A Lightweight Video Subtitle Translation Tool | 一款轻量级视频字幕翻译工具
Language is no longer a barrier to watching movies! | 语言不再是观影的阻碍!
LightVT is an easy-to-use video subtitle translation software that uses advanced AI technology to help movie enthusiasts overcome language barriers and discover excellent film and television works from different cultures.
- 🎬 Video Subtitle Extraction: Intelligently extract subtitle content from video files
- 🌍 Smart Translation: High-quality multilingual translation based on large language models
- 📝 Subtitle File Translation: Directly process SRT subtitle files
- 🗒️ Plain Text Translation: Support for translating plain text files, such as .txt
- ⚡ GPU Acceleration: Support CUDA acceleration for significantly improved processing speed
- 🔧 Simple Interface: User-friendly graphical interface with intuitive operation
| Component | Purpose | Verification Command |
|---|---|---|
| FFmpeg | Subtitle extraction | ffmpeg -version |
| CUDA | Translation, GPU acceleration | nvcc --version |
| NVIDIA Driver | Translation, GPU support | nvidia-smi |
To use the translation function, you must have a large language model. The source code does not include large model files.
If you don't have one on your computer, download model files from the following sources:
Recommended models:
- Minimum configuration: Qwen3-8B (suitable for 8GB+ VRAM)
- Recommended configuration: Gemma3-12B and above performance models (suitable for 12GB+ VRAM)
FFmpeg Not Found
Problem: "ffmpeg command not found" error
Solution:
- Download FFmpeg
- Add FFmpeg to system PATH
- Restart terminal and verify:
ffmpeg -version
CUDA Related Errors
Problem: GPU acceleration not working
Solution:
- Confirm NVIDIA driver installation:
nvidia-smi - Confirm CUDA Toolkit installation:
nvcc --version - Check CUDA version compatibility (supports 11.8+)
- If problems persist, disable GPU acceleration and use CPU mode
Out of Memory
Problem: Insufficient VRAM when processing large files
Solution:
- Reduce GPU layer settings
- Use smaller model files
- Process long videos in segments
- ✅ Basic subtitle extraction functionality
- ✅ AI translation integration
- ✅ GPU acceleration support
- ✅ Graphical user interface
- ✅ Smart glossary
- 🔄 Cloud model support
- 🔄 Bilingual subtitle output
- llama.cpp - High-performance LLM inference
- FFmpeg - Multimedia processing framework
- CustomTkinter - Modern Python GUI framework
- HuggingFace - Model hosting platform
If this project helps you, please give us a ⭐
LightVT 是一款简单易用的视频字幕翻译软件,使用先进的AI技术帮助影视爱好者跨越语言障碍,发掘并欣赏来自不同文化的优秀影视作品。
- 🎬 视频字幕提取:从视频文件中智能提取字幕内容
- 🌍 智能翻译:基于大语言模型的高质量多语言翻译
- 📝 字幕文件翻译:直接处理 SRT 字幕文件
- 🗒️ 纯文本翻译:支持翻译纯文本文件,比如.txt
- ⚡ GPU 加速:支持 CUDA 加速,显著提升处理速度
- 🔧 简洁界面:用户友好的图形界面,操作简单直观
| 组件 | 用途 | 验证命令 |
|---|---|---|
| FFmpeg | 字幕提取 | ffmpeg -version |
| CUDA | 翻译,GPU加速 | nvcc --version |
| NVIDIA驱动 | 翻译,GPU支持 | nvidia-smi |
要使用翻译功能,必须要有大语言模型,源代码不包含大模型文件。
如果你的电脑上没有,从以下来源下载模型文件:
推荐模型:
- 最低配置:Qwen3-8B (适合8GB+显存)
- 推荐配置:Gemma3-12B及以上性能的模型(适合12GB+显存)
FFmpeg 未找到
问题:提示 "ffmpeg command not found"
解决方案:
- 下载 FFmpeg
- 将 FFmpeg 添加到系统 PATH
- 重启终端并验证:
ffmpeg -version
CUDA 相关错误
问题:GPU 加速不工作
解决方案:
- 确认安装 NVIDIA 驱动:
nvidia-smi - 确认安装 CUDA Toolkit:
nvcc --version - 检查 CUDA 版本兼容性(支持 11.8+)
- 如果仍有问题,可禁用 GPU 加速使用 CPU 模式
内存不足
问题:处理大文件时显存不足
解决方案:
- 减少 GPU 层数设置
- 使用较小的模型文件
- 分段处理长视频
- ✅ 基础字幕提取功能
- ✅ AI 翻译集成
- ✅ GPU 加速支持
- ✅ 图形用户界面
- ✅ 智能术语表
- 🔄 云端模型支持
- 🔄 输出双语字幕
- llama.cpp - 高性能 LLM 推理
- FFmpeg - 多媒体处理框架
- CustomTkinter - 现代化Python GUI 框架
- HuggingFace - 模型托管平台


