In this paper the authors describe the development of a session system whic
h follows human performance and creates an improvisational accompaniment. O
ne of the features of this system is that it can use acoustic instruments s
uch as pianos and guitars instead of MIDI instruments which are common in c
onventional session systems. In a session system complex sounds such as cho
rds must be input and the input musical information must be analyzed in rea
l time. The authors' system can recognize chord names in real time through
signal processing that takes advantage of the fact that instrument sounds a
re harmonics, while observing the restriction that detailed musical informa
tion such as the pitch of the sounds comprising a chord or the accuracy of
the duration of sound production are not analyzed. The recognition rate for
four-note chords using thirds was shown in an experiment to exceed 95% usi
ng this method. In addition, the function which extracts the smallest repet
ition unit from a piece of music automatically allows for a session system
that does not need to know the score beforehand. (C) 2001 Scripta Technica.