Acoustical analysis of tracheoesophageal voice

Abstract

Acoustical voice analysis of laryngectomees is a complicated matter because of the often weak periodicity of the voice and the high noise component. This study consists of a feasibility study and validation of an acoustical tracheoesophageal (TE) voice analysis on a sustained vowel based upon recordings of 66 laryngectomees from four clinics in three European countries. Based on reliability analysis of the acoustical data, TE voices can be objectively divided in three categories: (I) good voices with low-frequency harmonics and noise taking over at the higher frequencies; (II) moderate voices consisting of repetitive bursts of sound energy with low repetition rate and a weak periodicity due to high levels of noise, even at the low frequencies; (III) poor voices with no detectable or very weak fundamental frequency or envelope periodicity. The voice samples from category I and II correlate well with perceptually analyzed voice quality parameters, which supports the robustness and validation of this acoustical analysis method to analyze TE voices.