Exploring the Benefits of Interaction with Audio Continuators

I recently completed my Master’s thesis in Music Technology, which focused on creating an Interactive Audio Continuator (IAC) (Lal, 2014). The Continuator is an interactive music system proposed by (Pachet, 2002), which automatically learns musical patterns from MIDI information using Variable-Length Markov Chains (VLMC). A user is then allowed to interact with the learned patterns in order to automatically create new, stylistically consistent music. In the work of (Kosta, 2012) and (Marchini, 2010), VLMC are applied to timbre and pitch features extracted from audio in order to create Audio Continuators; however, these systems are not interactive. In my work, I re-created the majority of Kosta’s framework, improved some of the machine learning methods, and proposed a paradigm and basic proof-of-concept for interacting with Audio Continuators. I would like to continue researching and exploring interaction with IACs, and propose that interacting with these systems may yield educational and therapeutic benefits.