More Intelligence and Less Clouds in Our Smart Homes

Cloud-based services are useful, sometimes necessary for the users. However, especially in smart homes, exploiting cloud services for enabling multimodal human–computer interfaces, e.g., for speech recognition and synthesis, face recognition, face expression interpretation, and so on, is nowadays not necessary and, in several cases, disadvantageous for the end user. We argue that in those cases “local computing”, i.e., processing multimedia data locally, where they are acquired, is currently a better choice. We describe, as a proof of concept, the Red architecture, a totally local virtual assistant for home automation, and illustrate its abilities.