I've been working to set up the CMU-Sphinx voice recognition software to play around with. Unfortunately, the sound the program is receiving from my microphones is flawed. The parts that come through sound clear, but segments of varying length are dropped seemly randomly and very frequently. The end result tends to be about half the length of the original. I heard it directly using their test AudioTool found here: CMU-Sphinx AudioTool. I noticed the same kind of missing data using the video recording features in the Cheese Webcam Booth that comes installed on Ubuntu. I know it's not a hardware problem, because when I use Audacity to record audio there are no problems. I would suspect that it has to do with Java in some way, but that's complete speculation. I'm currently using Ubuntu 11.04. If you have any suggestions or if there's more information you need from me to help me, please let me know. Thank you much!