Electronic composer Holly Herndon has launched a new project, a custom voice instrument and "digital twin" nicknamed Holly+.
The instrument, created in partnership with Never Before Heard Sounds, uses machine learning technology to reproduce user-submitted audio in a computer-generated approximation of Holly's singing voice.
Users can upload any audio file up to 5 minutes in length, which is then transformed using a "deep neural network" that was trained on recordings of Holly Herndon's vocals.
The project makes use of technology similar to that of the "vocal deepfakes" that have been dominating headlines over the past year, which reproduce the voices of artists and celebrities using AI. What makes Holly+ different is that it's fully authorised and open for public use.
Herndon has even set up a DAO (Decentralised Autonomous Organization) co-operative to own the IP for Holly+. Any funds generated from the usage and licensing of the tools will be redirected to the co-op to be shared amongst artists using them, and support further development.
Herndon is no stranger to the intersection between music and technology. Her latest album, Proto, was created in collaboration with an AI program named "Spawn", which she developed alongside Mat Dryhurst.
Find out more about Holly+ on Holly Herndon's website.