OpenAI is currently testing the development of a new tool that can generate music using text or voice instructions, according to a report by The Information. According to the report, the company is collaborating with students from the prestigious Juilliard School in New York, who are helping to build a dataset to train the artificial intelligence. The goal, it was reported, is to enable the system to create guitar accompaniment for an existing song or add a soundtrack to a video—automatically and AI-based.
At this stage, it is unclear how advanced the project is or what stage of development it is in. A source cited in the report stated that Juilliard students were asked to mark and explain musical scores so that they could serve as raw material for teaching the system. This is not only an unusual collaboration between the world of technology and the world of art, but also another step by OpenAI in the field of automatic content creation—a field in which it has already experimented in the past, but has not yet presented an official music product to the public.
OpenAI’s interest in music is not surprising given the growing competition: Companies like Suno and ElevenLabs already offer tools for creating songs and voice clips based on artificial intelligence. However, as the technology advances, so too are the voices warning of the flooding of streaming platforms with machine-generated content.