Abstract :
Testing methods, such as `articulation¿, `immediate appreciation¿, etc., at present available for providing a direct measure of the performance or effectiveness of a link for speech communication are not individually or even collectively applicable over any wide ranges of levels of performance or types of link. Many practical links fall outside the range over which these methods are reliable, and there is need for a general method that will both provide a safeguard against unrealistic results and serve, if possible, to extend the usefulness of existing techniques. The paper examines the mechanism of conversation over a speech link and deduces a sequence of criteria each applicable over an appropriate range of performance. A general assessment method follows from this, and the corresponding experimental techniques (which depend on the use of ordinary untrained subjects) and some applications are discussed. Some existing assessment methods can be fitted into the framework provided by the general method and can then be used with increased confidence. Assessments as described in the paper are of little value for indicating the merit a link has for the user. Their most valuable use is to facilitate (as part of the process known as rating) realistic comparisons of links with Standard Speech Links whose evaluation can be studied in actual use.
