In the present invention, a CPU acquires an achievement level based on the difficulty level of a song by evaluating the playing skills of the user based on a comparison of a plurality musical notation data expressing each note composing the song and music playing data generated in response to the song being played and inputted by the user; updates state variables (achievement level rank “achieve” and achievement level rank change “change”) in a value function table “value” based on the acquired achievement level; and gives advice of a type specified by the most effective (maximum value) advice type “action” retrieved from the value function table “value” [achieve][change][action] based on the updated state variables to the user playing and inputting the song.