Friday, September 23, 2005

CMU Sphinx

I tried a voice recognize program which was written by one of my colleague, during university time, she can't remember what is the library that used in the program, I found out is Sphinx, an opensource project from CMU. The program was written in Visual Basic and running on Windows platform. It used some sampling data to convert the voice to text, it support English and Bahasa. Since the program more focus on Bahasa, is more accurate when use in Bahasa mode. Only certain words in the sampling database will be recognized, but is good enough, very impressive. The sampling data is quite big, about 150M. I told my colleague, there is a few area that we can apply the technology. I have a Sun server which do not have a console login, every time I need to telnet into the server to shutdown the system, may be I can use the sphinx as the interface to shutdown the server, by just saying, "Server shutdown!". How nice it will be! Another area that we can use it in our office is the PBX phone system, by integrated into the Asterisk. When anyone call in to our office, if they do not know a person's extention number, he can just say the name, then the system will connect to the person's extention automatically. Anyway, that's for fun and brain storm, no one is going to spend the time on doing it.

No comments: