🎧 SmartPlaylist AI

SmartPlaylist AI is a machine learning project that clusters music tracks based on their audio features, allowing users to group songs with similar characteristics (like energy, danceability, and acousticness) into distinct playlists.

🌟 Features

• Audio Feature Analysis: Uses standard Spotify audio metrics to represent songs.
• ML Clustering: Implements K-Means Clustering to discover natural groupings of music.
• Optimal Cluster Selection: Utilizes the Elbow Method to determine the most effective number of clusters ($k$).
• Interactive UI: A user-friendly Gradio interface for real-time cluster prediction.
• Cloud Deployment: Fully deployed and hosted on Hugging Face Spaces.

🛠️ Tech Stack

• Language: Python 3.x
• Machine Learning: scikit-learn (K-Means, StandardScaler)
• Data Manipulation: pandas, numpy
• User Interface: gradio
• Model Persistence: joblib
• Visualization: matplotlib, seaborn
• Deployment: huggingface_hub

📊 Dataset

The model is trained on the Spotify Audio Features Dataset from Kaggle, which includes metrics such as:

• Acousticness: Whether the track sounds acoustic.
• Danceability: Suitability for dancing.
• Energy: Perceptual measure of intensity.
• Loudness: Overall loudness in dB.
• Valence: Musical positiveness (happy/sad).
• Tempo: Beats per minute (BPM).

🚀 Getting Started

Prerequisites

• Python 3.10+
• A Kaggle account and API token (kaggle.json).

Installation

1. Clone the repository:

   git clone <your-repo-url>
   cd "SmartPlaylist AI"

2. Install dependencies:

   pip install -r requirements.txt

3. Setup Kaggle API: Place your kaggle.json file in ~/.kaggle/ (Linux/Mac) or C:\Users\<User>\.kaggle\ (Windows).

Usage

1. Download the Data:

   python download_data.py

2. Train the Model:

   python train_model.py

   This generates model.pkl, scaler.pkl, and an elbow_plot.png.

3. Run the Application:

   python app.py

   Open the provided local URL in your browser to start predicting clusters!

🌐 Live Demo

The application is deployed on Hugging Face Spaces: 👉 SmartPlaylist AI on Hugging Face

📈 Evaluation

The model is evaluated using the Silhouette Score, measuring how similar a song is to its own cluster compared to other clusters.