VidLingo - AI-Powered Video Translation & Dubbing

Welcome to the VidLingo project! This platform is designed to automate the process of video translation and dubbing using a modular, AI-powered pipeline.

🚀 How to Run the Suite

The entire VidLingo suite is managed via Docker Compose, allowing for easy, modular execution of each service. All generated media will appear in the local downloads folder.

1. Build All Services: Builds the Docker images for all modules.

   docker-compose build

2. Run the Downloader (Module 1): Downloads a video from a given URL into the shared ./downloads folder.

   docker-compose run yt-downloader "https://www.youtube.com/watch?v=your-video-id"

3. Run the Transcriber (Module 2): Scans the ./downloads folder for videos and generates a timestamped transcription JSON file.

   docker-compose run transcriber

4. Run the Translator (Module 3): Scans the ./downloads folder for transcription files and translates them using the configured cloud AI. You can specify the target language using the TARGET_LANGUAGE environment variable. Note: This module requires a GEMINI_API_KEY to be set in a .env file at the project root. See .env.example for format.

   docker-compose run --env TARGET_LANGUAGE=French translator
   # Or for Polish (default):
   # docker-compose run translator

5. Run the TTS (Module 4): Generates a new audio track from the translated text, using Microsoft Edge's text-to-speech engine, mixes it with the original background audio, and remuxes it into a final video.

   docker-compose run tts

📁 Local File Workflow

If you want to process a local video file (instead of downloading from YouTube):

1. Place your video file (e.g., my_local_video.mp4) directly into the C:\VidLingo\downloads folder on your host machine.
2. Skip the yt-downloader step.
3. Start the pipeline from the transcriber service:

   docker-compose run transcriber
   docker-compose run translator
   docker-compose run tts

   The transcriber will automatically find your local video file in the downloads folder and initiate the rest of the dubbing process.

---

📦 Modules

This project is built with a modular, service-oriented architecture. Each service has its own README file for detailed information.

Module 1: YouTube Downloader (/services/yt-downloader)

• Status: ✅ Complete
• Description: A containerized Python service for media acquisition. More details in its local README.

Module 2: AI Transcription Engine (/services/transcriber)

• Status: ✅ Complete
• Description: A high-performance transcription service using faster-whisper on CPU. More details in its local README.

Module 3: Cloud AI Translator (/services/translator)

• Status: ✅ Complete
• Description: A cloud-native translation service using Google's Gemini API for dubbing-ready text. More details in its local README. Configuration: Requires a GEMINI_API_KEY in a .env file at the project root.

Module 4: AI TTS Service (/services/tts)

• Status: ✅ Complete
• Description: The final module, responsible for synthesizing dubbed audio using Microsoft Edge's TTS engine and mixing it into the final video. More details in its local README.

🛠️ Tech Stack

• Backend: Python 3.11
• AI / ML: faster-whisper, Google Gemini API, Microsoft Edge TTS
• Containerization: Docker, Docker Compose
• Core Libraries: yt-dlp, pydub, FFmpeg
• Automation: Git

Contributing

This project is in its initial development phase. Contribution guidelines will be established as the project matures.

---