README.md

transcribee-worker

a transcription worker for the transcribee open-source transcription platform

This component connects to the backend, picks up jobs and runs transcription / forced-alignment / speaker-identification jobs. It thus does all the CPU / GPU intense heavy lifting.

Installation

We use pdm for dependency management. To install all dependencies locally, run:

pdm install

Configuration

You can configure the worker by setting environment variables. For a full list of those, see the attributes of the Settings class in config.py Instead of setting environment variables, you can also specify them in the .env-file. They will automatically be read from there. See .env.example for an example of how this file could look.

The most important settings are explained here:

The worker needs multiple models, which will automatically be downloaded if they to not exist yet. They will by default be stored in ./.data/models. You can change this directory by setting MODELS_DIR