Open source speech foundation model that runs locally on CPU in real-time