This repository contains a Speaker Diarisation Webservice powered by pyannote. and CLAM. This webservice is developed at the Centre of Language and Speech Technology, Radboud University, Nijmegen.
For end-users and hosting partners, we provide a container image that ships with a web interface. You can pull a prebuilt image from the Docker Hub registry using docker as follows:
$ docker pull proycon/diarisationservice
Run the container as follows:
$ docker run -v /path/to/your/data:/data --env HF_TOKEN=$HF_TOKEN -p 8080:80 proycon/diarisationservice
Ensure that the directory you pass is writable. Replace $HF_TOKEN with your Huggingface token and accept the license agreements for the following models:
- https://huggingface.co/pyannote/speaker-diarization-3.1
- https://huggingface.co/pyannote/segmentation-3.0
Assuming you run locally, the web interface can then be accessed on http://127.0.0.1:8080/.
If you want to deploy this service on your own infrastructure, you will want to set some of the environment variables
defined in the Dockerfile when running the container, most notably the ones regarding authentication, as this is by
default disabled and as such NOT suitable for production deployments.