Acoustic diagnosis of voice disturbance using an analytical model of the vocal tract

Laryngeal cancer and voice disturbance share the initial symptom of hoarseness of voice. By analyzing changes in the voice, these diseases might be diagnosed. Because it is important to detect these diseases as early as possible, there is demand for a simple and highly accurate diagnostic method. The GRBAS (grade, roughness, breathiness, asthenia, strain) scale is used as an acoustic diagnosis method for these disorders but its objectivity is not well established. Instead, more accurate diagnosis is possible by capturing the waveform of the flow velocity at the vocal cords. The aim of this study was to enable diagnosis of laryngeal cancer and voice disturbance by identifying the sound-source waveform from voice measurements. For acoustic analysis of the vocal tract, we modeled the air in the vocal tract as concentrated masses connected by linear springs and dampers. The vocal tract shape was identified by making the natural frequencies of the analytical model correspond to the measured formant frequencies. The sound-source waveform was calculated from the measured voice waveform. To assess the validity of the model, we measured actual voices and used the model to identify the vocal tract shapes and corresponding sound-source waveforms. The identified waveforms were similar to the Rosenberg wave, which is an approximation of the actual vocal sound-source waveform. Because of the influence of local solutions, multiple vocal tract shapes could be identified from a single sample. However, mathematical analysis showed that these differed in amplitude of sound-source waveform only, which does not affect the shape of the waveform. From this, we conclude that our proposed methods are valid.