🥁 AirPlay – Real-Time Gesture-Controlled Drum Kit

Play a virtual drum kit using just your hands and a webcam — no physical instruments required.
AirDrum uses computer vision and hand tracking to detect drumming gestures in real time and play the corresponding drum sounds.

---

🎬 Demo

Point your index finger at one of the four quadrants on screen and strike downward to play a drum!

Quadrant	Drum
Top-Left	🎩 Hi-Hat
Top-Right	🦵 Kick Drum
Bottom-Left	🔔 Cymbal
Bottom-Right	🥁 Snare

---

🛠️ How It Works

1. Your webcam feed is captured and processed frame-by-frame using OpenCV.
2. MediaPipe Hands tracks up to 2 hands and identifies 21 landmarks per hand.
3. The tip of your index finger (landmark 8) is used as the "drumstick".
4. A depth-change algorithm detects a forward striking motion by monitoring the Z-axis movement of the fingertip between frames.
5. When a strike is detected in a quadrant, pygame plays the corresponding drum sound.
6. A cooldown timer prevents repeated accidental triggers from a single gesture.

---

📦 Requirements

• Python 3.8+
• A working webcam

Install dependencies

pip install opencv-python mediapipe pygame

---

📁 Project Structure

AirDrum/
│
├── Play.py           # Main application
│
├── snare.wav         # Snare drum sound
├── cymbal.wav        # Cymbal sound
├── kick.wav          # Kick drum sound
├── hihat.wav         # Hi-hat sound
│
├── snare.png         # Snare icon (transparent PNG)
├── cymbal.png        # Cymbal icon (transparent PNG)
├── kick.png          # Kick icon (transparent PNG)
└── hihat.png         # Hi-hat icon (transparent PNG)

---

▶️ Running the App

python Play.py

• The app opens in fullscreen mode.
• Press q to quit.

---

⚙️ Configuration

You can tweak these variables at the top of Play.py:

Variable	Default	Description
cooldown	0.25	Seconds between allowed hits on the same drum
depth_threshold	0.025	Sensitivity of the strike detection
min_detection_confidence	0.7	MediaPipe hand detection confidence
min_tracking_confidence	0.7	MediaPipe hand tracking confidence

---

🧠 Tech Stack

Technology	Role
OpenCV	Webcam capture, frame rendering, visual overlays
MediaPipe	Real-time hand landmark tracking
pygame	Low-latency audio playback
Python	Core application logic

---

🚧 Known Limitations

• Works best in good lighting conditions.
• Accuracy may decrease if hands are partially out of frame.
• Z-depth from MediaPipe is a relative estimate, not true 3D depth — works well for strike detection but is sensitive to hand angle.

---

🤝 Contributing

Pull requests are welcome! Some ideas for improvement:

• Add more drum zones or customisable layouts
• Record and playback drum sessions
• Add a visual beat indicator / metronome
• Support MIDI output

---

📄 License

This project is open source and available under the MIT License.