CSUN 2006/Speech Recognition

I'm a big fan, and long time user of Speech Recognition - Voice-to-Text technology. I've never liked Dragon much, generally finding it hard to train, unforgiving of vocal variation, clunky in interface, and too expensive - but all that said, I know that there are huge fans out there. I do use ViaVoice very successfully, and have used it with second-graders, tbi clients, people struggling with ms, and many, many other difficult situations (I use the Pro-USB 10 version). I'v also had some luck with the speech recognition software included in Windows XP, though this hardly is the solution for everyone.

Wednesday at CSUN I saw two new exciting options. One, SpeakQ, from Quillsoft, which makes the word prediction software WordQ looks like something you need in your "toolbox." It is not just simple, the youngest child could master it, but it is the first of these software packages to allow a non-reader or very weak reader to train themselves - without a whispering helper or a tape recorder. The software simply says, "say this," and the user repeats the phrase. Brilliant! SpeakQ is an add-on to WordQ, and costs more than ViaVoice (though still half what Dragon costs) when purchased together, but it's a lot of tool for $350. (US)

I also saw a preview of the accessibility features in the new Microsoft Vista operating system. This needs to become the standard for schools and governments - many which still use the Windows 2000 system, a very poor choice for those with access issues. In Vista, the magnifier is a little better, the on-screen keyboard the same, but it is much easier to get to the Access Control Panel (which is now a part of set up), and which even contains a series of questions that will lead users to the best choices in accessibility solutions. The Narrator (screen reader) will now allow any SAPI-compatible voice to be attached, so text-to-speech output quality will be improved.

The biggest change is in the Speech Recognition/Voice Command program that is built in. This has been improved dramatically, and now features a brilliantly simple interface: "There are two major types of commands you can use with Speech Recognition: ‘Say what you see,’ and ‘Click what you see’. In addition, for times these commands might not work, like if you don’t know the name of a certain toolbar button, you can say ‘Show numbers.’ Numbers will then appear over everything on the screen and you can say those numbers to click the object under them." Look at this report, download the pre-release user guide here. Look at the speech commands here.

And one last solution... A company called Talk Technologies is selling "Voice Sylencer," a microphone that will allow voice-to-text users to operate their systems in noisy, and non-private environments. It is "soundproof" microphone you can speak into in office cubicles, in classrooms, perhaps on trains and planes, and may solve the "you can't take notes with this" issue in voice-to-text. It also has great possibilities for "echoing captioners" working in classrooms and meetings (more on that technology soon). (less than $200 US for the USB mic system).

from sunny Los Angeles... Ira Socol