Authors:
Damyan Damyanov
1
and
Vassil Galabov
2
Affiliations:
1
Technical University of Sofia, Faculty of Automation, Department for Industrial Automation, Bulgaria
;
2
Technical University of Sofia, Bulgaria
Keyword(s):
“pulse source – filter” model of speech production, speech communication in control systems.
Abstract:
In practice, when analysing, processing and generating signals, it is often assumed, that the process is of type “pulse source - filter”. Examples include the speech production process according to the theory of Fant,
analysis of shockwaves, ECG, EEG, seismology. For determination of the parameters of the filter many methods exist, most of which require the assumption of linear, all-pole model of the filter. It dates from the
time when the computational power of the processing systems was very low. From the 50-s on, many computational effective algorithms have been created. Their complexity is almost an order smaller
compared to those, using other models of the filter. In frequency domain the all-pole filter describes very well the processes, for which it has been created. In most of the practical solutions it has became classics,
and what follows is his application for other purposes, for which it may be inappropriate. In this paper, some general properties in the application of lin
ear all-pole filter and pulse-source for generating of periodical
signal are reviewed. These properties explain some phenomena of the modelled real process and give better interpretation for the constraints, which come out from the implementation of such model.
(More)