Abstract: A vocabulary memory which contains descriptions of vocabulary elements is addressable by a code which is formed by a number designating a vocabulary element, but the set of useful codes for a given user comprises gaps (32-38) between groups of consecutive numbers which are referred to as segments (24-31). The vocabulary memory contains segments which are arranged end to end and the receiver comprises an auxiliary memory which contains, for a given segment (27), the smallest and largest code values of this segment and the position in the vocabulary memory of descriptions corresponding to the codes contained in this segment, as well as the addresses (225, 230) in said auxiliary memory of data of two other segments (25, 30) of the sequence of segments, some of which are situated above and some of which are situated below said given segment etc. in order to form a classification in the form of a tree.

Abstract: The speech synthesizer generates signals for a loudspeaker on the basis of symbols defining the words to be reproduced.In accordance with the invention, the same speech synthesizer is provided with means for processing symbols of three different types, i.e.:a type representing phonetically the elements of a word to be reproduced,or a type formed by a number designating a quantity or the designation of a major route,or a type formed by a number designating a sequence prestored in a memory of the speech synthesizer and corresponding to a phrase part.A series of symbols arriving at the speech synthesizer comprises successive sequences of symbols which are arranged end to end, said sequences being of three types, a type recognition character ($, .vertline., ) being placed at each end of a sequence of a given type.