Search for:

About Our Research

The framework guiding the programmatic line of research involves the interplay of three key components: basic research (understanding normal and impaired auditory perception), translational research (digital signal processing techniques for hearing aids and automatic speech recognition), and modeling (information-theoretic and neural models of perception). The central goal of this research is to improve speech understanding and decrease listening effort in hearing aid users. Engineering advances in hearing aid technology have far surpassed the current knowledge base in hearing sciences, thereby hindering the ability of clinical audiologists to make evidence-based decisions about the use of hearing aid features for individual patients. Hearing aids can change sounds in many ways that may or may not benefit speech understanding. Most research in this area relies on proprietary hearing aids and their features, making it difficult to establish causal relations between controlled changes in signal processing and an individual’s perception of those changes.

To address these problems, a sophisticated hearing aid simulator was developed, capable of replicating generic features in commercial hearing aids. The simulations are customized for each participant following clinically-accepted best practices. Flexible parameters include the filter bank, number of independent channels, and compression parameters. This simulator has been an instrumental tool in at least 14 peer-reviewed articles and 5 NIH grants. It has also led to the discovery of knowledge, resulting in three United States patents. These simulations and reverse engineering allow for conducting double-blind, randomized controlled trials with clinically realistic amplification and algorithms. In addition, they enable the exploration of why certain hearing aid features affect perception and what might explain individual differences in hearing aid benefit using laboratory measures and models of processing at the sensory, neural, and cognitive levels.

Past and Present Research Areas

Wide dynamic range compression (WDRC) is a ubiquitous feature in hearing aids used to repackage information in the amplitude domain to enhance signal audibility while maintaining listening comfort.  It is the last stage of processing after the incoming signals have been processed by hearing aid features like directionality, noise reduction, speech enhancement, filtering, frequency lowering, etc.  Since our research interests focus on enhancing these features or developing new ones, we created a hearing aid simulator.  While the hearing aid simulator was initially designed to be a tool of our research, it has also been the subject of our research on the effects of compression parameters on speech intelligibility.

Alexander, J.M., and Masterson, K.M. (2015).  Effects of WDRC release time and number of channels on output SNR and speech recognition.  Ear and Hearing, 36, e35-e49.

Rallapalli, V., and Alexander, J.M. (2019).  Effects of noise and reverberation on speech recognition with variants of a multichannel adaptive dynamic range compression scheme.  International Journal of Audiology, 58, 661-669.

Alexander, J.M., and Rallapalli, V., (2017).  Acoustic and perceptual effects of amplitude and frequency compression on high-frequency speech.  Journal of the Acoustical Society of America, 142, 908-923.

Brennan, M.A., McCreery, R., Kopun, J., Alexander, J.M., Lewis, D., and Stelmachowicz, P.G. (2016).  Masking release in children with hearing loss when using amplification.  Journal of Speech Language and Hearing Research, 59, 110-121.

Frequency lowering is used to move speech information and other environmental sounds from frequency regions where auditory transduction is very poor or nonexistent to lower frequency regions where sensory coding is better.  Every major hearing aid manufacturer worldwide has a version of frequency lowering in their hearing aids.  We have created a simulation of the nonlinear frequency compression technique used by the world’s largest hearing aid manufacturer.  This allows us to conduct the necessary parametric manipulations that control frequency lowering and relate their effects on speech acoustics to perception.  We have recently done the same for the latest frequency-lowering technique, adaptive nonlinear frequency compression, which forms the basis for my latest projects in the lab and models for speech perception.  We have also developed the patented inverse-frequency compression algorithm.  It is designed for those with severe-to-profound hearing loss who need frequency lowering the most but cannot benefit from current technology because it operates in a frequency range they cannot hear.  Finally, we have created and tested another version of frequency lowering that improves current methods for those with less than severe-to-profound hearing loss.

Alexander, J.M. (2019).  The s-sh confusion test and the effects of frequency lowering.  Journal of Speech Language and Hearing Research, 62, 1486-1505.

Alexander, J.M. (2016).  Nonlinear frequency compression: Influence of start frequency and input bandwidth on consonant and vowel recognition.  Journal of the Acoustical Society of America, 139, 938-957.

Alexander, J.M., Kopun, J.G., and Stelmachowicz, P.G. (2014).  Effects of frequency compression and frequency transposition on fricative and affricate perception in listeners with normal hearing and mild to moderate hearing loss.  Ear and Hearing, 35, 519-532.

“System and method for selective enhancement of speech signals” (Patent No. US 9,706,314), R.L. Jenison, K.R. Kluender, and J.M. Alexander, Wisconsin Alumni Research Foundation.  Patent issued: June 11, 2017.

