MuseMix

• 

  MuseMix – Where your memories become mixtapes, and your moods become art.

We were inspired by the emotional storytelling power of old-school mixtapes — those personal playlists people used to craft to express feelings words couldn’t capture. There was something intimate and magical about curating a set of songs that could speak for your mood, your memory, or your moment. As technology evolves, we saw an opportunity to revive that nostalgic experience through the lens of generative AI. This led to the creation of MuseMix — a creative web app that blends music, poetry, and visuals to turn emotions into shareable mixtape zines.

At its core, MuseMix allows users to input a short reflection or memory — for example, “walking home under October rain” — and transforms that into a personalized digital zine. Specifically, the output includes a mood-matching Spotify playlist, poetic song blurbs written by AI, cover art generated based on the user’s mood, and a stylized zine assembled into a downloadable PDF. In this way, MuseMix becomes a bridge between feeling and form — turning inner experiences into something you can hear, see, and hold.

To bring this idea to life, we used a modular, agent-based architecture. Each agent was responsible for a distinct task: the Mood Analyst interpreted the emotional tone of the input, the Playlist Generator curated fitting songs via the Spotify API, the Blurb Writer crafted poetic captions, the Cover Art Generator produced visuals using TinySD or SDXL, and the Zine Assembler formatted everything into a cohesive layout using WeasyPrint or ReportLab. To coordinate all these agents smoothly, we used Groq’s LLaMA 3 for fast, context-aware text generation and built the frontend in Streamlit. We also experimented with LangGraph to orchestrate multi-agent workflows.

However, the process wasn’t without challenges. One of the biggest hurdles was mapping abstract emotions to specific musical genres — especially when dealing with nuanced moods like nostalgia, longing, or wonder. Additionally, coordinating the outputs of multiple AI agents in real time, while ensuring the final PDF remained aesthetically consistent, required careful engineering and tuning.

Despite these obstacles, we’re proud of what we accomplished. We successfully built a functioning pipeline that generates emotionally resonant, artfully designed zines in under a minute. More importantly, we created something that feels human — a tool that invites reflection, expression, and connection.

Throughout this journey, we gained valuable insights into prompt engineering, emotion-driven personalization, and real-time generative applications. We also saw firsthand how AI can enhance — rather than replace — human creativity, helping users express what they feel in beautifully crafted ways.

Looking ahead, we plan to introduce new features such as customizable zines with handwritten notes or personal photos, support for multiple languages and regional music styles, and a “Gift Mode” for creating zines for friends or loved ones. We’re also aiming to launch MuseMix as a public-facing app on Hugging Face Spaces, making it accessible to creators, students, and journaling enthusiasts around the world.

In the end, MuseMix isn’t just a project — it’s an emotionally intelligent creative platform that turns thoughts into art, and feelings into shared experiences.

Built With

• groq
• llma
• python
• react
• spotify