It’d be amazing if speech-to-text could take into account context as well: Greek if I’m speaking Greek, Korean if I’m speaking Korean, or for (int i = 0; i < count; ++i) if I’m dictating code.