DADABOTS

DADABOTS acts as a cross between a band, a hackathon team, and an ephemeral research lab. The project combines music creation with software engineering and research, operating as one integrated effort rather than a conventional outfit. It frames its work around collaboration, experimentation, and a DIY ethos in the context of live and recorded music.

The process centers on creating music with neural networks and customized tooling rather than standard instruments. It emphasizes unsupervised learning, data-driven exploration, and iterative training to generate new audio material. The approach highlights the challenges of training nets, including choosing parameters and evaluating outputs before shaping them into albums or live performances.

Core capabilities include building neural synthesis into music production, training multiple nets to explore different sounds, and curating generated material for human consumption. It discusses using high-performance hardware and specialized algorithms to produce long-form audio pieces, then arranging them into coherent releases.

Plans and scope mentioned include ongoing collaboration with artists and the aim to broaden access to neural music tools. The project references the role of neural synthesis in future audio software and notes hardware costs and parallel software development. It also notes the desire to open processes and expand collaborations beyond the current partners.

Use cases

• Experimental music creation using neural networks
• Collaboration with artists for new sound design