Algorithmic Whales

The three-part composition explores the sound worlds of two rave models (Realtime Audio Variation autoEncoder) trained by the Intelligent Instruments Lab on humpback whale songs and water sounds. The sound of the soprano saxophone is resynthesised in real time by the humpback whale model, transferring the timbre of the humpback whale songs to the saxophone sound. In a second voice, the encoding process of the timbre transfer is bypassed and the decoding parameters are controlled directly by an algorithm. The result are strange, machine-like variations of the whale songs, with which the saxophonist enters into a dialogue as the composition progresses. The conversation between the two voices is slowed down and fed into the water-trained rave model, creating the third voice, a slowly evolving soundscape that is directly influenced by the other voices.