We research the benefit of performing split model inference on speech models using the edge devices and the cloud infrastructure