Latest Developments In Music Generation

Lately, the entire AI community feels like AI agents and LLMs are the only things happening in AI. But that’s not true, it is sad that other cool ideas do not get as much attention as they should. So, today we are going to dive deep into music generation and look into FluxMusic.

The reason I want you to read this blog is that people in AI should be exposed to new ideas, outside of LLMs, I feel somehow a lot of AI engineers just don’t know enough tricks and rely too much on API calls and copying code from HuggingFace. So, without further ado, let’s jump into music generation.

Table Of Content

• Rectified Flow
• Transformer-based Diffusion Models
• What Is MelSpectrogram?
• Model Architecture Of FluxMusic
• Conclusion

Photo by Possessed Photography on Unsplash

Here’s Part II:

WAVTOKENIZER: Tokenizer For Audio Language Modeling

The first concept that we want to talk about in music generation is Rectified Flow.

Rectified Flow