I’ve been wanting to dive deeper into AI, so I’m kicking off a project that combines a few of my favorite things: web development, music, and creative AI. The goal? Build a web app tailored for songwriters, especially guitarists, that can generate the “bones” of a song from a text prompt, giving you a foundation to craft your song from.
Project Overview
Imagine you’re a guitarist looking for inspiration and type in something like “a sad punk rock song using the blues scale.” The app would generate a song with chord progressions, key information, and a structure with labeled sections (verse, chorus, bridge, etc.). I’m focusing on a clean, responsive design and secure user accounts, so users can save and revisit their creations.
Tech Stack and AI Tools
The stack will be Next.js for the frontend with Tailwind CSS and shadcn/ui for styling. I’m planning on using Supabase for authentication and database management, keeping the backend scalable and secure. But for the AI portion, I’ll be experimenting with some specialized libraries:
- NLP for Prompt Parsing: I’m considering using
naturalto extract musical elements like mood, genre, and scale from user prompts. - Music Theory Handling: The
tonallibrary looks promising for managing scales, chords, and music theory-related tasks. - Chord Diagrams:
svguitarmight be a good option for generating chord diagrams, although this feature is still optional at this point.
Why AI?
AI has always intrigued me, especially when it comes to blending creativity with technology. This project gives me a unique opportunity to explore how AI can push the boundaries of music creation. By leveraging NLP and music theory, I can create an app that responds to natural language prompts in a way that feels intuitive and musical. It’s exciting to think about AI helping musicians break through creative blocks and inspiring new ideas.
Right now, this is just a sprout 🌱, and I’ll be documenting my progress along the way as the project evolves. Stay tuned!