🎵 Song Identification, Mood Detection & Music Recommendation Website

A beginner-friendly web application where users can upload a short audio clip, and the system will:

1. Identify the song
2. Detect the mood of the song using Machine Learning
3. Recommend similar songs

🚀 Project Overview

Music platforms like Shazam and Spotify analyze audio signals to recognize songs and suggest similar music.

🤝 Contributers

• Ruvaifa
• Sourish
• Ayush

🧠 System Workflow

1. User uploads an audio clip (5–15 seconds)
2. Backend saves and processes the audio
3. Song is identified using a music recognition API
4. Audio features are extracted
5. ML model predicts the mood
6. Similar songs are recommended
7. Results are displayed on the website

🏗️ Tech Stack

🌐 Frontend

• ⚛️ React 18.3 - Modern UI framework
• 📘 TypeScript 5.5 - Type-safe development
• 🎨 Tailwind CSS - Utility-first styling
• 🎭 Framer Motion - Smooth animations

⚙️ Backend

• 🚀 FastAPI - High-performance Python framework
• 🐘 PostgreSQL - Robust relational database
• 🔥 Supabase - Real-time database & auth

🎧 Audio Processing

• 🎼 Librosa – audio feature extraction
• 📊 NumPy – numerical operations

🧠 Machine Learning

• 🎲 Scikit-learn
• 🤖 Pre-trained classical ML models

🎶 Music APIs

• 🎙️ ACRCloud / Shazam API – song identification
• 🎧 Spotify Web API – recommendations

---

🧠 Machine Learning Models Used

This project intentionally avoids deep learning to keep it beginner-friendly.

🎼 1️⃣ Audio Feature Extraction (Input to ML Models)

Using Librosa, we extract numerical features from the audio clip:

• Tempo (BPM)
• Zero Crossing Rate
• Spectral Centroid
• Spectral Bandwidth
• RMS Energy
• MFCCs (Mel-Frequency Cepstral Coefficients)

These features convert audio into a numerical vector, which ML models can understand.

---

😊 2️⃣ Mood Detection Model (Core ML Part)

🎯 Goal

Classify a song into one of the moods:

• Happy
• Sad
• Calm
• Energetic

---

🔹 ML Models

You can start with any one of these:

✅ Logistic Regression

• Easy to understand
• Fast to train
• Good baseline model

---

📊 Training the ML Model

• Dataset: Pre-labeled music datasets (e.g., GTZAN, Free Music Archive)
• Labels: Mood category
• Train-test split: 80/20
• Evaluation metrics:
  • Accuracy
  • Confusion matrix

The trained model is saved using joblib or pickle.

---

🔮 Prediction Flow

1. Audio uploaded by user
2. Features extracted
3. Features passed to trained ML model
4. Model outputs predicted mood
5. Mood shown on frontend

🎧 3️⃣ Recommendation Logic (ML + API Based)

🔹 Option 1: API-Based Recommendation

• Use Spotify Audio Features:
  • Valence
  • Energy
  • Tempo
• Recommend songs with similar values

🔹 Option 2: ML-Based Similarity

• Use Cosine Similarity
• Compare feature vectors of songs
• Recommend top N similar songs

📌 Beginners can start with API-based recommendations.

📁 Project Structure

tune-trace/
│
├── frontend/                  # React + TypeScript
│   ├── src/
│   │   ├── components/        # UI components
│   │   │   ├── AudioUpload.tsx
│   │   │   ├── SongResult.tsx
│   │   │   └── Recommendations.tsx
│   │   │
│   │   ├── services/          # API calls
│   │   │   └── api.ts
│   │   │
│   │   ├── App.tsx
│   │   └── main.tsx
│   │
│   └── package.json
│
├── backend/                   # FastAPI backend
│   ├── main.py                # API entry point
│   ├── audio.py               # Audio upload & processing
│   ├── song.py                # Song identification
│   ├── mood.py                # Mood prediction (ML)
│   ├── recommend.py           # Song recommendations
│   ├── ml_model.pkl           # Trained ML model
│   └── requirements.txt
│
├── uploads/                   # Temporary audio files
│
├── README.md
├── .gitignore
└── .env.example

---

🧪 Future Improvements

• CNN-based audio classification
• Spectrogram-based deep learning
• User-personalized recommendations
• Multi-label mood prediction

🤝 Contributing

We welcome contributions! Please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---