A sound belt removably mounted on diapers includes a housing with two opposite flat surfaces on which one or more switch actuators are exposed. A removably mounted overlay with pictorial representations that cooperatively coincide with the switches permit a baby to activate them. Prerecorded sounds, including messages, are generated in response to the actuation of the switches and these messages are relevant to the pictorial representations so that the baby user learns what they means and aids him/her in communicating his/her desires and/or needs.