The current methodologies of Automatic Speech Recognition are not tied to specific platforms.  This can help to the development of voice-driving applications using speech-processing libraries. In this line, our research group has rebuilt the code of IVORY, a methodology of isolated word speech recognition, as functions without dependencies, then able of concurrent execution. As a by-product, the resulting code is as portable as to suit in a generic system with minimal audio processing abilities. This paper shows this portability by mean of the development of a speech recognition application over a DSP board, coupling the ASR functions and an audio application, with lesser modifications.