New automated AI speech-to-text integration for the NORDUnet on-premise Kaltura service

The Open AI Whisper-based Speech-to-Text integrated with the NORDUnet Kaltura on-premise service is using the large Whisper models currently whisper-large-v3 plus one additional local model per NREN.

All Nordic languages is enabled; Swedish, Norwegian, Danish, Finnish, and Icelandic, in addition English, French, German and Spanish is enabled, additional languages can be enabled.

We will continue to optimise and improve the user experience of the service.

The service is running on dedicated hardware located in the NORDUnet datacenters, with no data being transferred outside the NORDUnet on-premise environment.

Also see the description for the Speech2text transcription service

Information about Open AI Whisper can be found here: https://github.com/openai/whisper