Peggy B. Nelson - US grants

Factors affecting hard-of-hearing (HOH) listeners' use of formant transitions in speech will be investigated in three experiments. Listeners will be selected on the basis of: 1) hearing loss: (a) mild-to- moderate or (b) moderate-to-severe, 2) audiometric configuration: (a) flat or (b) sloping, and 3) amount of auditory experience: (a) extensive, or (b) limited, as define by age of hearing loss onset, years of amplification use, and home/school mode of communication. The three experiments will use different approaches for the study of formant transition use by HOH listeners. These are: 1) investigation of listeners' use of formant transitions for glide and voiced stop consonant identification as a function of hearing loss and auditory experience. Spoken stimuli will be modified so that formant transitions are the only available acoustic cues for consonant identification. Transition audibility will be determined based on threshold measurements for each of the transition onset and offset frequencies. An audibility metric will be developed and applied to describe the effects of transition audibility on transition use. The relationships between use of formant transitions and the listeners' auditory experience will also be examined. 2) comparison of listeners' frequency/temporal discrimination for transition-like stimuli with transition use for consonant identification. The transition-like stimuli will simulate the transition characteristics found in the glide/stop identification stimuli. Relationships between transition discrimination and transition use will be described. 3) investigation of HOH listeners' perceptual weighting of formant transitions compared to other acoustic cues to the stop/glide contrast. Both synthesized and natural speech stimuli will be developed that vary along two acoustic continua (formant transition and amplitude onset). Analysis of the perceptual importance of the two acoustic cues, auditory experience, and degree of hearing loss will be completed. In all experiments, listeners with normal hearing thresholds will be used as controls.

The overall goal of this project is to evaluate promising signal processing techniques to enhance speech spectral characteristics for improved speech understanding by hard-of-hearing (HoH) listeners. These spectral enhancements include: 1) deepening of between-formant spectral valleys, 2) frequency scaling of speech, and 3) combinations of both deepening and scaling. Many hard-of- hearing (HoH) listeners can resolve low-frequency spectral cues normally but they are often less able to resolve frequency information in higher-frequency signals. In speech, higher- frequency second formants (F2), which provide important information for phoneme distinctions, are not easily resolved by these HoH listeners. Sinusoidal modeling techniques will be used to detect spectral peaks in rapidly changing syllables, and to deepen spectral valleys between peaks. Sound design software will be used to accomplish modest frequency scaling (up to 30 percent reduction) while retaining good sound quality. It is anticipated that these combined schemes will produce good-quality speech signals with formants that are more easily resolved by HoH listeners. Two groups of listeners with moderate to severe hearing loss will be identified as having better or poorer spectral resolution abilities based on frequency discrimination and spectral peak detection tasks. These listeners will then identify unmodified stimuli and stimuli that have been enhanced by the proposed processing chemes. Results will demonstrate the benefits of each scheme individually, and of the combined schemes. Early results using pilot processed stimulil show promise toward the goal of improving speech perception abilities of listeners with moderate to severe hearing loss. Results from the proposed investigation will provide important additional pilot data that will serve as a basis for future investigations of speech enhancements. A long-term goal of this research is the development of real-time enhancement algorithms that can be implemented into wearable amplification devices.

DESCRIPTION (provided by applicant): Listeners with sensorineural hearing loss experience significantly more masking and significantly less release from masking than do listeners with normal hearing sensitivity. When background noise is intermittent, normal- hearing (NH) listeners take advantage of the momentary dips in the noise to understand the target signal. This benefit is termed masking release (MR). Hearing-impaired (HI) listeners do not experience the same degree of MR as NH listeners. As a result, HI listeners cannot understand speech in noisy environments at the same signal-to-noise ratios in which NH listeners can understand successfully. At the same time, listeners with hearing loss frequently complain that their amplification devices are not satisfactory in background noise. We propose to further our understanding of the masking and MR experienced by listeners with hearing loss and to investigate the causes and potential sources of remediation. In Experiment 1 we will test MR in speech recognition experiments across a wide range of noise and signal levels and bandwidths. We will test NH and HI listeners, as well as NH listeners whose thresholds will be elevated by noise to match those of the HI listeners. We will measure listeners'sentence recognition in quiet, steady noise, and fluctuating noise and will define MR as the performance difference between steady and fluctuating noise. We hypothesize based on preliminary results that signal audibility will not fully explain the reduced MR of HI listeners, and that additional variability is explained by HI listeners'reduced access to spectrotemporal information about the speech in the dips of noise. We further hypothesize that underlying changes in cochlear physiology can explain HI listeners'difficulty in conditions of intermittent noise, and we propose to measure cochlear function in Experiment 2. We will measure basilar membrane input/output responses, masking period patterns, modulation masking, and frequency selectivity to describe HI listeners'cochlear function. We will use results from Experiment 2 to compute the adjusted effective audibility experienced by HI listeners taking into account the loss of cochlear gain experienced by HI listeners. In Section 3 we will model and simulate the effects of cochlear hearing loss based on speech recognition and psychoacoustic test results obtained from the same listeners. We hypothesize that these models and simulations will provide accurate and useful predictions of MR in young HI listeners. We will use the resulting models and simulations as empirically based first steps in evaluating signal processing for noise reduction. In later years the models can also be applied further to listeners with severe hearing loss, and to elderly listeners with hearing loss. Overall, the findings will contribute to our understanding of the nature of hearing loss, the effects of reduced audibility and cochlear gain, and the implications for understanding speech in a variety of noise backgrounds. We hypothesize that improving our understanding of the nature of reduced masking release will improve our ability to develop signal processing algorithms that can more effectively remediate HI listeners'difficulties in noise. PUBLIC HEALTH RELEVANCE Results from the proposed experiments and models will help understand hearing-impaired listeners'difficulty understanding speech in background noise. We anticipate that models developed in this project will lay the groundwork for evaluating signal-processing algorithms to maximize speech intelligibility in noise.

