|
|
--- |
|
|
title: Copyright Purpose Song Recommender |
|
|
emoji: 🚀 |
|
|
colorFrom: blue |
|
|
colorTo: purple |
|
|
sdk: gradio |
|
|
sdk_version: 5.24.0 |
|
|
app_file: app.py |
|
|
pinned: false |
|
|
short_description: AI-powered song recommendation engine for copyright purpose. |
|
|
--- |
|
|
|
|
|
# 🎧 AI-Powered Song Recommender for Creative Projects |
|
|
|
|
|
Welcome to the **Copyrighted-Purpose Song Recommender** — an intelligent assistant that helps you find the perfect song for your project using natural language. |
|
|
|
|
|
Whether you're working on: |
|
|
- 🎬 a cinematic scene |
|
|
- 📱 a viral social media video |
|
|
- 📢 a brand campaign |
|
|
- 💔 an emotional film |
|
|
- etc etc |
|
|
|
|
|
This app will analyze your needs and recommend songs using real music intelligence powered by: |
|
|
- **Spotify audio features** (valence, energy, tempo, danceability, etc.) |
|
|
- **YouTube metadata** (likes, comments, views) |
|
|
- **LLM parsing** (DeepSeek LLM for production, and still on research for compatible CPU model in HF demo.) |
|
|
|
|
|
--- |
|
|
|
|
|
## 💡 How It Works |
|
|
|
|
|
1. **Describe your project** in natural language (e.g. _"I want a sad, emotional breakup song for a scene for a K-drama"_). |
|
|
2. The system **extracts structured intent** like mood, tempo, genre, etc. |
|
|
3. Songs are recommended from our catalog based on matching features. |
|
|
4. ✅ You can give feedback (👍/👎) and export the results. |
|
|
5. As an overview of the outcome, go to notebook.ipynb -> 4.2 Temporary result |
|
|
 |
|
|
|
|
|
--- |
|
|
|
|
|
## ✨ Features |
|
|
|
|
|
- 🎼 **Valence + Danceability mood filtering** |
|
|
- 🔁 **Reference song similarity matching** |
|
|
- 📊 **YouTube popularity sorting** |
|
|
- 🧠 **LLM-powered input parsing** |
|
|
- 💬 **Feedback logging with export** |
|
|
- 📁 **Downloadable CSV for results & feedback** |
|
|
|
|
|
--- |
|
|
|
|
|
## 🚀 Try It Now |
|
|
|
|
|
1. Enter a description of your scene or campaign |
|
|
2. Click **🔍 Recommend Songs** |
|
|
3. Optionally give feedback or export results |
|
|
|
|
|
--- |
|
|
|
|
|
## 🛠️ Built With |
|
|
|
|
|
- `Gradio` – UI |
|
|
- `Transformers` – LLM model loading |
|
|
- `LangChain` – for prompt templates and parsing |
|
|
- `Pandas` – data filtering and exporting |
|
|
- `DeepSeek LLM 7B Chat` – for input parsing (production) |
|
|
- This space is using compatible model for CPU, yet still on research the best model. |
|
|
|
|
|
--- |
|
|
|
|
|
## 📂 About the Dataset |
|
|
|
|
|
- The song catalog includes audio features from Spotify and engagement stats from YouTube. |
|
|
- You can modify or expand this dataset in your fork. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🙏 Credits |
|
|
|
|
|
Developed by Shavira Z (personal project), for use in music publishing song user/clients. |
|
|
|
|
|
--- |
|
|
|
|
|
## 📬 Feedback or Suggestions? |
|
|
|
|
|
Feel free to reach out or leave an issue. We’re happy to collaborate or improve this experience! |
|
|
|
|
|
Email: [email protected] |
|
|
|
|
|
LinkedIn: https://linkedin.com/in/shavirazh |
|
|
|
|
|
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference |
