Abstract

We introduce a novel dual-switch control paradigm based on the simultaneous measurement of frontalis muscle mechanomyography (MMG) and vocalizations (humming) using a single contact microphone attached to the forehead. Vibrations of the face and skull during vocalization are manifested as periodic high-frequency components in the microphone signal recorded at the forehead. The presence of these periodic components is detected by a normalized cross-correlation function, while muscle contractions are detected using a continuous wavelet transform method. The dual-switch provides two independent binary control signals. Eleven participants, including one individual with severe physical disabilities, participated in a cued activation task in which the dual-switch exhibited sensitivities and specificities of 96.8±3% and 98.4±1%, respectively for vocalizations, and 99.7±0.5% and 99.2±0.5%, respectively for muscle contractions. Since skin vibrations due to voiced sounds and muscle contractions have non-overlapping dominant bandwidths, the performance of the MMG switch was not affected by vocalizations. This new integrated MMG-vocalization access solution affords the user two binary switches from a single access site, and may thus augment access alternatives for certain individuals with severe physical disabilities.

Copyright © 2010 IPEM. Published by Elsevier Ltd. All rights reserved.