This paper describes the testing of a multi-modal user interface for a television. The purpose of the test was to (1) investigate the overall acceptability of voice control for (television) remote control functions; (2) empirically investigate the impact of recognition error rates on acceptability; and (3) gather preference information relative to specific remote control functions. A Wizard of Oz method was used to simulate speech control at four different speech recognition accuracy rates. The four recognition accuracy rates used were 100%, 95%, 90%, and 85%. The results indicate that overall, voice-activated remote controls at all levels of accuracy were preferred over the manual remote control.
