OpenAI’s New AI Music Generator: What We Know

▼ Summary
– OpenAI is developing a tool that generates music from text and audio prompts, as reported by The Information.
– The tool could add music to videos or provide instrumental accompaniment to vocal tracks, according to sources.
– It is unclear when the tool will launch or if it will be a standalone product or integrated with existing OpenAI services.
– OpenAI is collaborating with Juilliard School students to annotate musical scores for training data.
– While OpenAI has previous generative music models, recent focus has been on audio models for text-to-speech and speech-to-text.
Rumors are swirling about a significant new project from OpenAI, reportedly a sophisticated AI music generator capable of creating music from simple text descriptions and audio inputs. This development, if confirmed, would mark a major expansion of the company’s creative AI portfolio beyond its established text and video generation tools.
According to sources familiar with the matter, this innovative tool could serve multiple creative purposes. One potential application involves seamlessly adding custom musical scores to pre-existing video footage. Another exciting possibility is its ability to generate instrumental accompaniments, such as a guitar track, to complement a standalone vocal recording. The exact launch timeline remains uncertain, and it is not yet known whether this will be a standalone product or a feature integrated into existing platforms like ChatGPT or the video generation app Sora.
To ensure the model produces high-quality, musically coherent output, OpenAI is said to be collaborating with students from the prestigious Juilliard School. These musicians are reportedly assisting in the critical process of annotating musical scores. This annotated data is essential for training the AI to understand the intricate relationships between musical notation, theory, and the resulting sound.
This is not OpenAI’s first foray into AI-generated music, though previous models were released before the massive success of ChatGPT. More recently, the company’s public audio efforts have centered on voice synthesis and transcription technologies. OpenAI is entering a field with established competitors, including tech giant Google and the specialized AI music startup Suno, both of which have already released their own generative music models.
(Source: TechCrunch)