The purpose of this paper is to describe the use of a speaker-adaptive system to help resolve many of the difficult recognition problems encountered in the automobile environment. In brief, the system adapts by continuously monitoring the speaker's voice and updating a “phone table” that is tuned to the speaker. The limited vocabulary templates consist of a series of fixed pointers into the varying phone table. The phone table actually consists of two parts: voiced periodic sounds and aperiodic sounds. The periodic sounds in the table tune the system to the driver and the aperiodic portion tunes the system to the acoustic environment. Impulse noises are handled effectively by the proprietary signal processing system currently used in SCOTT's Coretechs VET 3 speech terminal. Methods will be presented to improve the system's ability to function in various noise environments and its ability to adapt to new environments and speakers.