An outer mirror and a vehicle interphone are provided for transmitting a sound to the outside from a vehicle cabin and vice versa. Sound outside the vehicle cabin is transmitted to the cabin from a microphone-loudspeaker provided in a visor cover of the outer mirror. The microphone-loudspeaker is supported on a frame having higher rigidity than the visor cover, which is mounted to a vehicle body via an electric accommodating unit and a door mirror base to promote shock resistance. The visor cover includes a sound-conducting opening in its bottom portion, and the microphone-loudspeaker is made waterproof to promote weather resistance.