The intelligibility of speech declines at high intensities, both for normally hearing and hearing impaired listeners, but for the hearing impaired this """"""""rollover"""""""" effect poses a special problem. For these listeners, the levels of hearing aid amplification required to make speech audible in a region of hearing loss may produce sufficient rollover to result in either no intelligibility gain or even some loss as a result of amplification However, a recent discovery in this laboratory indicates that rollover can be defeated and intelligibility preserved by reducing speech to an array of noncontiguous speech bands having vertical filter slopes (i.e., rectangular bands) and widths substantially narrower than a critical band. Preliminary experiments employing normally hearing listeners presented with 500 Hz low-pass speech, having an intelligibility of about 13 percent, have shown that adding an array of ten 4 percent wide rectangular speech bands, spaced at 1/3-octave intervals from 1000 to 8000 Hz, produced a 75 percent intelligibility increase both at a moderate level and when the array was presented at an intensity previously found to produce substantial rollover for wideband speech in the same frequency range. Further planned experiments will employ normally hearing listeners to determine the bandwidths and inter-band spacings (band densities) for the stimulus arrays that provide the greatest intelligibility with the least vulnerability to rollover at high intensities. Then, in collaboration with the director and audiologists at Community Audiology Services of UWM, experiments with hearing-impaired listeners will be conducted to evaluate the effectiveness of replacing wideband speech in regions of hearing loss with amplified arrays of sub-critical width bands. Listeners will have audiologically normal or minimally impaired hearing sensitivity up to 500 Hz and will have bilaterally symmetrical, mild-to-profound sensorineural hearing loss at higher frequencies, with audiometric configurations ranging from flat to steeply sloping. A series of experiments will compare the new reduced-spectrum amplification strategy with full-spectrum amplification to determine how the effectiveness of the new strategy varies with the severity and frequency region of hearing loss. Additional experiments will determine the effects of other factors, including the type of high-frequency emphasis and compression applied to the rectangular band arrays. A second, related set of experiments will use a methodology developed in this laboratory to examine the role of lateral suppression as a means of further enhancing the control of rollover across the sub-critical band arrays. A final experiment will use rectangular passbands to directly measure the effective dynamic range of speech across the spectrum, which can only be estimated by the Speech Intelligibility Index (SII) procedure used widely to estimate passband intelligibility for the design and evaluation of auditory prostheses. The findings from this experiment will inform our choice of relative levels for the subcritical-widh bands used for the reduction of rollover, and may also provide supplementary information for use in calculation by the SII.

Public Health Relevance

The prevalence of hearing impairment in the US is estimated to be nearly 50 percent for adults 48 to 87 years of age. Unfortunately, the effectiveness of hearing aids, used to amplify speech in order to improve its audibility for these listeners, is hampered by the loss of intelligibility, known as the rollover effect that normally occurs when speech is presented at high intensities. The present study will evaluate a new speech processing strategy that may substantially increase the benefit of hearing aid amplification by reducing or eliminating the rollover effect. !

National Institute of Health (NIH)
National Institute on Deafness and Other Communication Disorders (NIDCD)
Research Project (R01)
Project #
Application #
Study Section
Language and Communication Study Section (LCOM)
Program Officer
Shekim, Lana O
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Wisconsin Milwaukee
Schools of Arts and Sciences
United States
Zip Code
Warren, Richard M; Bashford Jr, James A; Lenz, Peter W (2017) Critical bandwidth speech: Arrays of subcritical band speech maintain near-ceiling intelligibility at high amplitudes. J Acoust Soc Am 141:EL222
Bashford Jr, James A; Warren, Richard M; Lenz, Peter W (2017) Maintaining intelligibility at high intensities with arrays of subcritical width speech bands and interpolated noise. J Acoust Soc Am 142:EL299
Bashford Jr, James A; Warren, Richard M; Lenz, Peter W (2015) How broadband speech may avoid neural firing rate saturation at high intensities and maintain intelligibility. J Acoust Soc Am 137:EL340-6
Bashford Jr, James A; Warren, Richard M; Lenz, Peter W (2013) Maintaining intelligibility at high speech intensities: evidence of lateral inhibition in the lower auditory pathway. J Acoust Soc Am 134:EL119-25
Warren, R M; Bashford, J A; Lenz, P W (2013) How Broadband Speech May Avoid Neural Firing Rate Saturation at High Intensities and Maintain Intelligibility. Proc Meet Acoust 13:3426
Warren, Richard M; Bashford Jr, James A; Lenz, Peter W (2013) When intelligibilities of paired speech bands do not behave the way they are supposed to. J Acoust Soc Am 134:EL244-50
Bashford, J A; Warren, R M; Lenz, P W (2013) When Spectral Smearing Can Increase Speech Intelligibility. Proc Meet Acoust 19:60118-60124
Bashford Jr, James A; Warren, Richard M; Lenz, Peter W (2011) Enhancing the intelligibility of high intensity speech: Evidence of inhibition in the lower auditory pathway. Proc Meet Acoust 12:
Warren, Richard M; Bashford, James A; Lenz, Peter W (2011) An alternative to the computational Speech Intelligibility Index estimates: direct measurement of rectangular passband intelligibilities. J Exp Psychol Hum Percept Perform 37:296-302
Bashford, James A; Warren, Richard M; Lenz, Peter W (2010) When noise vocoding can improve the intelligibility of sub-critical band speech. Proc Meet Acoust 9:60001-600019

Showing the most recent 10 out of 19 publications