Efficacy of linear frequency transposition on consonant identification in quiet and in noise

Frequency transposition has gained renewed interest in recent years. This type of processing takes sounds in the unaidable high-frequency region and moves them to the lower frequency region. One concern is that the transposed sounds mask or distort the original low-frequency sounds and lead to a poorer performance. On the other hand, experience with transposition may allow the listeners to relearn the new auditory percepts and benefit from transposition.

The current study was designed to examine the effect of linear frequency transposition on consonant identification in quiet (50 dB SPL and 68 dB SPL) and in noise at three intervals--the initial fit, after one month of use (along with auditory training), and a further one month of use (without directed training) of transposition.

A single-blind, factorial repeated-measures design was used to study the effect of test conditions (three) and hearing aid setting/time interval (four) on consonant identification.

Eight adults with a severe-to-profound high-frequency sensorineural hearing loss participated.

Participants were fit with the Widex m4-m behind-the-ear hearing aids binaurally in the frequency transposition mode, and their speech scores were measured initially. They wore the hearing aids home for one month and were instructed to complete a self-paced "bottom-up" training regimen. They returned after the training, and their speech performance was measured. They wore the hearing aids home for another month, but they were not instructed to complete any auditory training. Their speech performance was again measured at the end of the two-month trial.

Consonant performance was measured with a nonsense syllable test (ORCA-NST) that was developed at this facility (Office of Research in Clinical Amplification [Widex]). The test conditions included testing in quiet at 50 dB SPL and 68 dB SPL, and at 68 dB SPL in noise (SNR [signal-to-noise ratio] = +5). The hearing aid conditions included no transposition at initial fit (V1), transposition at initial fit (V2), transposition at one month post-fit (V3), and transposition at 2 months post-fit (V4). Identification scores were analyzed for each individual phoneme and phonemic class. Repeated-measures ANOVA were conducted using SPSS software to examine significant differences.

For all test conditions (50 dB SPL in quiet, 68 dB SPL in quiet, and 68 dB SPL in noise), a statistically significant difference (p < 0.05 level) was reached between the transposition condition measured at two months postfitting and the initial fitting (with and without transposition) for fricatives only. The difference between transposition and the no-transposition conditions at the 50 dB SPL condition was also significant for the initial and one-month intervals. Analysis of individual phonemes showed a decrease in the number of confusions and an increase in the number of correct identification over time.

Linear frequency transposition improved fricative identification over time. Proper candidate selection with appropriate training is necessary to fully realize the potential benefit of this type of processing.