1990 — 1993 |
Kluender, Keith R |
R29Activity Code Description: Undocumented code - click on the grant title for more information. |
Spectral Temporal Factors in Perception of Fluent Speech @ University of Wisconsin Madison
The overall goal of the proposed research is to investigate the internal spectral temporal space within which speech perception takes place. A guiding premise of the proposed work is the assumption that many of the most important phenomena of speech perception can be explained in terms of general auditory processes. This in mind, experiments are designed to exploit mechanisms listeners use to maintain continuity of speech and other complex sounds under normal environmental conditions which frequently are not optimal. Many studies have demonstrated that listeners are most likely to maintain continuity of a sound source that does not undergo extreme spectral transformation over too short a period of time, but little is known about what sorts of transformation are tolerable. Given that transformations that traverse relatively little space are more likely to be perceived as continuous, the proposed research is designed to map out a spectral temporal space and to describe the and model principles governing perceptual continuity of speech. Along the way, a number of fundamental questions will be addressed concerning the structure of phonetic inventories. The Principle Investigator is co-author of a somewhat controversial theoretical position that holds that many of the most widespread phonetic regularities can be explained by efforts of language communities to arrange their phonetic inventories in a way that exploits auditory predispositions of listeners. This "auditory enhancement hypothesis" is put to a number of critical tests in attempts to explain near-universal tendencies to use certain types of vowels. It is argued that, for example, general tendencies of languages to have higher pitch for high vowels, and more nasalization of low vowels can be explained in terms of making these vowels auditorily maximally distinctive. By exploiting mechanisms that aid the listener in maintaining perceptual continuity, one can empirically assess the perceptual distance between vowel sounds that are in accordance with these phonetic regularities and vowel sounds that are not. Finally, an extensive modeling effort is described that investigates whether an unsupervised neural network model can be developed that can account for the behavior of human listeners in the proposed experiments.
|
0.961 |
1994 |
Kluender, Keith R |
R29Activity Code Description: Undocumented code - click on the grant title for more information. |
Spectral-Temporal Factors in Perception of Fluent Speech @ University of Wisconsin Madison
The overall goal of the proposed research is to investigate the internal spectral temporal space within which speech perception takes place. A guiding premise of the proposed work is the assumption that many of the most important phenomena of speech perception can be explained in terms of general auditory processes. This in mind, experiments are designed to exploit mechanisms listeners use to maintain continuity of speech and other complex sounds under normal environmental conditions which frequently are not optimal. Many studies have demonstrated that listeners are most likely to maintain continuity of a sound source that does not undergo extreme spectral transformation over too short a period of time, but little is known about what sorts of transformation are tolerable. Given that transformations that traverse relatively little space are more likely to be perceived as continuous, the proposed research is designed to map out a spectral temporal space and to describe the and model principles governing perceptual continuity of speech. Along the way, a number of fundamental questions will be addressed concerning the structure of phonetic inventories. The Principle Investigator is co-author of a somewhat controversial theoretical position that holds that many of the most widespread phonetic regularities can be explained by efforts of language communities to arrange their phonetic inventories in a way that exploits auditory predispositions of listeners. This "auditory enhancement hypothesis" is put to a number of critical tests in attempts to explain near-universal tendencies to use certain types of vowels. It is argued that, for example, general tendencies of languages to have higher pitch for high vowels, and more nasalization of low vowels can be explained in terms of making these vowels auditorily maximally distinctive. By exploiting mechanisms that aid the listener in maintaining perceptual continuity, one can empirically assess the perceptual distance between vowel sounds that are in accordance with these phonetic regularities and vowel sounds that are not. Finally, an extensive modeling effort is described that investigates whether an unsupervised neural network model can be developed that can account for the behavior of human listeners in the proposed experiments.
