Skip to main content

Medical detector and method for extra pulse detection


Detection and quantification of extra pulses in medical readings via analysis- by-synthesis based signal modelling of cyclic components, and subsequent modelling of the extra pulse component.


Extra pulses mixed to a cyclic pulse train component were found to characterize a distinct type of dysphonic (hoarse) voice quality. However, the option of automatically detecting extra pulses in irregular phonation was not considered in the past. The proposed medical detector and method for extra pulse detection accomplishes this task.

Bilder & Videos


The invention bases on analysis-by-synthesis and signal modelling. First, the fundamental frequency of the cyclic signal component is obtained. Second, the cyclic signal component is synthesized. A unit pulse train with the fundamental frequency is cross-correlated with the input signal to obtain an average cyclic pulse shape. Jitter and shimmer are estimated via iterative minimization of the modelling error. Fourier coefficients are obtained from the average pulse shape by Fourier transformation, and fed into a Fourier synthesizer, together with the fundamental frequency. Third, the cyclic signal component is subtracted from the input signal. Fourth, an extra pulse trigger means estimates the timing of extra pulses from the remainder of the signal. Finally, analysis results are displayed to the user.


The detector provides an objective means to test a signal for the presence of extra pulses and it includes several features that improve reliability and accuracy. It is much less labour-intensive than manual approaches. The detector can be easily used to administer diagnostic tests.


  • Diagnostic procedures in clinical practice
  • Assessment and monitoring of voice/speech quality
  • Medical data analysis
  • Non-medical data analysis Andrea Kolbus
+43 1 40160-25204
Peter-Jordan-Straße 70/III
1090 Wien




Medical reading analysis, Signal modelling, Event detection, Analysis-by-synthesis

Angebot Anbieter-Website

Kontakt | Geschäftsstelle

TechnologieAllianz e. V.
Christiane Bach-Kaienburg

c/o PROvendis GmbH
Schloßstr. 11-15
D-45468 Mülheim an der Ruhr