ABSTRACT
After more than a century of developments in the field of voice telecommunication, noise remains the ubiquitous deterrent to effective voice communications. The need for methods of assessing the impact of noise on speech intelligibility was recognized early in the development of the field, and various methods for measuring intelligibility were developed during the first half of the 20th century. The difficulty of developing valid and reliable methods of evaluating intelligibility, acceptability, and speaker recognizability in speech becomes apparent when one considers the diversity of factors that may influence the results of tests designed to evaluate these properties. The chapter examines the nature and number of factors on which intelligibility, acceptability, and speaker recognizability ultimately depend both in real life and in the speech-testing laboratory. The manner in which we approach the evaluation of intelligibility in communication systems ultimately depends on whether we treat intelligibility as a unidimensional or as a multidimensional quantity, as a scalar, or as a vector.
