This paper proposes an experimental approach to an IEC-assisted form of musical improvisation through practice-based research and aesthetics discourse on evolutionary art. The harmonic series of a snare drum acts as both source and fitness function for a real-time artistic performance environment, called “Populous Oscillation”, created in Max/MSP. The processing of the patch is based on the partials of the acoustic drum that, by way of genetic algorithms, populate, evolve and control electronic sound synthesis timbre, tonality and rhythmical patterns in connection with an artificial neural network. This presents a dynamic human-computer performance as either an augmentation of the acoustic instrument or a self-contained, mutating artificial agent. The paper seeks to map novel approaches to generative electro-acoustic music by suggesting a rapid variety and flow of fitness in both genetic algorithms and neural networks resembling modes of listening in free improvisation. The results span from prototype recordings, discussions of aesthetic judgment in improvisation as an evolutionary process, open organization of variety and perspectives on further system development, all anchored in the author’s primary practice as an artist.