Extant Problems in Analysis-Synthesis Telephony
James L. Flanagan · The Journal of the Acoustical Society of America · 1959
Efficient voice communication implies transmission of the minimum information necessary to specify a speech event and to evoke an appropriate response. Human speakers generate information at rates which are much less than the capacity of conventional telephone channels. This disparity has motivated considerable research in analysis-synthesis techniques for speech transmission. One premise is common to most of the analysis-synthesis schemes; namely, the transmission system should be constrained so that the appropriate message ensemble contains only the sounds of human speech. One attempts, therefore, to build into the communication link as many as possible of the constraints characterizing the production and the perception of speech. Further, one hopes to manipulate the parameters of the constraints with as few independent, low information-rate signals as possible. This paper attempts to review briefly the progress which has been made with analysis-synthesis telephony, and endeavors to appraise the current status of the research. It undertakes to suggest how recent work in speech production may find application in transmission systems, and to point up some of the problems confronting further development. Finally, it considers briefly the potentialities of simulating speech systems with digital computers.