video_auto_trans

Video Auto Translation & Processing is an automated pipeline for video translation, vocal/music separation, voice cloning (TTS), subtitle generation, subtitle watermark removal (delogo), and video re-rendering. It provides a set of scripts that can be run individually or as a full pipeline via a command-line interface or an interactive terminal UI. The system requires FFmpeg to be installed and added to the system PATH. Configuration is handled through several files, including assets/_curren_key for API keys, assets/ffmpeg_config.json for FFmpeg settings, assets/render_config.json for style, .env for environment variables, and src/configuration.py for source configuration. Installation involves installing base libraries common for CPU and GPU, followed by hardware-specific libraries for either CPU-only or NVIDIA GPU setups. The pipeline includes steps for separating vocals and music (s0_demucs.py), transcribing audio to text using Whisper (s1_transcribe.py), translating subtitles (s2_translates.py), generating speech via Google TTS (s3_Google_speech.py) or voice cloning TTS via XTTS-v2 (s3_AI_speech.py), creating word-level or karaoke SRT files (s4_srt.py), auto-detecting hardsub blur boxes (s5_ocr_delogo.py) with a GUI editor (s5.1_ocr_delogo_gui.py), and rendering the final video (s6_video_complex.py). The entire pipeline can be executed sequentially with exec.py, or an interactive terminal UI dashboard can be accessed with menu.py.