transcribe-worker

Worker for the SUNET transcription service

Development environment setup

1. Create a virtual environment:

   python -m venv venv

2. Activate the virtual environment:

   source venv/bin/activate

3. Install the required packages:

   pip install -r requirements.txt

4. Edit the environment settings, should be in a file named .env. The following settings should be sufficient for most cases:

   DEBUG=True
   API_BACKEND_URL="http://localhost:8000"
   API_VERSION="v1"
   WORKERS=2
   WHISPER_CPP_PATH=<Path to whisper.cpp>
   FILE_STORAGE_DIR="<Your file storage directory>"

5. Build and install whisper.cpp, see https://github.com/ggml-org/whisper.cpp for details.

6. Download the needed Whisper models. From the transcriber-worker directory run:

   ./download_models.sh

7. Run the worker:

   python3 main.py --foreground --debug