pluja/whishper

What it solves

Whishper provides a completely local, private way to transcribe audio and video files into text and create subtitles without relying on cloud services.

How it works

It uses a suite of components including a web-based user interface, a backend coordinator, and a transcription API powered by FasterWhisper. Users can upload files or provide URLs (via yt-dlp) for processing. For translation, it integrates with LibreTranslate, and it uses MongoDB for data storage, all running on the user's own hardware (CPU or NVIDIA GPU).

Who it’s for

Content creators, privacy-conscious users, and anyone needing a self-hosted tool for audio transcription and subtitle editing.

Highlights

• 100% Local: All transcription, translation, and editing happen on your own machine, enabling offline use.
• Integrated Subtitle Editor: Features tools for segment splitting, insertion, and characters-per-second (CPS) warnings.
• Flexible Input: Supports direct file uploads and URLs from any source compatible with yt-dlp.
• Multiple Export Formats: Transcriptions can be downloaded as TXT, JSON, VTT, or SRT.