This paper describes building of the basic components, particularly speech recognition and synthesis, of a speech-to-speech translation system. This work is described within the framework of the "Tongues: small footprint speech-to-speech translation device" developed at CMU and Lockheed Martin for use by US Army Chaplains.