Comment on Open source audio transcription model that obscures sensitive info in realtime