YouTube Video Translator

This project translates YouTube videos by extracting audio, transcribing it, and translating the text. The translated subtitles are then added back to the video.

Features

• Download YouTube videos
• Extract audio from videos
• Transcribe audio to text
• Translate text between Vietnamese and English
• Create subtitle files and convert them to VTT format
• Serve the translated video with subtitles through a web interface

Prerequisites

• Docker
• Docker Compose

Installation

1. Clone the repository:

   git clone https://gitlab.com/your-username/youtube-video-translator.git
   cd video_translator

2. Build and run the Docker containers using Docker Compose

   docker compose up --build

Usage

• Open your web browser and go to http://localhost:5000.
• Enter the YouTube link of the video you want to translate.
• Select the translation direction (Vietnamese to English or English to Vietnamese).
• Submit the form and wait for the processing to complete.
• The translated video with subtitles will be available for download.

Note

• First time you using it will download model from hugging face so you need to wait

Project Structure

• app.py: The main Flask application.
• youtube_downloader.py: Module for downloading YouTube videos.
• audio_extractor.py: Module for extracting audio from video files.
• transcriber.py: Module for transcribing audio to text.
• translator.py: Module for translating text using the transformers pipeline.
• subtitle_creator.py: Module for creating subtitle files.
• utils.py: Utility functions, including converting SRT files to VTT.
• config.py: Configuration file containing model names and device settings.
• requirements.txt: Python dependencies.
• Dockerfile: Dockerfile for building the Docker image.
• docker-compose.yml: Docker Compose configuration file.

Configuration

The configuration settings, including model names and device settings, are located in config.py.

Dependencies

• Python 3.9
• Flask
• transformers
• sentencepiece
• torch
• torchaudio
• accelerate
• yt-dlp
• openai-whisper
• moviepy

Development

To run the application locally without Docker, follow these steps:

Install the required Python packages: If you using Window you need to install ffmpeg follow this link: Install.

pip install -r requirements.txt
sudo apt install ffmpeg

Run the Flask application:

python app.py

Open your web browser and go to http://localhost:5000.

License

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

Acknowledgments

• Transformers by Hugging Face
• PyTube for downloading YouTube videos
• Whisper for transcription