DESCRIPTION (provided by applicant): Hearing aid users experience considerable dissatisfaction with hearing aids in noise. Current audiogram-based methods for setting hearing aid parameters (multifrequency gain and compression) are based on predicted loudness and/or intelligibility of speech in quiet; there is no research-based prescription method for setting parameters for speech in noise. At the same time, there is a growing body of evidence that hearing-impaired persons vary considerably in their ability to understand speech in noise, even when their audiograms are very similar. Given this variance in speech understanding in noise, it is reasonable to expect that the variance across hearing aid users in the optimal hearing settings for noisy environments will be at least as large. If this is the case, then the current methods for setting hearing aid parameters for use in noise are almost certainly producing suboptimal results for many hearing aid users. The advent of wireless control of hearing aids provides an opportunity to address directly questions related to the hearing aid parameter settings that users find most beneficial. Availability of a sophisticated system for simulating different sound environments will make is possible to collect a substantial body of data on hearing aid setting preferences in different sound environments that are still under close experimental control. Work proposed here is a systematic exploration of how preferred hearing aid settings vary across hearing aid users and across sound environments. The following hypotheses will be tested: a. Preferred settings are not the same for quiet settings as they are for noisy settings. b. Variance in preferred settings among hearing aid users is larger for noisy settings than it is for quiet settings. c. Preferred noisy settings can be predicted from clinical nd laboratory measures of speech understanding in noise. d. User-adjusted settings will provide greater perceived user benefit in both quiet and noisy situations than settings derived using current clinical practices for setting hearing aids. e. Some users will obtain benefit from hearing aids that allow at-will user adjustments in the field. Overall, results will improve hearing aid outcomes by providing new methods for setting hearing aids for noisy settings, and for allowing users to set their own hearing aids in daily use. Both outcomes have the potential of increasing satisfaction with hearing aids by increasing the perceived benefit of hearing aid settings made in the clinic and in everyday life.

The human senses provide the information to the brain about the world around us. Deficits in visual, auditory or other sensory inputs have a major impact on quality of life, including education, employment and social engagement, which in turn places a major burden on the US economy. The scale of this problem is large: According to reports by the World Health Organization (WHO) and the National Institutes of Health, there are approximately 5 million Americans with vision impairment and around 36 million Americans with some form of hearing loss. There is broad recognition of the need for interdisciplinary collaboration for translational research on disabilities. Research into the development of more effective assistive technologies and environmental modifications requires interdisciplinary expertise that unites a fundamental understanding of the basic sensory science (vision, audition, motor control, speech and language) with deep technical expertise in engineering, computer science, and other related fields. This National Science Foundation Research Traineeship (NRT) award to the University of Minnesota will enable the formation of an integrated and interdisciplinary training program in sensory science through the joint strengths of the university's Center for Cognitive Sciences and Center for Applied and Translational Sensory Science. The program's goal is to prepare future scientists to combine engineering approaches with scientific knowledge and methods so that they are in a position to develop the next generation of sensory aids that will improve the quality of life for Americans with sensory loss. The program will serve students from computer science, engineering, kinesiology, psychology, and speech-language-hearing disciplines through courses, research opportunities, internships in the medical-devices industry, and public outreach activities. A total of 49 different PhD students will be enrolled in the program over the course of the 5 years, 18 of whom will receive NRT fellowships.

This project creates a new interdisciplinary graduate training program that encompasses the following key aims: 1. Establish an academic program in the form of a graduate minor through which students will receive the focused, multi-disciplinary educational background they will need to address critical challenges in the development of assistive technologies for sensory loss. 2. Advance interdisciplinary research in translational sensory science through the development of structural mechanisms that support the formation of interdisciplinary research teams. 3. Provide extra-curricular training and networking opportunities to the trainees through supplementary mechanisms, including weekly journal clubs, summer and winter workshops, spring and fall research symposia, and an annual fall retreat. Students will also participate in public engagement such as preparation of podcasts on research topics of interest to the stakeholders. 4. Provide practical career development opportunities through personalized contacts with local and national companies. Many trainees will eventually seek employment in industry, and those who continue in academia should seek to engage in collaborative work with industry partners. In an effort to provide students with real-world experience that can inform their research and post-graduation career choices, the program will offer the trainees 8-10 week summer internship opportunities in industry-based applied research.

The NSF Research Traineeship (NRT) Program is designed to encourage the development and implementation of bold, new potentially transformative models for STEM graduate education training. The Traineeship Track is dedicated to effective training of STEM graduate students in high priority interdisciplinary research areas, through comprehensive traineeship models that are innovative, evidence-based, and aligned with changing workforce and research needs.