John R. Bareham
Consultant in Electroacoust., 20 Rogers Ave., Marlborough, MA 01752
Hands-free telephones often are designed with performance features that make it difficult to obtain meaningful electroacoustic measurements using simple methods. Compressors, limiters, threshold detectors, noise guards, and other more complex speech-sensitive functions change the telephone's electroacoustic characteristics depending on the type and level of signal applied to it. To realistically account for such variables, electroacoustic characteristics of several hands-free telephones were measured using a binary method, in which two signals were applied to the device simultaneously. The first signal was a bias, which temporarily caused the telephone to operate in a well-defined manner. Bias signals used in this work included noise, noise bursts, and speech fragments. The second signal was used for the actual measurement. It was applied at a low level relative to the bias, so the effect of the bias was not disturbed. The measurement signal used was a series of very fast sine sweeps, processed using an enhanced time delay spectrometry algorithm. This method was chosen to enable simulated free-field measurement of the telephone in an ordinary room. The measurement setup will be described, and measurement results will be shown and explained. Benefits and limitations of the method will be discussed.