Thursday, March 18, 2004
NASA - picking up subvocal speech
Voice recognition as an input method has never caught on, for 2 main reasons. Firstly, it's not very accurate, but hopefully that wil change over time as more processor power can be thrown at the problem and more accurate models of human speech and language can be deployed.
The second problem is a practical one: unless you are in an environment no-one minds you talking but everyone else is quiet, it's a bit tricky to use voice input reliably and politely. NASA's on picking up subvocal speech might be the answer - by measuring the nerve impulses sent to your throat as you read or speak silently to yourself, their system is able to recognise words without you actually making any noise. of course, with systems like this around, you'll have to be careful what you say to yourself when writing an email about your boss :)
Voice recognition as an input method has never caught on, for 2 main reasons. Firstly, it's not very accurate, but hopefully that wil change over time as more processor power can be thrown at the problem and more accurate models of human speech and language can be deployed.
The second problem is a practical one: unless you are in an environment no-one minds you talking but everyone else is quiet, it's a bit tricky to use voice input reliably and politely. NASA's on picking up subvocal speech might be the answer - by measuring the nerve impulses sent to your throat as you read or speak silently to yourself, their system is able to recognise words without you actually making any noise. of course, with systems like this around, you'll have to be careful what you say to yourself when writing an email about your boss :)
(c) 2003-2005 Russell Beale
Atom
RSS