OpenAI is reportedly exploring a new frontier in generative AI music creation. According to a report from The Information[1], the company is working on a tool capable of producing original music from both text and audio prompts. As part of its early research, OpenAI has reportedly collaborated with students from The Juilliard School, one of the world’s most prestigious performing arts institutions, to help develop high-quality training data.
Sources familiar with the project told The Information that OpenAI envisions a system capable of creating everything from guitar accompaniments for vocal tracks to soundtracks for videos, allowing users to generate a full-fledged song with minimal input.
While details on the project’s current stage remain unclear, one source indicated that Juilliard students were asked to annotate music score data that could be used to train the model to better understand musical structures and styles.
This isn’t OpenAI’s first attempt at music generation. The company has previously experimented with AI-driven audio models, though none have yet reached the scale or public availability of its text or image tools. The move also places OpenAI in direct competition with emerging platforms like Suno and ElevenLabs, which are developing their own AI-based music solutions.
As AI-generated music continues to grow, so do the challenges from concerns over originality and licensing to a rising flood of automated content on streaming services. The rapid evolution of these technologies suggests that the intersection of AI and art is only beginning to take shape.
References
- ^ The Information (www.theinformation.com)