🎵 AI Music Generator using LSTM & Deep Learning

An AI-powered music generation system that learns musical patterns from MIDI datasets using an LSTM (Long Short-Term Memory) neural network and generates original music in MIDI, WAV, and MP3 formats.

---

📖 Overview

AI Music Generator is a Deep Learning project that uses an LSTM neural network to learn musical patterns from MIDI music files. After training on a dataset, the model generates new musical note sequences, converts them into MIDI, WAV, and MP3 formats, and allows users to download the generated music through a modern Flask web application.

---

✨ Features

• 🎼 Train an AI model using MIDI datasets
• 🧠 LSTM-based Deep Learning model
• 🎹 Extract musical notes using Music21
• 🎵 Generate original AI music
• 💾 Save generated music as MIDI (.mid)
• 🔊 Convert MIDI to WAV
• 🎧 Convert WAV to MP3
• 📥 Download generated MP3 music
• 🌐 Modern Flask Web Interface
• 📊 Interactive AI Music Generator Dashboard
• 🎨 Responsive User Interface

---

🛠 Technologies Used

• Python
• TensorFlow
• Keras
• Flask
• Music21
• NumPy
• HTML5
• CSS3
• JavaScript
• FluidSynth
• FFmpeg
• Pydub

---

📂 Project Structure

AI-Music-Generation/
│
├── dataset/
│
├── static/
│   ├── style.css
│   ├── script.js
│   ├── generated_music.wav
│   └── generated_music.mp3
│
├── templates/
│   └── index.html
│
├── preprocess.py
├── train.py
├── generate.py
├── midi_to_audio.py
├── app.py
├── requirements.txt
├── README.md
├── notes.pkl
├── music_model.keras
├── best_model.keras
└── generated_music.mid

---

🚀 Installation

Clone Repository

git clone https://github.com/yourusername/AI-Music-Generation.git

Install Dependencies

pip install -r requirements.txt

Install FFmpeg

Download FFmpeg and add its bin folder to your system PATH.

Install FluidSynth

Install FluidSynth and download a SoundFont (.sf2) file (e.g., GeneralUser GS).

---

▶️ Run the Project

1. Preprocess Dataset

python preprocess.py

2. Train the Model

python train.py

3. Launch the Website

python app.py

Open your browser:

http://127.0.0.1:5000

Now you can:

• Train the model
• Generate AI Music
• Convert MIDI to MP3
• Download MP3

---

📊 Output

After successful generation, the project creates:

• 🎵 generated_music.mid
• 🔊 generated_music.wav
• 🎧 generated_music.mp3
• 🤖 music_model.keras
• 💾 best_model.keras

---

🚀 Future Improvements

• 🎼 Multiple instrument support
• 🎹 Genre-specific music generation
• 🎧 Built-in music player
• ☁️ Deploy on Render or Railway
• 🤖 Transformer-based music generation
• 🎵 Real-time AI music generation

---

👩‍💻 Author

Zakia Chandio

AI & Machine Learning Developer

---

⭐ If you like this project

Give this repository a ⭐ on GitHub.