#Introduction This project to build a Sphinx4 Neural network Hybrid. Those familiar with the speech recognition will know that currently Sphinx4 (albeit by the time we wrote this) is the best open source speech synthesis application available to researchers. The tool is based on hidden markov model and is quite effective. alternatively, neural networks started as champions in pattern recognition but its development was hampered by tons of issues which over time as technology matures get fewwer. This project is an attempt then to fuse the best portions of the two so as to come up with a crossbreed (hybreed) of both. to achieve this, we will use the front end module of sphinx4 for pre processing and feature extraction then use encog neural network toolkit to perform recognition. then contrast the findings against sphinx4 perfomance.