Dead Set Radio is an experiment in the design of musical automata.  The intent is to explore the extent to which principles of music theory and composition from global genres can be emulated by rudimentary artificial intelligence.  The design is informed by Hopfield and Wilson-Cowen neural network models.  The final program could be considered a tool for musicians; the prototype produces audio, MIDI and sheet music that can be used by musicians for their own work. 

While the mathematical and musical integrity of the program is taken seriously, the music is not the complete ends of the project.  The core components were designed during a period in which I was unable to pursue music due to the severity of other obligations.  We live in an era that demands more of our lifespan and yields less in return.  The strain of this environment on artists and musicians is immense, and digital media serves as a tantalizing crutch to those who manage to produce at all.  On a broader level, the project is intended to get observers to question which aspects of the self they are willing to forfeit to automation.

Dead Set Radio is a challenge to listeners and musicians to keep their music human while cultural aesthetic standards are  challenged.  The output of the prototype forced me to reevaluate my musical sensibilities, and I am sure it will do the same for many others.

The prototype has been designed in MATLAB, and has already produced over 30 hours of audio.  My intent is to create an open-source standalone,  and to use it for some or all of the projects below, organized by time and resources required:

1. Exhibitions and/or recordings of music generated by the program.
2. A version that generates music in real time or with microphone data.
3. An online database application that learns from user input.
4. Live performances by musicians playing works produced or inspired by the output.

If you wish to follow this project or offer your talents or support, please check this page for future updates.