All the Noises: Hijacking Listening Machines for Performative Research by John Bowers & Owen Green
Research into machine listening has intensified in recent years creating a variety of techniques for recognising musical features suitable, for example, in musicological analysis or commercial application in song recognition. Within NIME, several projects exist seeking to make these techniques useful in real-time music making. However, we debate whether the functionally-oriented approaches inherited from engineering domains that much machine listening research manifests is fully suited to the exploratory, divergent, boundary-stretching, uncertainty-seeking, playful and irreverent orientations of many artists. To explore this, we engaged in a concerted collaborative design exercise in which many different listening algorithms were implemented and presented with input which challenged their customary range of application and the implicit norms of musicality which research can take for granted. An immersive 3D spatialised multichannel environment was created in which the algorithms could be explored in a hybrid installation/performance/lecture form of research presentation. The paper closes with reflections on the creative value of ‘hijacking’ formal approaches into deviant contexts, the typically undocumented practical know-how required to make algorithms work, the productivity of a playfully irreverent relationship between engineering and artistic approaches to NIME, and a sketch of a sonocybernetic aesthetics for our work.
