Estimating Glottal Voicing Source Characteristics by Measuring and Modeling the Acceleration of the Skin on the Neck

Abstract

Quantifying vocal function is becoming increasingly important both for corroborating clinicians' subjective impressions during a voice evaluation and for assessing the effectiveness of surgery or voice therapy. Accelerometers offer significant potential for measuring voice use because they are immune to environmental noise, inexpensive, and suitable for ambulatory monitoring. This work investigates the extent to which measures of the acceleration of the skin on the neck (near the larynx) can be used to estimate important vocal function parameters. Simultaneous acceleration, acoustic, and airflow recordings from one male subject with a normal voice were made to initially examine relationships between the acceleration signal and the acoustic or airflow signals. A vocal system model is proposed to provide insights into these relationships. The model's acceleration-derived estimates of maximum flow declination rate (MFDR) and sound pressure level (SPL) are compared to direct measures obtained from the airflow (MFDR) and acoustic (SPL) signals at the lips. Results show high correlations between the estimated and measured MFDR (r=0.86) and SPL (r=0.94) values, suggesting that clinically relevant parameters can be derived from accelerometer data using a vocal system model.