johunsang / kekedubing

What is kekedubing?

Kekedubing is a local-first web app for translating, dubbing, and captioning videos without touching the cloud. Drop in a local file or paste a YouTube URL, pick your source and target languages, choose a synthetic voice, and export a dubbed MP4 with burned-in subtitles. It also lets you draw rectangular blur boxes on the video for privacy editing. The backend runs Python with FastAPI, speech-to-text uses faster-whisper, translation uses Argos Translate local models, and video rendering uses FFmpeg. A Live Interpreter tab handles streaming URL dubbing in real-time.

Why is it gaining traction?

The pitch is simple: your videos never leave your machine. For creators tired of uploading content to cloud services, this is a privacy win. The Live Interpreter mode is the standout feature -- it groups Whisper segments into 3-7 second chunks, translates and synthesizes each one, then pads or trims the audio to fit the original timing. The result plays in sync with the muted video. It supports 1,800+ video sites via yt-dlp and offers 35 subtitle style presets, which covers most creator workflows without leaving the app.

Who should use this?

Content creators localizing videos for international audiences will get the most value. Privacy-conscious users who want to dub or subtitle without cloud dependencies should consider it. Developers building automated video pipelines can extend it since everything runs locally. It is not a polished SaaS product -- expect to run Docker commands and tune environment variables.

Verdict

At 10 stars, this is an early-stage project with minimal community validation. The 1.0% credibility score reflects that reality. Setup is straightforward with Docker, but long videos are CPU-intensive and first runs download models on demand. Worth trying if you need local video dubbing, but treat it as experimental rather than production-ready.