To effectively train for singing technique by visually or auditorily feeding back the extent of singing technique.

A pitch extracting device 16 detects the pitch and amplitude of the input voice from a microphone 14. Singing parameter data on the interval, volume, vibrator, portamento, attack, tremolo, etc., regarding the singing technique are outputted from a singing parameter data output means 40 and tutor parameter data as models for the spinging parameters are outputted from a matching data output means 44. A matching means 46 matches the singing parameter data against the tutor parameter data and outputs state parameters. A video control means 48 and a sound control means 50 gives variations to the image on a display 56 and the sound from a speaker 58 based on the video control scenario 52 and sound control scenario 54 according to the state parameters.