19th INTERNATIONAL CONGRESS ON ACOUSTICS MADRID, 2-7 SEPTEMBER 2007 SPEECH INTELLIGIBILITY OF TWO GROUPS OF CORDECTOMIZED PATIENTS AFTER LARYNGOFISSURE AND LASER SURGERY

Abstract

Speech intelligibility is inversely related to the noise generated in the vocal folds, in the resonance cavities, and in the environment. In this study the intelligibility of two cordectomized groups of patients, treated with two different surgical techniques, was analysed. One group underwent laryngofissure with conventional surgery; the other underwent surgery by laser. Each group recorded a list of 48 random items, words, and pseudowords. These lists were presented under five different conditions of S/N in an anechoic room with TDH 48 headphones. The stimuli were presented under two conditions of no noise and under three conditions the items were masked with white noise, in which S/N was +3dB, 0dB, -3dB and -6dB. The results show that without masking, intelligibility was very similar for both groups of patients. However, when the speech was masked, intelligibility was better for laryngofissure patients. Moreover, the difference in the recognition scores between the two techniques increased with the degree of masking. INTRODUCTION. The source-filter theory of speech production [1] explains speech production as the convolution of two factors: a source of vibration, which is accomplished by the vocal folds; and a resonator, located in the oral tract. So this relation can be expressed thus: S = Fv * Tr (Eq. 1) where S is speech; Fv is the source of vibration, and Tr is the oral tract or the transfer function, radiation of energy from the lips and mouth being negligible for characterization of the speech signal. To a great extent, the mechanism of source generation is independent of the filtering process. The characteristics of the source are not influenced by, nor depend on the acoustic properties of filters. During prephonation, the vocal folds are adducted, closing the glottis and obstructing the airflow. When the vocal folds abduct they start to vibrate due to the joint action of the pressure exerted by the sub-glottal retained air plus the Bernouilli effect, produced when the airflow passes from the trachea to the open glottis, a smaller opening than the trachea. Vibration of the vocal folds induces the resonances of the oral tract, an anatomical structure that looks like a 17cm long tube (depending on age and gender). In this tube the tongue modulates areas of different section and length, which acquires different resonance properties, amplifying certain bands of frequencies and attenuating others [2]. In patients with larynx diseases, if a tumour is localized in the vocal cords and it is in the first stages of development, it is possible to operate directly on the cords (cordectomy) preserving the other laryngeal structures. In the case of bilateral cordectomy, the patient can use the ventricular folds to speak or the postoperative rearrangement of the laryngeal structures [3]. The

3 Figures and Tables

Cite this paper

@inproceedings{Garrigues200719thIC, title={19th INTERNATIONAL CONGRESS ON ACOUSTICS MADRID, 2-7 SEPTEMBER 2007 SPEECH INTELLIGIBILITY OF TWO GROUPS OF CORDECTOMIZED PATIENTS AFTER LARYNGOFISSURE AND LASER SURGERY}, author={J. L. Garrigues and Jennifer V. Romero and Jason Cerda}, year={2007} }