For a system to be able to generate realtime accompaniment to
previously unknown songs, it must predict their harmonic development,
i.e. the chords to be played. We claim that such a system must combine
long-term experience, to identify typical chord sequences (e.g. II–V
and II–V–I), with ‘on-the-fly’ adaptation to
track-recurrent structures (e.g. choruses and refrains) of the particular
song being played. We have implemented a prediction system using a neural
network model that encompasses prior knowledge about typical chord
sequences. The results achieved are very encouraging, and rather better
than those reported in the literature. However, our predictor could not
adapt its behaviour to the idiosyncrasies of each song, since online
learning is difficult in neural networks. In this paper, we propose an
extension to our previous work by the inclusion of a rule-based sequence
tracker, which detects recurrent chord sequences while the song is being
performed. We show that this hybrid model, which combines a neural network predictor with a rule-based sequence tracker, improves the
system's performance.