Alphabet-oriented and user-oriented source coding theories

The aim of this, and of the follow-up paper [16], is to present the mathematics used in the technical field of speech coding. This paper remains mostly within a ,,classical" source coding theory based on alphabetic distortion measures. We prove a new general joint source/channel coding theorem for these measures. By an example of two users we demonstrate that the "classical" theory, which is in the above stated sense alphabet-oriented, need not always be user-oriented. The first of these users is a statistician and the second one is a telephone owner. At the end we prove a statement which opens the possibility to build up a spectrum-oriented source coding theory, which is user-oriented for a special category of users, e.g. for the telephone owner. Such a theory is presented in [16].