OpenAI reportedly developing new generative music tool (techcrunch.com)

OpenAI is reportedly building a generative music tool that can create music from text and audio prompts, according to The Information. Sources say the system could be used to add background scores to videos or generate guitar accompaniment for existing vocal tracks. It’s unclear when the tool might launch or whether it will ship as a standalone product or be folded into ChatGPT and OpenAI’s video app Sora. The company is said to be working with Juilliard students to annotate scores for training data—suggesting an emphasis on symbolic score alignment as well as audio conditioning. For the AI/ML community this signals renewed investment in multimodal audio generation and music-specific training regimes. Text-plus-audio prompting implies models that can combine semantic instructions with timbral or melodic examples, enabling finer control than pure text-to-music systems. Using annotated scores could improve musical structure, harmony and instrumentation in outputs, but also raises questions about dataset provenance and copyright. The move follows OpenAI’s earlier (pre-ChatGPT) music experiments and its more recent focus on speech models; competitors such as Google and Suno already offer generative music work, so this could accelerate capabilities and deployment patterns for creative tools across audio production and content creation.