Two modifications in the acoustic source signal for an electrolarynx



Journal Title

Journal ISSN

Volume Title



Two treatments for improving the intelligibility and voice quality of an electrolarynx were developed and tested, using an electrolarynx synthesizer which permitted designs to be implemented by connecting function blocks in patch board fashion. The output of currently available electrolarynges is an acoustic pulse train of a fixed frequency. Their sound is quite monotonous, and their frequency spectrum exhibits resonances that give speech a harsh, somewhat nasal quality. The misplaced emphasis on certain frequencies in the speech spectrum may cause some vowel sounds to be distorted. First, an anti-resonance treatment that convolved the transducer impulse response with a three pulse group was developed, to suppress the tendency of the transucer to ring after the initial impulse. Second, a frequency inflection treatment was implemented, based on the effect of the lowest vocal tract resonance on the glottal source. When the first format drops in frequency, the coupling to the glottal source will raise the fundamental frequency (Berg, 1958). A listener test was performed on the speech of three laryngectomees using the new treatments. Statistical analysis confirmed that both treatments were significantly more intelligible than the unmodified electrolarynx whose transducer they shared, though tests of voice acceptability labeled all the devices as sounding non-normal. The results indicated that the two treatments do produce an improvement in intelligibility. They outscored an available electrolarynx with the same transducer; but they did not score as well as a different available electrolarynx using a throat vibrator transducer. The great amount of experience our electrolarynx users have had with this type of electrolarynx may explain this result. The results demonstrate the need for future experiments using similar techniques with better transducers.