“Enhancing Perception of Frequency-Lowered Speech” (Patent Nos. US 9,173,041 B2; US 10,083,702 B2), J.M. Alexander, Purdue Research Foundation.  Patents issued: October 27, 2015; September 25, 2018.

“Hybrid Expansive Frequency Compression for Enhancing Speech Perception for Individuals with High-Frequency Hearing Loss” (Patent No. US 11,961,529), J.M. Alexander, Purdue Research Foundation.  Patent issued: April 16, 2024.

    A significant challenge in advancing hearing aid performance is to overcome the distortion of frequency information important for understanding speech caused by damage to the inner ear.  Current hearing aid strategies that compress the amplitude and/or frequency range of speech can contribute further to the distortion.  We conducted research and made improvements to an algorithm that operates in real-time to enhance the salient features of speech.  The Contrast Enhancement (CE) algorithm implements dynamic compressive gain and lateral inhibitory sidebands across channels in a modified winner-take-all circuit, which together produce a form of suppression that sharpens the dynamic spectrum.  One innovative aspect of the CE algorithm is how it operates across successive speech sounds to enhance signature changes in their frequency composition.  Normal-hearing listeners identified spectrally smeared consonants and vowels in quiet and in noise.  The processing improved consonant and vowel identification, especially in noise.  For consonants, the most consistent improvement was for place of articulation.  This is encouraging for hearing aid applications because confusions between consonants differing in place are a persistent problem for listeners with sensorineural hearing loss.  The CE algorithm has also successfully enhanced automatic speech recognition, especially in noisy environments.

    Kwon, M. (2014).  Modification of Computational Auditory Scene Analysis (CASA) for Noise-robust Acoustic Feature.  Ph.D. Thesis, Purdue University.

    Alexander, J.M., Jenison, R.L., and Kluender, K.R. (2011).  Real-time contrast enhancement to improve speech recognition.  PLoS ONE, 6(9), e24630.  doi: 10.1371/journal.pone.0024630.

    “System and method for selective enhancement of speech signals” (Patent No. US 9,706,314 B2), R.L. Jenison, K.R. Kluender, and J.M. Alexander, Wisconsin Alumni Research Foundation.  Patent issued: June 11, 2017.

      Signal processing algorithms, including speech enhancement and frequency lowering, have been developed for individuals with sensorineural hearing loss in an attempt to partially restore degraded or absent speech cues.  Because signal processing techniques like these use more than simple gain and attenuation to recode speech, traditional or modified audibility-based models of speech intelligibility cannot fully capture the change in potential information associated with the increased saliency or distortion of particular speech cues.  These models quantify information in the acoustic signal in terms of how the excitation pattern or the neural firing along the cochlea changes over time.  Findings suggest that including information from slower modulations may improve predictions across a wider variety of conditions.

      Lllanos, F., Alexander, J.M., Stilp, C.E., Kluender, K.R. (2017).  Power spectral entropy as an information-theoretic correlate of manner of articulation in American English.  Journal of the Acoustical Society of America- Express Letters, 141, EL127.

      Rallapalli, V., Alexander, J.M. (2015).  Neural-Scaled Entropy predicts the effects of nonlinear frequency compression on speech perception.  Journal of the Acoustical Society of America, 138, 3061-3072.

      Stilp, C.E., Alexander, J.M., Kiefte, M., Kluender, K.R. (2010).  Auditory color constancy: Calibration to reliable spectral properties across speech and nonspeech contexts and targets.  Attention, Perception, and Psychophysics, 72, 470-480.

      Stilp, C.E., Kiefte, M., Alexander, J.M., Kluender, K.R. (2010).  Cochlea-scaled spectral entropy predicts rate-invariant intelligibility of temporally distorted sentences.  Journal of the Acoustical Society of America, 128, 2112-2126.

      The models and algorithms identified in the previous contributions build off theories about how various processes code speech information along the auditory pathway.  These theories assume that sensory systems are efficient information processors that respond best to changes in the signal across time and frequency.

      DeRoy Milvae, K., Alexander, J.M., and Strickland, E.A. (2021).  The relationship between ipsilateral cochlear gain reduction and speech-in-noise recognition at positive and negative signal-to-noise ratios.  Journal of the Acoustical Society of America, 149, 3449-3461.

      Plotkowski, A., and Alexander, J.M. (2016).  A sequential sentence test paradigm using revised PRESTO sentence lists.  Journal of the American Academy of Audiology, 27, 647-660.

      Alexander, J.M., and Kluender, K.R. (2010).  Temporal properties of perceptual calibration to local and broad spectral characteristics of a listening context.  Journal of the Acoustical Society of America, 128, 3597-3613.

      Alexander, J.M., and Kluender, K.R. (2009).  Relativity of spectral tilt change in stop consonant perception by hearing-impaired listeners.  Journal of Speech, Language, and Hearing Research, 52, 653-670.