Research Shorts: Using AI in contemporary composition

Zakiya Leeming

This week I’ve been talking to Zakiya Leeming about her work with artificial intelligence (AI) and music, and it’s great to be speaking with a composer about this. AI is a pressing issue not only for artists across all disciplines, but also for all those in arts and music education. Through Zakiya’s artistic practice, composition becomes a kind of research in itself.

She explains: ‘My work explores recent and emerging AI technologies and considers how they can complement, expand or inform a composition practice. I have been surprised to find how diverse both the technologies and the uses of these can be and the impact it has had on my practice, including making new compositional methods available and informing the development of new conceptual frameworks. I have also been surprised by how human-driven and human-centric these can be. This research highlighted how AI technologies can be created and adopted for use in art.’

Zakiya told me that her interest in AI and music came out of her work thinking about composition informed by science and technology, and her work with a new tool, PRiSM SampleRNN, which can be used to learn and generate raw audio from a custom dataset.

‘I began with a series of creative questions designed to explore both the technology and its potential impact on artistic practices. The research uncovered a wellspring of potential from the incorporation of new AI technologies in a contemporary composition practice. I am currently working on a new series of works centred on the uses and implications of AI for voice in collaboration with voice artist Emma Clarke.’

During the project, Zakiya composed a series of works that explored AI technologies in different ways. She went on to explain ‘This included designing datasets to test how variables affected outcomes, and the design of methodologies for the creation, development, and performance of each work. The scope of the research included ethical questions about opacity of training data and consent to its use, and highlighted possible alternatives to pre-trained AI models such as the use of custom AI tools that allow datasets to be limited to material created through an artist’s own practice.’

This research demonstrates how AI can be used in a variety of ways to enhance a musical composition and performance practice. Zakiya explained that it ‘could be of use for artists and educators alike looking for strategies to explore AI in composition and performance practice. It highlights a distinction between pre-trained and custom AI, and how datasets can be limited to artist or students’ own original work, as well as how they can be modified for different creative results. This research can also start conversations about creative attribution, authorship and creative roles when artists engage with AI technologies in their practice.’

Interview by Dr Sarah K. Whitfield – Research Manager for Music Mark

Learn more about Zakiya’s compositions

Listen to the AI project and read more about them