Ly-Lyric App

• 

  Search for a song

• 

  Songs with similar lyrics

• 

  Songwriting Assistant

• 

  Lyrics Story

• 

  Translate Lyrics

• 

  Mood-Based Recommendation (Coming soon)

• 

  Song to Art Generator (Coming soon)

Inspiration

Have always heard great things about Snowflake and its capabilities, especially for data warehouse, Cortex Search and LLM integrations like Mistral-large2. This hackathon gave me the perfect opportunity to dive in, explore how everything works together, and build something creative that showcases the potential of these technologies.

What it does

Ly-Lyric App is an interactive platform for music lovers and lyricists. It allows users to:

• Search for songs by title, artist, or lyrics snippets.
  
• Discover songs with similar lyrics using semantic search.
  
• Generate creative lyric drafts inspired by existing songs.
  
• Uncover the backstory behind lyrics.
  
• Translate lyrics into multiple languages.
  

It brings together advanced Snowflake Cortex capabilities and LLMs to deliver a unique experience.

How we built it

Using:

• Streamlit for the user interface.
  
• Snowflake Cortex Search for retrieving song lyrics and metadata.
  
• Mistral-large2 for lyric generation, storytelling, and translation.
  
• Python for backend integration and quick iteration.
  

Challenges

Starting just five days ago, had to quickly familiarize myself with Snowflake, Snowflake Cortex Search, LLM features, and how they work together. Ensuring seamless integration, handling session state in Streamlit, and generating consistent outputs within a tight timeline were challenging but very fun and rewarding.

Accomplishments

• Successfully leveraging Snowflake Cortex Search and Mistral-large2 to build a minimal functional and interactive app.
  
• Creating a platform that seamlessly integrates multiple advanced features like semantic search, lyric generation, and translation.
  
• Learning and implementing new technologies in a limited timeframe.
  

Learnings

• Snowflake’s powerful Cortex Search and its integration with LLMs.
  
• Building user-friendly applications with Streamlit.
  
• How to handle prompt engineering, semantic search, state management in real-world applications.
  

What's next for Ly-Lyric App

• Optimize the performance overall.
• Expand the features, including mood-based recommendations and song-to-art generation.
  
• Add more personalization options to enhance the user experience.
• Improve the accuracy and creativity of outputs through further experimentation with Snowflake Cortex and Mistral-large2.
  

HAPPY EXPERIMENTING !! HAPPY BUILDING !!!

Built With

• llm
• mistral-large2
• python
• rag
• snowflake-cortex
• snowflake-db
• streamlit