Partial laryngectomy has an influence on voicing; its degree depends on the particular surgery. During therapy the patient's voice has to be evaluated from time to time. Changes in criteria like volume, intelligibility and use of prosody have to be found and reported. The evaluation is subjectively and the method is time-consuming and expensive for doctor and patient. Therefore an automatic and objective method would be desirable.

In our work, we examine how good the patients' speech is recognized by an automatic speech recognition system and whether voice quality can (partially) be evaluated automatically. For this goal, the evaluation results of the automatic system and a human expert group have show a certain correlation. The patients' self-evaluation (SF-36, V-RQOL, VHI, Trierer Skalen) will also be part of an automatically computed measure which describes the voice quality.

In the precedent project SVcheck (http://www5.informatik.uni-erlangen.de/Forschung/Projekte/SVcheck/?language=en) on the analysis of continuous speech after laryngectomy, it was shown that there is a high correlation between the evaluation by human experts according to clinically relevant criteria and the word accuracy computed by an automatic speech recognizer or automatically computed prosodic features. This experience serves as the basis for the new project in which we will examine the effects of specific surgery on the mentioned parameters more in detail. The variance of pathology is very large as e.g. only one vocal fold or the false vocal folds (ventricular folds) may have been removed.