Nasality Detection in EMG-based Speech Interfaces

Nasality is a very important characteristic of several languages, especially European Portuguese. This paper addresses the challenge of nasality detection in EMG-based speech interfaces. By combining EMG data with real time imaging information, we explore the existence of useful information on the EMG data about the velum movement. Results indicate that is possible to “detect” zones of nasality, particularly by sensors positioned below the ear between the mastoid process and the mandible. By conducting an exploratory classification experiment, accuracy rates as high as 79.6% have been achieved for nasal vowel detection, paving the way for the development of an EMG-based speech interface for languages with strong nasal characteristics.