Limitations and Future Vision of Generative Music AI

Explore the current limitations and future prospects of generative music AI systems in this comprehensive video lecture. Delve into the challenges faced by text-to-music models, the drawbacks of deep learning approaches, and issues with music representation. Discover potential solutions to these problems, including ways to overcome the "curse of deep learning" and improve music representation techniques. Examine the importance of user interfaces in generative music systems and learn about procedural issues in research methodology. Gain insights into bridging the gap between engineering and domain knowledge in the field. Conclude with key takeaways and information about an upcoming generative music AI workshop in Barcelona.

Intro
Text-to-music limitations
The curse of DL
Deep learning limitations
Solving the curse of DL
Music representation limitations
Music representation: Way forward
User interfaces
Problems with research procedure
Research proedure: Way forward
Engineer / domain knowledge divide
Key takeaways
What's up next?
Workshop info