Keyboards Must Die
A status report on speech recognition in the Linux world...
"Even in the Linux and FOSS world, voice recognition software has a long way to go. It's a complex problem for a number of reasons: the software has to differentiate between commands and dictation, and it has to learn to recognize how you pronounce words. English is a particular nightmare; we can't make reliable spellcheckers because of homonyms, and those same little words also give voice software fits.
"Way back in the olden days (around 1999) I had copies of Dragon NaturallySpeaking and ViaVoice to test. They ran only on Windows, of course, but they weren't bad. The main problem was a lack of computing power- our state-of-the-art PCs were Pentium II 166s. So I spent a fair bit of time waiting for the computer to catch up..."


