To provide a technique to add various performance expressions to the voice.

A table is stored in the relationship storing region 122 of a karaoke device 1 to show the relationships between the consonant information showing the kinds of consonants and the musical sound control information showing the kinds of tones. When a user of the karaoke device 1 sings, the user's voice is collected through the microphone 15. The controller 11 classifies the voice data showing the collected voice into a periodic section and an aperiodic section and finds out the kinds of consonants according to the spectrum patterns in the aperiodic sections. Then, the controller 11 specifies a musical instrument matching the kind of consonants by referring to the relationship storing region 122 and creates the MIDI data showing the tone of the specified musical instrument. The voice processor 16 supplies the analog signals matching the MIDI data to the loudspeaker 17, and the loudspeaker 17 emits the sound of the musical instrument matching the consonant of the user's voice.

COPYRIGHT: (C)2008,JPO&INPIT