A local media transcription service that indexes a video library and transcribes it with a local Whisper server. Designed to run unattended on a home server.

Personal tooling project — works for my own media library, not packaged for general use.

How It Works

Scan — indexer walks the media directory, probes each video (duration, size), and records which files are already transcribed
Transcribe — the worker pulls pending files, sends audio to a local Whisper server, and saves transcripts next to the source
Continuous — a loop mode keeps the library up to date as new media arrives

Commands

python mt.py scan          # (re)index the media library
python mt.py transcribe    # process all pending videos once
python mt.py continuous    # run the worker in a loop
python mt.py status        # show library/transcription stats

Project Structure

Path	Description
`mt.py`	CLI entry point
`src/indexer.py`	Library scanning and indexing
`src/worker.py`	Transcription worker (one-shot and continuous)
`systemd/`	`media-transcriber.service` and `whisper-server.service` units
`setup.sh`	Initial environment setup

Deployment

Install the systemd units for an unattended setup:

sudo cp systemd/media-transcriber.service systemd/whisper-server.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now whisper-server media-transcriber

Requirements

Python 3.10+
A running Whisper server (see systemd/whisper-server.service)
ffmpeg for audio probing/extraction