Chosen Techniques for Processing of Pseudowhisper of the Laryngectomized

In this paper we present novel signal processing techniques for improvement of intelligibility of the so-called pseudowhisper, which is a kind of the residual speech, typical for the rest speaking abilities of a major part of the laryngectomized patients, i.e., of those, who went through the larynx (vocal cords) removal surgery. First of all we show how it is possible to suppress annoying vocal artifacts and other unusual disturbances, which are present in a typical pseudowhisper, such as tongue clicks, plosive sounds (consonants) and the tracheal outlet breath sounds, which typically occur in a pseudowhisper of the laryngectomized. Furthermore, this paper deals with an additional and very effective pseudowhisper postprocessing technique consisting in inserting vowels and other related voiced speech segments into the previously cleaned and denoised pseudowhisper (by means of the presented techniques) in order to make it drastically more comprehensible. Illustrative results of the intelligibility tests carried out by the authors before and after applying the suggested signal processing techniques are presented.