Abstract

An all-audio, hands-free and eyes-free interactive tutor provides time-efficient and reduced-boredom instruction by varying a course of instruction based on the correct and incorrect responses of the user so as to frequently repeat and provide positive feedback for poorly-learned items of knowledge, to periodically refresh well-learned items and to suggest that rest be taken or to switch to easier material whenever too many human errors have occurred. The audio interactive tutor includes an audio input module, in the preferred embodiment a voice recognition unit based on a finite state grammar, an audio output module, a course of study and a user model. In a preferred embodiment, the course of study provides for human error control as well as for voice recognition unit error control, which synergistically cooperate to render the tutor substantially fault-tolerant and therewith enables the employment of commercially-available but not excessively accurate voice recognition units. The tutor may be embodied to operate on a PC or workstation, or may be embodied to operate as a stand-alone tutor. Spanish language instruction is disclosed in an exemplary embodiment.