MediaTranscribe

The Problem

Every decent transcription tool wants you to upload your recording to someone else's server and pay by the minute. That's a hard sell when the audio is a client call, an interview, or anything you'd rather not hand over.

The Solution

I built a transcription app that runs entirely on your own machine. Whisper does the speech-to-text locally, FFmpeg pulls the audio out of video files, and a Redis-backed queue handles the long jobs so the browser never sits there waiting.

The Results

• Local transcription with whisper.cpp, no external API
• FFmpeg audio extraction and validation, uploads up to 300 MB
• Background queue with live processing status
• Editable segments with timestamps, exported as SRT or TXT