We researched and developed a new system for score-following of orchestra performances that uses audio-signal amplitude.In the existing score-following system,pitches extracted from the sounds produced during the performance using spectral analysis.These sounds are matched with the pitches notated in the score.However,in orchestra music,many instruments are often played at the same time and the fundamental tones and overtones of these instruments overlap,making pitch-detection very difficult.However,the envelope of the amplitude of the performance forms distinct peaks at the note-onset positions.Therefore,we proposed and implemented a real-time score-following system that matches the amplitude peaks extracted from the performance sounds with the amplitude peaks that can be predicted by reading the score.