A device for inputting voice commands into a system such as a car navigation system includes a microphone for inputting voice commands, a display panel for displaying a help frame when required, a memory for storing a list of the voice commands and their respective usage frequencies in the past, and a controller for controlling operation of an entire system. A user confirms a voice command before inputting it, referring to a command list displayed on the display panel as a help frame. The command list shows all of the voice commands in a reverse order of their usage frequencies in the past. The commands which have not been frequently used and are not familiar to the user are shown in an earlier part of the list. Therefore, the commands for which confirmation are necessary can be quickly found in the list. Those commands which have been used in excess of a predetermined frequency level in the past may be eliminated from the list.