deusjin

deusjin / subforge

Public

Rust CLI for AI subtitle workflows: transcribe, segment, translate, evaluate, and burn or mux subtitles.

32
4
89% credibility
Found May 31, 2026 at 32 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

SubForge is a video subtitle pipeline tool that takes a video file and automatically produces translated, quality-checked subtitles — either burned directly into the video or attached as a separate subtitle track. It handles the full workflow from speech recognition through translation quality assurance to final video rendering, making subtitle production repeatable and organized rather than a one-off mess of scripts and manual work.

How It Works

1
🎬 You have a video that needs subtitles

You just recorded a video and realize it needs subtitles for a wider audience. Maybe it's a tutorial, a vlog, or content you want to share with people who speak different languages.

2
⚙️ Everything gets set up automatically

You run a simple setup command and the tool installs everything it needs on its own — speech recognition, translation tools, and video processing all configured together without you having to figure out any of it.

3
🎤 Your video gets transcribed

The tool listens to your video and converts the speech into text, breaking it into natural subtitle segments that match how people actually talk.

4
🌐 Your subtitles get translated

You pick the language you want and the tool translates your subtitles, keeping important terms consistent and checking that the translation quality is good.

5
You choose how to deliver your subtitles
🔥
Burn subtitles into the video

Subtitles are permanently embedded in the video — great for platforms like YouTube where you want them always visible.

📎
Attach as a separate track

Subtitles stay as a separate file attached to the video — viewers can toggle them on or off, and the video stays in original quality.

🎉 Your subtitled video is ready

Your video now has professional bilingual subtitles, burned in or attached, and you're ready to share it with the world.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 32 to 32 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is subforge?

SubForge is a Rust CLI that turns video subtitle production into a reproducible pipeline. You feed it a video, and it handles the entire chain: speech-to-text transcription, intelligent subtitle segmentation, translation via Google, Bing, or LLM backends, quality estimation with automatic refinement, and finally burns subtitles into the video or muxes them as a soft subtitle track. It solves the problem of juggling scattered scripts, model paths, ffmpeg flags, and manual rework every time you need subtitles.

Why is it gaining traction?

The hook is that it makes the whole workflow explicit and repeatable instead of a pile of one-off scripts. Local faster-whisper transcription with CUDA support means you can skip cloud API costs and latency. The LLM translation pipeline includes terminology extraction, translation memory, and GEMBA-MQM quality scoring that automatically retries low-quality segments. For creators who process videos repeatedly, that reproducibility matters. The `subforge process video.mp4` one-liner that handles everything end-to-end is the killer feature developers keep mentioning.

Who should use this?

Video localization teams, course creators, and content localizers who process the same type of content repeatedly. If you're burning through videos and getting tired of duct-taping together transcription, translation, and video encoding tools, SubForge replaces that stack. It's less useful for one-off subtitle tweaks where you just need to tweak timing or fix a line.

Verdict

SubForge is a well-architected Rust CLI with solid fundamentals (160+ tests, CI pipeline, proper error handling), but at 32 stars and version 0.2.0, it's still early-stage software. The credibility score of 0.8999999761581421% reflects that maturity gap. If you want a turnkey subtitle pipeline and don't mind building from source, it's worth trying. Just expect to do some manual configuration and keep an eye on releases as the project evolves.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.