We are searching for a Java Developer who can create a custom red5 implementation which calls to the sphinx4 engine to parse the audio recorded. This is for an e-learning project using speech recognition technology to help students learn how to read. The Sphinx software is really sensitive and needs a TON of configuration to load in the recognition patterns/dictionaries. Sphinx is a native wrapper for the JS engine (java speech). This will really be cutting edge for html5.. we will have to use something called socket.io We will stream the microphone to the server. There is an open source FMS alternative called Red5 that will have to be customized two that when it received the end of stream command it will convert the flv that it generates to a wav. Then, that wav will be passed to the speech to text engine The speech to text engine will generate text and then that will have to pass back to the client. (This will be advanced because its not rest per-se) It's going to be a push to the webpage, so we will have to use socket.io and node.js, which will allow the server to send messages to the client without polling the sever constantly The trick is a Java developer who can marry these two things together on the server-- the rest is easy. socket.io and node.js are super new technologies for html5 Again, it would be great if the java developer has experience with the Java Speech Engine, which establishes standards for defining recognition patterns Sphinx specifically is a major plus Keywords: Web Programming