Preliminary test of a real-time, interactive silent speech interface based on electromagnetic articulograph

A silent speech interface (SSI) maps articula-tory movement data to speech output. Alt-hough still in experimental stages, silent speech interfaces hold significant potential for facilitating oral communication in persons after laryngectomy or with other severe voice impairments. Despite the recent efforts on si-lent speech recognition algorithm develop-ment using offline data analysis, online test of SSIs have rarely been conducted. In this paper, we present a preliminary, online test of a real-time, interactive SSI based on electro-magnetic motion tracking. The SSI played back synthesized speech sounds in response to the user’s tongue and lip movements. Three English talkers participated in this test, where they mouthed (silently articulated) phrases using the device to complete a phrase-reading task. Among the three partici-pants, 96.67 % to 100 % of the mouthed phrases were correctly recognized and corre-sponding synthesized sounds were played af-ter a short delay. Furthermore, one participant demonstrated the feasibility of using the SSI for a short conversation. The experimental re-sults demonstrated the feasibility and poten-tial of silent speech interfaces based on elec-tromagnetic articulograph for future clinical applications.