OpenAI is said to be in the advanced stages of building a new generative AI music tool. It looks like the company is planning to tap into the fast-growing AI-driven audio creation space with this new tool. The tool is said to allow users to generate original music, soundtracks, and multi-vocal tracks simply by typing text or providing audio prompts. Something similar to how ChatGPT handles text and Sora manages video.
According to reports circulating online, users will be able to create instrumental tracks, background music for videos, or add guitar or drum accompaniment to existing vocals. It will also support multi-vocal generation and AI-powered mixing.
The system is expected to offer granular control over the output, allowing users to tweak style, tone, and energy to move quickly from rough ideas to finished musical pieces. This flexibility could make it useful for everything from songwriting to ad production.
However, the key differentiator here appears to be the training data. OpenAI is said to be collaborating with students from The Juilliard School to create annotated musical scores for model training. This partnership could help the company avoid the copyright issues that have plagued other AI music startups while ensuring higher-quality, musically accurate outputs.
Obviously, there is no word yet on when the tool will launch or whether it will be available as a standalone app or on OpenAI’s existing platforms. The company may also explore enterprise use cases, like enabling ad agencies to quickly produce jingles, soundtracks, or voice harmonies directly from text prompts.
Though it’s important to keep in mind that the generative music space is currently facing heavy scrutiny, startups like Suno and Udio are under fire for allegedly using copyrighted material to train their systems. OpenAI seems keen to avoid similar controversies; its Juilliard collaboration and emphasis on ethical data sourcing indicate a more cautious and compliant approach.
