Some example Musebot music:



Quiet melodic

Autechre style noise


Mission Statement

The aim of the Musebot project is to establish a playful and experimental platform for research, education and making, that will stimulate interest and advance innovation in MuMe. Above all the Musebot project is a collaborative, creative experiment: we invite you to join us in making autonomous software agents that work together to make original music. These software agents will run on a network of computers creating music together in a “musebot ensemble” for a public audience, at a number of special events throughout 2015 and beyond.

Each software agent should correspond roughly to a single “instrumental part” in a piece of music, like a bassline or a drum beat. If we make these agents smart, then the resulting music will be coherent and continually evolving in interesting ways.

“Why do this?”

There has been a lot of research in MuMe systems, and the results are impressive. But a lot of the creative work is in standalone systems that compose or perform live with human improvisers. This is a daunting task and the results can be opaque. It is hard for people to share their ideas or their code, or work out ways that their systems might be incorporated into creative workflows. Musebots, by contrast, are small modular units that are designed to be shared and studied by others. By making collaboration central, the Musebot project forces us to be transparent in how our systems work.

“Is it a robot jam?”

We’d prefer to think of it like this: imagine composing music in any Digital Audio Workstation, but then replacing each of the tracks with an autonomous music-making software agent, that has to work with the other tracks to make the final piece of music. It’s a study into intelligent musical self-organisation. “Human musicians having a jam” can make for a useful metaphor, but computers can do things differently, so we prefer not to fixate on that metaphor. Either way, getting software agents to work together requires thinking about how music is constructed, and working out shared paradigms for its automation.

“Sounds ambitious!”

Yes, we want nothing less than to advance music AI, but there’s a lighter side… we want to explore the full range of creative expression musebot ensembles have to offer. If you have a bit of experience using one of the common computer music environments such as MaxMSP, PD, Processing or SuperCollider you’ll be able to make a simple musebot quickly by adapting our examples. You can download any number of musebots and the musebot conductor (a custom program that manages the musebots using network communication), and set up your own musebot ensemble.

Related Links

Skip to toolbar