In a camera, apparatus and methods for recording or storing voiced in messages in message memory associated with an exposed image frame for reproduction as an annotation with prints made from the exposed image frames. The camera includes a voice recognition system for voice recognition of words spoken by the user before the words are stored in message memory. Fixed and adjustable vocabularies are provided for use in the voice recognition. The adjustable vocabulary may be loaded into the camera by a vocabulary memory card or through an interface with a docking station of a personal computer or vending machine so that an adjustable vocabulary of words of interest to the camera user or related to a scene or event of photographic interest may be employed in the voice recognition. The messages may be recorded on magnetic film layers or stored in camera on-board memory or in detachable message memory modules to be forwarded with the exposed film strip to a photofinisher for read out and printing on the prints. The messages may be stored or recorded by the user in real time with each exposure or at an earlier or later time.