Zach Evans is the head of the Harmonai research lab, the music research wing of Stability AI, working on open-source generative AI tools for musicians.
"We're launching new models and features in Stable Audio (www.stableaudio.com) including the ability to upload your own audio for style transfer. We're also working on new open-source text-to-audio models and improvements to Dance Diffusion." Zach Evans Links
Chapters
00:00 Introduction and Background 01:26 Dance Diffusion: An Overview 14:34 Inference and Output Generation 25:00 The Subset of Sound and Music 30:16 Current Projects and Future Plans 34:29 Why is GPU faster for this kind of stuff? 36:07 Main use case of text-based version 37:49 Challenges with text prompts for audio 39:18 Improving control mechanisms for models 42:06 The core process of generative AI 43:56 The future of AI in the audio space 46:03 AI models and overfitting 47:19 Automated mixing and mastering 50:22 Personal reinforcement learning 53:08 AI as a tool, not a replacement for humans 56:33 The impact of AI on jobs and artistry 58:23 The importance of personal taste and ownership in art 01:00:32 The complexity of AI's impact on the music industry 01:01:36 Where to find Zach Evans and his work