Actually basic voice recognition is done in the TV itself. By 'basic' I mean things like switching it on or off, changing channels, adjusting volume or navigati

Actually basic voice recognition is done in the TV itself. By 'basic' I mean things like switching it on or off, changing channels, adjusting volume or navigating the menu. Also this is limited to a few selected languages (most notably English and Korean, not sure which others). As soon as you get to things like entering text into text inputs or want to control your TV in 'unsupported' language, this gets delegated to cloud services.