Most otoacoustic emissions are relatively weak as measured in the ear canal. A system noise level of −20 to −30 dB SPL is required in order to measure some emissions of interest, which means microphone noise spectrum levels of −20 dB SPL or better are needed. In the case of stimulated emissions such as the cubic distortion product at the frequency 2f1−f2, the distortion product is often more than 80 dB below the level of the f1 and f2 primary tones. As a result, the intermodulation distortion requirements on the amplifiers and earphones are severe−typically 0.003% or so−requiring the use of two separate earphones which must be acoustically isolated from each other. Special transducer constructions and couplings designed to meet these needs will be described, as will a new set of transducers specifically designed to be used in an inexpensive infant screening device which can be carried in the pocket like a stethoscope.