🎬 AI Video Editor

An open-source AI-powered video editor that removes silence, adds subtitles, overlays relevant images, and enhances videos automatically.

** Disclaimer **

Currently, This only works on Podcast videos where there are less people in the frame and only 1 or 2 people talking at a time. I am still working on to make it work greatly on every kind of video. Till then, enjoy!! :D

🌟 Features

✅ Filler & Silence Removal - Removes unwanted pauses & filler words.
✅ Face Tracking & Smart Cropping - Automatically focuses on speakers.
✅ Live Subtitles - Generates accurate subtitles in real-time.
✅ Keyword-Based Image Overlay - Finds relevant images & places them in the video.

---

🚀 Upcoming Features

🔜 End-to-End Automation - Just type a prompt, and the video edits itself!
🔜 Locally Run LLMs - No API requests needed, making video editing even faster!

---

📌 How It Works?

Below is a high-level overview of how the AI Video Editor processes videos:

graph TD;
    A[Upload Video] --> B[Extract Audio]
    B --> C[Transcribe Audio]
    C --> D[Detect Silence & Fillers]
    D --> E[Trim Silence & Fillers]
    E --> F[Face Tracking & Speaker Detection]
    F --> G[Merge Audio & Video]
    G --> H[Generate & Overlay Subtitles]
    H --> I[Extract Keywords]
    I --> J[Download & Overlay Images]
    J --> K[Final Processed Video]

Loading

Each step is explained in detail in the notebooks/ folder.

---

🛠️ Installation

1️⃣ Clone the Repository

git clone https://github.com/end-9214/AI-Video_editor.git
cd AI-Video_editor

2️⃣ Create a Virtual Environment & Install Dependencies

python -m venv venv
source venv/bin/activate  # On macOS/Linux
venv\Scripts\activate     # On Windows

pip install -r requirements.txt

3️⃣ Check if PyTorch is CUDA-Enabled

Before running the project, ensure that PyTorch is installed with CUDA support:

import torch
print("CUDA Available:", torch.cuda.is_available())

If False, install the correct PyTorch version:

pip uninstall torch torchvision torchaudio
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

4️⃣ Add Your API Key

Create a .env file inside the Scripts/ folder:

touch Scripts/.env

Open the file and add:

GROQ_API_KEY=your_api_key_here

5️⃣ Run the App

streamlit run app.py

---

📂 Project Structure

AI-Video-Editor/
│── app.py                  # Streamlit App
│── requirements.txt         # Required Dependencies
│── processed_videos/        # Stores intermediate processing files
│── final_outputs/           # Stores final processed videos
│── notebooks/               # Detailed explanations of each step
│── Scripts/                 # All processing scripts
│   ├── Convert_to_Audio.py
│   ├── Transcription_script.py
│   ├── Silence_and_fillers_removal.py
│   ├── Face_Tracking.py
│   ├── Generate_Subtitles.py
│   ├── Overlay_Images.py
│   ├── .env                 # Store API keys here
│   └── ...
└── README.md                # Project Documentation

---

📖 Understanding Each Step

Each processing step has been explained in Jupyter notebooks inside the notebooks/ folder.
You can check them out if you want to understand the logic behind each step.

📌 Example Notebooks:

• notebooks/01_AI_Trim.ipynb
• notebooks/02_Face_Subtitles_and_Images.ipynb

---

📊 Results

Here are some example results from the AI Video Editor:

---

🎯 Vision

The goal of this project is to make video editing as simple as typing a prompt!
This AI-powered tool automates the entire process, allowing users to edit videos in no time. 🚀

🆕 Upcoming Feature:
✅ Locally Run LLMs - No API requests needed! This will make video processing faster and remove API limitations.

---

💡 Contributing

This project is open-source and welcomes contributions from everyone!
If you have new ideas or improvements, feel free to open a PR. 🚀

---

📧 Contact

If you have any questions, feel free to reach out!
📩 Email: karamveersingh2003111@gmail.com
🌍 GitHub: end-9214