This repository contains the code and methodology for an AI-powered Arabic video translation pipeline, incorporating speech-to-text, machine translation, text-to-speech, and lip-syncing. This system enables real-time Arabic video dubbing with high-quality lip synchronization and natural voice cloning.
- Speech-to-Text: Converts Arabic speech to text with high accuracy.
- Machine Translation: Translates Arabic text into the target language.
- Text-to-Speech: Generates natural-sounding speech from translated text.
- Lip-Syncing: Aligns lip movements with the generated speech for realistic dubbing.
##Usage Run the pipeline using the following command: python main.py --input video.mp4 --language en This command takes an Arabic video (video.mp4), processes it, and generates an English-dubbed version with synchronized lip movements.
We assess the pipeline's performance using industry-standard metrics: -Word Error Rate (WER): Measures the accuracy of speech-to-text conversion. -BLEU Score: Evaluates machine translation quality. -Lip-Sync Accuracy: Assesses how well the generated speech matches lip movements.
To install dependencies, run:
pip install -r requirements.txt