Inverse analysis of vocal sound source by acoustic analysis of the vocal tract

Diseases occurring near the vocal cords, such as laryngeal cancer, often cause voice disturbance as an initial symptom. As an acoustic diagnostic method for such diseases, the GRBAS (grade, roughness, breathiness, asthenia, strain) scale is widely used, but its objectivity is not well established. Instead, more accurate diagnosis may be possible by capturing the waveform of the volume velocity at the vocal cords (the vocal sound-source waveform). The aim of this study is to enable diagnosis of diseases near the vocal cords by identifying the sound-source waveform from voice measurements. In the proposed method, an analytical model of the vocal tract is used to identify the sound source. The air inside the vocal tract is modeled as concentrated masses connected by linear springs and dampers. The vocal tract shape is identified by making the natural frequencies of the analytical model correspond to the measured formant frequencies. The sound-source waveform is calculated from the analytical model by applying the measured voice (sound pressure) to the lip position of the identified vocal tract. To assess the validity of the proposed method, an experimental device was made to simulate the human voice mechanism. The device is equipped with artificial vocal cords made of a urethane elastomer that are self-excited by air flow. The sound pressure equivalent to the voice was measured using a microphone set at the lip position of the experimental device, and the flow velocity at the artificial vocal cords was measured using a laser Doppler velocimeter (LDV). To assess the model's validity, the sound-source waveform identified from the measured sound pressure was compared with the waveform measured using the LDV