Speech Recognition


Last night I finally got the time to play around with speech recognition, and would like to share my experience here.

 

Speech recognition, as you know, is not something new. While it is used a lot to build phone based application - the sometimes annoying robotic phone operator -, there are not so much web based or desktop based applications.

 

From my personal point of view, there are two major issues that might play a big part in why speech recognition is not so popular for web based, or desktop application.

 

Reason 1: Reliability

Speech recognition engine only understand words if you say it in a certain way or in a certain pace. You say it too fast, it might not understand you. You talk with an accent, it might not understand you. Note: I was actually almost screaming last night trying to get the program I wrote, before it finally understand me. (You can actually teach the machine to recognize more voices)

 

Reason 2: Customer base

Most computers these days have a mouse and keyboard. Microphone, however, might not be as common.

 

In spite of the shortcomings, however, I still find it very cool and has lots of potential to do great stuff. I am sure future advancement in the field will only help it further. One of the advancement that I noticed the most is in the way the engine sounds now. In the past, they sound totally like a robot, and now it actually sound more like a person.

 

www.monk-e-mail.com is an example of a site that utilizes speech recognition technology.

Windows Vista will actually implement speech recognition to allow user to do certain task by saying it, which I am sure will help boost the field.

 

Add comment