Voice Recognition on Linux


I was pleasently suprised to get an email from a developer working on the CMU Sphinx project. He has good news the project is certainly live and kicking which is very good news indeed.

Paul Lamere is working on Sphinx - 4 a java version of CMU Sphinx. He had seen my laments about voice reconition on linux and very kindly gave me an update on sphinx 4. It seems the java version is leaner and meaner and the oldC versions.

I didn't have much success with the C version, couldn't even get the application to start up. You will say, since it's a open source application I am supposed to dig deeper to find out what's wrong and possibly even fix it, or provide enough information for the developers to fix it. Unfortunately there are so many new technologies for me to study each day 24 hours is hardly enough. So sphinx went to the back burner.

Getting back to why I am pleased with the java implementation of sphinx gathering momentum is that this is indeed one area where the much abused 'write once run anywhere' cliche is true. If you are carefull not to use the vendor specific features of the Java Speech API you stand a good chance of getting your code to run on both linux and windows.

comments powered by Disqus