Abstract
Background: Since the development of word-recognition materials to test the transmission properties of auditory devices and human auditory systems, a carrier sentence or phrase (e.g., Say the word) has been used to preface the test word. For practical reasons, only the amplitude of the carrier phrase was somewhat controlled. The current American National Standards Institute standard for audiometers continues to specify the level of the test word should be the same communication level as the carrier phrase. Purpose: The development of an amplitude calibration protocol for use with short-duration speech signals that are characterized by substantial amplitude modulations is described. Research Design: Protocol 1 evaluated the average maximum root-mean-square (rms) amplitudes of 12.5-, 25-, 50-, and 100-ms voiced phoneme segments of each test word in 0.0227-ms increments to determine the segment duration to use. Protocol 2 used the 50-ms segment with the maximum rms amplitude among the 200 words in each list to normalize independently the amplitudes of the carrier phrases and test words to a target rms amplitude for each speaker. Study Sample: Digital copies of the 200 monosyllabic words in three versions of Northwestern University Auditory Test No. 6 (NU-6) and one version of the W-22 each spoken by a different speaker were evaluated using the numeric digital values transcribed from the audio files. Two iterations of the protocol were compiled. Data Collection and Analysis: In-house routines were used to analyze the waveform data, the results of which were evaluated with central tendency statistical analyses. Results: The finalized protocol is based on the rms amplitude of a 50-ms segment of the sustained, voiced phoneme of each test word. The protocol directly links the rms amplitudes of the calibration tone and of the 50-ms word segments as opposed to the currently used linking of the calibration tone rms amplitude to a peak meter deflection of the carrier phrase from which the amplitude of the test word is inferred. Conclusions: The effectiveness of the calibration protocol was demonstrated successfully on the four sets of word-recognition materials. The rms amplitude adjustments made independently to the individual carrier phrase and test-word utterances produced overall rms amplitudes for each of the four speakers that were homogenized slightly for the carrier phrases but substantially for many of the test words. Clinical Relevance Statement: The calibration protocol described provides an objective procedure that can be implemented and, most importantly, replicated with numeric accuracy to equate test-word (and carrier phrase) amplitudes among short speech signals like monosyllabic words and among speaker versions of those materials.