Creating a musical improvisation machine can be regarded as practical way to learn more about the character of its creator. Through the making of my own improvisation machine, a software patch cheekily named AIYA , I have begun to learn more about my own biases as a musician as well as how to dissect the complex interactions and customs that are at play during my improvisations. I attempted to translate the metaphorical black box of spontaneous human interaction into concrete algorithms for the machine’s behavior. In the process of doing so, I began to view my own position within an improvisation (both with humans and with machines) in a new perspective, which afforded me distance to my normal immersion as a performer playing the flute and allowed me to examine and strategically arrange the underlying principles of improvisation that drive the performance.
Performing with an improvisation machine also caused me to examine its potential to be a musical agent and question what it could mean to have agency in a performance. The machine was not only triggered by my behavior on the stage, but triggered behavior in my own playing in ways that I perceive are mimicking the spontaneous creation processes at work in human-to-human music improvisation. The impression of machine agency became an important point of investigation in the development of the machine.
From a technical perspective, the creation of an improvisation machine is not a new endeavor. Lewis’ pioneering works for Voyager (2003) can already be seen in the late 1980s, and works from IRCAM (OMax) in the late 1990s and other improvisation machines developed by Carey (2019), Hoffman and Weinberg (2011), and Lorway et al (2019) already shed light on the impact of such machines on the approach to improvised music performance. However, most improvisation machines, such as those mentioned above, do not take any movement activity from the performer(s) as inputs for the machine’s decision making. In my case, I wanted the improvisation machine to not only take my sounds as inputs, but also my movements (e.g., via the accelerometers/gyroscopes, cameras, full-room motion capture systems, etc.) and my relationship to the space.
This thesis serves as a technical and conceptual documentation of AIYA’s development, where each stage of its development is accompanied by a performance work created for it. In an attempt to organize the narrative of the creation of the machine, the left-hand column displays tags of concepts that each paragraph touches upon. Clicking on the tags will reveal a search result of all references of the tag from this thesis.
 “AIYA” is named after the Chinese expression of the same name — a versatile exclamation that can express shock, excitement, anger, or stress, depending on the context. As it also includes the abbreviation “AI,” the name “AIYA” pokes fun at the often-experienced shock/excitement/stress of working with machines when they attempt to be representations of intelligence.