Abstract
The thesis is concerned with the development of computer procedures for translating morse-coded transmissions back to the original message. Starting mainly from digitized signals which have already been recognised as morse, the thesis describes a number of alternative procedures and shows their performance on actual data. Particular emphasis was given to the following aspects of the problem: the development and testing of several different algorithms for estimating the mark and space parameters: the amalgamation of short marks and short spaces into adjacent intervals: the detection of stepchanges in the data: the discrimination of different key types used in generating morse. Using a comparison program, scores are given for the correlation between the text obtained by the computer program with that obtained by an operator. The transcription quality obtained ranges from approximately 70% for manual morse to 95% or better for automatic morse.