkshire accent... I found this out when I met my ex's grandfather years ago. Very embarrassing indeed). Then you have people that speak at different speeds. Some people talk slowly (Captain Kirk usually... only spoke three... words at a... time!), whilst some speak very fast and everyone else sits somewhere in the middle. Personally, I don't see voice recognition going anywhere major. Just like 3D TV, it's cool, but it's still just a passing phase. Regards, Andrew B aliensrcooluk at yahoo.co.uk ----- Original Message ----- From: "Ian King" <IanK at vulcan.com> To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at classiccmp.org> Sent: Friday, October 01, 2010 5:44 PM Subject: RE: Voice recognition will never kill the keyboard was: Re:Evolution of the Apple Mouse You're mixing two different arguments, but there is merit to both. Once upon a time, I worked on speech recognition and synthesis at Microsoft - we shipped the first Speech API that wasn't produced by Microsoft Research. (Instead of 137 interfaces, there were five.) It was a good product, offering quite accurate speaker-independent recognition, but it wasn't going to replace keyboards. For one thing: it's bad enough when you have a roomfull of people talking on telephones. Imagine if they were also talking to their computers! One of the problems speech recognition has yet to solve is the so-called 'cocktail party' scenario: a computer cannot pick out and focus on one voice among many. So if the woman in the next cubicle has a particularly loud, penetrating voice (not that that's ever happened to me), your results may vary.... I've also noticed that people who are talking to devices don't modulate their voices the same way people do when talking to other people. There's no sense of feedback, and volume rises. (It's interesting that the only place voice systems have really become successful is - on the phone!) Further, since the device isn't perfect but again provides no mitigating feedback upon error, people rapidly become frustrated with voice systems, once again leading to raised voices. IMHO a more important factor is that our application software is written for keyboard input, with varying support for mouse input. Spoken interaction is dramatically different from either command-line interaction or its translation to the GUI. This goes beyond UI, and requires a very different model of communication with our digital companions. Given the above objection, no one has put much time or money into tackling the problem. Speech recognition will not catch on either among vendors or customers if it's simply a replacement for the keyboard and the model of interaction it engenders. Regarding touchscreen devices, yes, I agree that mice are on their way out. Remember the cute little pop-out mouse on the HP Omnibook? As we move more to mobile devices (a category in which the iPad barely 'fits'!), physical pointing devices are awkward (including a stylus, which is easily lost). But once again we will need to make changes to our user interfaces: isn't it fun trying to select one line from a single-spaced list on a Web page? (I have an Android-based 7" tablet, which I'm coming to dearly love.) Also, there is a very significant difference between the mouse and the touchscreen: the former has a persistent cursor. The touchscreen has no 'default' for 'clicking' but instead enables all visible icons as potential action objects. This is more akin to the light-pen/gun devices that predate the mouse (the mouse was not the original pointing device). This subtle difference can make it challenging to port a mouse-based application to a touchscreen. While touchscreens will likely supplant the mouse, at least for the mobile computing world (which is becoming the dominant expression of information technology), speech recognition just isn't going to replace keyboards, simply because it's a fundamentally different means of communication. -- Ian ________________________________________ From: cctalk-bounces at classiccmp.org [cctalk-bounces at classiccmp.org] On Behalf Of Alexey Toptygin [alexeyt at freeshell.org] Sent: Thursday, September 30, 2010 6:49 PM To: General Discussion: On-Topic and Off-Topic Posts Subject: Voice recognition will never kill the keyboard was: Re: Evolution of the Apple Mouse On Thu, 30 Sep 2010, Liam Proven wrote: I know many, many people that can type faster than they can talk. And then there's programming. When every character matters, and many of them are punctuation, speech recoginition (and speech for that matter) falls flat on its face. It will never be faster to pronounce: print join(',', map $_->(), @$closures), "\n" for 1..$num; than it is to type it. And that's relatively readable; perl lets you code with >50% punctuation... When I see photos of the 'keyboardless' ipad, more often than not there's a virtual keyboard taking up half of that very expensive screen... I giggle and keep on typing :-) Alexey