jianchang512/pyvideotrans

Combines speech recognition, LLM-based translation, and text-to-speech synthesis in a unified pipeline with support for speaker diarization and multi-role dubbing. Integrates pluggable ASR models (Faster-Whisper, Qwen, WhisperX), translation backends (DeepSeek, ChatGPT, Ollama), and TTS engines (Edge-TTS, F5-TTS, CosyVoice), enabling both cloud API and fully local offline workflows. Provides interactive editing checkpoints throughout the translation chain plus a CLI interface for headless batch processing and server deployment.

16,496 stars. Actively maintained with 49 commits in the last 30 days.