AliPay Sponsor
WeChat Sponsor
English ｜ 中文 ｜
# 🌏 VT Video Translation Assistant Break the language barrier for your videos — one click to multilingual translation. From video upload to multilingual dubbing, from subtitle generation to final rendering — an all-in-one solution to make your content understood worldwide. > 🌐 Try Online: [https://vt.fa-tools.com](https://vt.fa-tools.com) --- ## 🚀 Overview **VT Video Translation Assistant** is a full-process multilingual video localization and distribution tool designed for creators, educators, brands, and teams. It integrates **speech separation, recognition, LLM translation, voice cloning, subtitle editing, and video synthesis** into one seamless workflow. Supports nearly 100 languages — including Chinese, English, Japanese, Korean, French, and German. --- ## 🧩 Key Features | Feature | Description | |----------|-------------| | 🎧 Voice Separation | Automatically separates vocals and background music | | 🗣️ Speech Recognition | Accurate ASR using Whisper model | | 🌍 AI Translation | Context-aware translation using Qwen or GPT | | 🧬 Speech Synthesis | Supports XTTS, Edge-TTS, Azure-TTS engines | | 🎛️ Voice Cloning | Preserves original tone and emotion | | 📝 Subtitle Editing | Real-time in-page editing without reloading | | 🎬 Video Rendering | Automatically generates final multilingual videos | --- ## ⚙️ Environment Setup ### System Dependencies ```bash apt-get update apt-get install -y vim git wget build-essential ffmpeg portaudio19-dev libasound2-dev fonts-noto-cjk fonts-noto-cjk-extra fonts-wqy-zenhei fonts-wqy-microhei ``` ### Python Environment ```bash conda create -n trvideo python=3.11 -y conda activate trvideo git clone https://gitee.com/fgai/video-translate.git cd video-translate pip install torch==2.5.0 torchvision==0.20.0 torchaudio==2.5.0 --index-url https://download.pytorch.org/whl/cu124 pip install -r requirements.txt ``` --- ## 📦 Model Downloads ```bash mkdir -p models # Whisper ASR model modelscope download --model Systran/faster-whisper-large-v2 --local_dir ./Systran/faster-whisper-large-v2 # XTTS speech synthesis model export HF_ENDPOINT=https://hf-mirror.com huggingface-cli download --resume-download coqui/XTTS-v2 --local-dir coqui/XTTS-v2 # Link to TTS model path ln -s /root/project/release/models/coqui/XTTS-v2 /root/.local/share/tts/tts_models--multilingual--multi-dataset--xtts_v2 ``` --- ## 🧭 Run Services ### 💡 Method 1: Manual Start #### Frontend (Django) ```bash cd frontend mkdir -p media/final_videos python manage.py makemigrations python manage.py migrate python manage.py runserver 0.0.0.0:31005 ``` #### Backend (FastAPI) ```bash cd backend mkdir -p media/videos media/vocals media/bgm media/videos_novocals \ media/srts media/tts media/final_videos uvicorn app.main:app --host 0.0.0.0 --port 31006 --workers 1 --reload ``` --- ### 💡 Method 2: One-Click Startup (Recommended) A startup script is included in the project root: ```bash cd video-translate chmod +x start.sh ./start.sh ``` This will automatically start both the **frontend (Django)** and **backend (FastAPI)** servers. Once started, open: * Frontend: [http://127.0.0.1:31005/](http://127.0.0.1:31005/) * Backend: [http://127.0.0.1:31006/](http://127.0.0.1:31006/) --- ## 🔍 API Test ```bash curl "http://localhost:31006/api/tasks" ``` --- ## 🧠 Architecture Overview ``` ├── backend/ # FastAPI backend service ├── frontend/ # Django frontend └── models/ # Model storage ``` --- ## 💡 Use Cases - Multilingual short video republishing - Course, lecture, and training localization - Product demos and brand internationalization --- ## 🧑💻 Tech Stack | Component | Technology | |------------|-------------| | Backend | FastAPI, SQLite, Uvicorn | | Frontend | Django, Bootstrap, jQuery | | Models | Whisper, XTTS, Edge-TTS, GPT/Qwen | | System | Ubuntu 22.04+, Conda, Docker | --- ## 💖 Support the Project If you like this project and wish to support its continued development, you can sponsor via QR code:
AliPay Sponsor
WeChat Sponsor