A fractal characterization of speech waveform graphs

The thrust of this thesis is to investigate the applicability of the fractal dimension as a parameter in some speech recognition applications. Almost no work has been done in this area; we would like to see how well fractals can be used in speech analysis. This thesis is a report of preliminary investigations done in attempting to combine these two seemingly diverse fields in an attempt to lead us one step closer to the ultimate goal of fully automatic speech recognition.