Dauphin: A Signal Processing Language - Statistical Signal Processing Made Easy

Abstract

Dauphin is a new statistical signal processing language designed for easier formulation of detection, classification and estimation algorithms. This paper demonstrates the ease of developing signal processing algorithms in Dauphin. We illustrate this by providing exemplar code for two classifiers: Bayesian and k-means, and for an estimator: the Kalman filter. In all cases, and especially the last named, the code provides a more conceptually defined approach to these problems than other languages such as Matlab. Some Dauphin features under development are also highlighted, for instance a infinite list construct called streams, which is designed to be used as a natural representation of random processes.