DDSP études for tenor saxophone & trumpet (2020). This work explores the use of the DDSP model, to introduce compositions, sound objects and performances in the practice of computer music and human-machine interactions. This artwork follows work on machine musicianship from researchers like Robert Rowe and George E. Lewis and explores AI as an instigator of compositional and performance explorations. The DDSP model ‘translates’ audio input material according to its understanding of what it was trained on and can transform inputs from a compositional, aesthetical and performative level. The use of the DDSP model presents machine-learning explorations through three layers of the creative process: composition, sound design and performance. Here, the études stemmed from training data of performances of tenor saxophone and trumpet and are referred to as machine learning performance translations. The études are mainly addressed as performances, and they should be listened to as aleatoric outputs from a machine musicianship filled with blemishes and moments of alluring idiosyncrasy.
These études stemmed from translations of its author and they explore the ability of this model towards transhumanist musical literacies that explore different modes of listening and languages outside of their respective instrument idioms. Input material became a reservoir of structural and conceptual possibilities – allowing the model to provide new and characteristic interpretations through idiomatic performances. The model harnessed content in flexible compositional formats and alluring idiomatic behaviours, providing new abstract and intuitive performances to the authors material. The model provided idiomatic performances characterized by peculiar levels of singular expression that exhibit idiosyncratic forms of machine musicianship, allowing each instrument model to become a sort of improvisation and performance partner. The DDSP model showcased behaviours that breakthrough the traditional hierarchies of the instruments that it attempted to reproduce. The ingenuity and expressive, disjointed behaviours of the model exhibited urges towards improvisational explorations that encourage modes beyond established practices of instrumental techniques of these instruments and showcase ‘extended DDSP techniques’ that alter the sonic characteristics available in the organic forms of these instruments.
These pieces should be approached and listened to as epistemic études that can inquire structuralist notions of composition and performance from a human and a non-human standpoint. Hence, these études are mainly dedicated to free, inhibited études: the work presented here focuses on idiomatic performative music and extends to the general topic of improvisational frames and musical literacies, exploring new modes of improvising languages, practices and compositions that exist inside and outside of the issue of practising established idioms.