Source code of three systems for speech search is available here:
Each archive has a README file included which should be helpful for your first steps with the systems.
If you have questions, comments or bug reports, feel free to contact me.
Michal Fapso (ifapso@fit.vutbr.cz)