An approach to the automatic transcription of stenotyped texts

This paper discusses the development of an automatic transcription system based upon the "Michela" shorthand machine. The aim of this sytem consists in transforming the continous text of pseudo-syllables, produced as output of the Michela machine, into a readable text. We describe the general methodology of our system, propose some techniques for "segmentation" and "aggregation", improve the output text quality by dictionary searching and give an evaluation of the text legibility.