Influence of Upper Airway Surgery on Voice and Speech Recognition.

Abstract

Upper airway surgery comprises a set of techniques that modify the anatomy of the vocal tract, including tonsillectomy and septoplasty. The objective of this work is to study the changes in acoustic parameters and the effects on the identification or verification of the speaker through the speech produced after the vocal tract surgeries, comparing them with a control group. A prospective study was performed between January 2019 and June 2019 including. The final study sample consisted of 84 patients who met the inclusion criteria. Of these, 31 underwent septoplasty, 26 tonsillectomy patients, and 27 controls. Demographic data and GRBAS evaluation were statistically evaluated. Tests were taken before surgery, 2 weeks after surgery and 3 months later. Furthermore, to establish the equal error rate, the recording of patients' voices was made with a succeeding acoustic analysis and programmed identification of the speaker through machine learning systems. A significant variance was observed in GRBAS, after surgery. Regarding acoustic parameters, a greater change was observed in the fundamental frequency at 2 weeks after surgery in the tonsillectomy group. Formants (F1-F3) and antiformants (AntiF1-AntiF3) changed in septoplasty group, not in tonsillectomy and control group at 3 months. When studying the impact of voice changes on the verification of the speaker through the speech, it was observed that there was a greater error in recognition in the tonsillectomy group at 2 weeks, coinciding with the results obtained in the rest of the parameters studied. Results suggest that upper airway surgery produces modifications in the vocal tract affecting GRBAS, acoustic parameters, including formants and antiformants, producing an effect on verification of the speaker through the speech.