Focused error correction and/or focused error detection is used in the information coding system according to the invention. The purpose of the present invention is to present a speech encoding method, in which the number of speech parameter bits on which error correction coding and/or error detection coding focuses is automatically adjusted in relation to the number of total speech parameter bits as the function of the quality of the information transfer connection. In the information encoding system according to the invention there is no need to reduce the number of bits used for speech encoding, due to which the voice quality of the speech remains high. In the information encoding system according to the invention the error correction and/or error detection is focused on the bits most important for the voice quality e.g. as the function of the C/I (Channel to Interference) - parameter describing the quality of the information transfer connection. The muting of speech synthesizing occurring in prior known systems on poor information transfer connection is in the information encoding method according to the invention reduced by using focused error detection.