SubForge

English

简体中文

English

简体中文

Appearance

SubForgeProfessional Video Subtitle Processing

Open Source · LLM-Powered · Process 14-minute video in 4 minutes, cost less than $0.002

Get Started
GitHub Repository
SubForge
⚡

Lightning Fast, Ultra Low Cost

Process 14-minute video in just 4 minutes, cost less than $0.002. Powered by Whisper + LLM stack for quality and speed.

🧠

LLM-Powered Intelligence

Beyond speech recognition. LLM semantic segmentation, auto error correction, terminology unification, expression optimization for professional results.

🌐

Multilingual Support

Recognize 99 languages, translate to 37 languages. Reflection translation mechanism ensures quality with precise timeline alignment.

📖

Fully Open Source & Free

GPL-3.0 license, no hidden fees. Run locally for complete data privacy control. Community-driven continuous improvement.

💻

No High-End Hardware

Run Whisper on CPU, optional GPU acceleration. Choose between cloud API or local offline processing.

📦

Batch Processing

Drag and drop media for automatic transcription, subtitle restructuring, and translation with live task progress.

✍️

Subtitle Review & Export

Review timestamps, source text, translations, and quality warnings in one workspace, then save or export clean SRT subtitles.

🔧

Advanced Features

VAD voice detection, vocal separation, word-level timestamps, manuscript matching, custom prompts, and more.

🖥️

Cross-Platform Desktop App

Windows and macOS desktop installers use a Next.js interface with a local FastAPI backend, real-time preview, and subtitle editing. Linux remains available through the CLI and source installation.

Interface Preview ​

Software Interface Preview

Quick Experience ​

bash
# Download and run the installer directly
# Or run from source
git clone https://github.com/henry1786580051-lang/SubForge.git
cd SubForge
run.bat
bash
# Use automatic installation script
git clone https://github.com/henry1786580051-lang/SubForge.git
cd SubForge
chmod +x run.sh
./run.sh

Why Choose SubForge? ​

  • 🎯 Efficient Processing: Full processing of a 14-minute video takes only about 4 minutes, costing less than ¥0.01
  • 🌟 Quality Assurance: Uses advanced Whisper models and large language models to ensure subtitle quality
  • 💡 Intelligent Optimization: Automatically corrects typos, unifies terminology, optimizes expressions
  • 🚀 Easy to Use: Drag and drop videos for fully automatic processing, no complex configuration needed

Core Features ​

Speech Recognition & Transcription ​

  • Supports Whisper API, FasterWhisper, WhisperCpp, and other engines
  • Supports 99 language recognition
  • Supports VAD (Voice Activity Detection)
  • Supports vocal separation

Intelligent Subtitle Processing ​

  • LLM semantic segmentation for smoother reading
  • Automatically optimizes terminology, code snippets, mathematical formulas
  • Supports manuscript matching to improve accuracy
  • Precise subtitle timeline alignment

High-Quality Translation ​

  • Supports LLM translation, Google Translate, Microsoft Azure Translator, and DeepLX
  • Reflection translation mechanism improves translation quality
  • Maintains complete timeline consistency
  • Supports 37 target languages

Subtitle Output ​

  • Keeps source media unchanged
  • Supports SRT, ASS, TXT, and JSON subtitle output
  • Provides bilingual layout and timeline-quality checks

Get Started ​

Ready to begin? Check out the Getting Started Guide to learn how to use SubForge.

Released under the GPL-3.0 License

Copyright © 2024-present henry1786580051-lang