A talk assisting system has a vibration sensor for detecting tones produced through a musical instrument and a signal interrupter connected between a microphone and a sound system; while a player is fingering a music tune on the musical instrument, pieces of vibration data express a wide amplitude of the vibrations, and a tone discriminator changes a state signal to an active high level; and the signal interrupter is responsive to the state signal so as to block the sound system from the voice signal, whereby the tones are not mixed with the voice.