|
0.961 |
2000 — 2003 |
Kluender, Keith R |
R01Activity Code Description: To support a discrete, specified, circumscribed project to be performed by the named investigator(s) in an area representing his or her specific interest and competencies. |
Spectral Enhancement of Speech @ University of Wisconsin Madison
The goal is to develop a model to explain how listeners perceive speech in the face of extreme context-sensitivity resulting from co-articulation. The ability of listeners to recover speech information, despite dramatic articulatory and acoustic assimilation between adjacent speech sounds, is remarkable and central to understanding speech perception. Recent results from this laboratory suggest that it may be possible to define and model quite general processes of auditory perception and learning which may provide a significant part of the explanation for findings demonstrating a correspondence between speech perception and production with perseverative co-circulation. To the extent do general auditory processes serve to accommodate acoustic consequences of co- articulation? More generally, does perceptual (spectral) contrast serve in part or whole to undo the assimilative nature of co-articulation and its acoustic consequences? More specifically, do general psychoacoustic enhancement effects, presumably due to relatively peripheral simple neural mechanisms, serve to provide spectral contrast for complex sounds such as speech? In addition to such basic auditory effects, what is the relative contribution of experience (learning) with co-articulated speech? What are the prospects for relatively elegant algorithms for signal processing that provide a desirable adjunct to amplification for hearing- impaired listeners? Toward these ends, studies with non-human animals provide a model of auditory processing unadulterated by experience with speech as well as a model to investigate the role of experience with co- variation among acoustic attributes consequent to co-articulated production. Here, animals permit precise control of experience with speech in establishing whether learning plays a significant role in accommodation of acoustic consequences of co-articulation while defining the relative contributions of experience and general auditory processes.
|
0.961 |
2004 — 2008 |
Kluender, Keith R |
R01Activity Code Description: To support a discrete, specified, circumscribed project to be performed by the named investigator(s) in an area representing his or her specific interest and competencies. |
Spectral Temporal Factors in Perception of Speech @ University of Wisconsin Madison
[unreadable] DESCRIPTION (provided by applicant): The overarching goal is to better understand auditory processes underlying perception of speech and other complex sounds. One current goal is to continue development of signal processing algorithms for spectral enhancement. The focus is upon improvement of speech recognition in the face of extreme context-sensitivity resulting from co articulation. This effort is directed toward incorporation in facilitative devices such as hearing aids. The algorithm, currently successful with normal hearing listeners with simulated hearing loss, will be further improved, and testing will be extended to a population of listeners with hearing impairment. [unreadable] [unreadable] The second major goal is to describe and model processes by which the auditory system maintains perceptual constancy despite variations commonly encountered across sound sources (such as different talkers) and across different sound environments. More specifically, reveal auditory mechanisms that serve to compensate for reliable spectral and temporal characteristics of a sound source and of the ambient environment more generally. The primary focus is to further understand perception of speech, and the manner in which these auditory processes explain how listeners understand speech across variations in characteristics of individual talkers, as well as their speaking rate and rhythm. If possible, these new findings will be incorporated into development of the spectral enhancement algorithm for clinical application. Finally, behavioral studies using a nonhuman mammal (chinchilla) will be conducted in order to establish a nonhuman model of auditory perceptual constancy that can be used in efforts to better understand the underlying physiology of spectral and temporal processing of complex sounds. [unreadable] [unreadable]
|
0.961 |
2009 — 2010 |
Kluender, Keith R |
RC1Activity Code Description: NIH Challenge Grants in Health and Science Research |
Contrast Enhancement to Improve Speech Recognition With Hearing Devices
DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (06): Enabling Technologies and specific Challenge Topic, 06-DC-101: Develop Improved Hearing Devices. It is projected that by 2030 there will be over 40 million adults and over 2 million children with hearing loss in the United States. The average reduction in earning potential for individuals with hearing aids is estimated to be $11,500 per year and $23,000 per year for individuals with untreated hearing loss. Total annual lost income of American workers with hearing loss is enormous. An important variable influencing these figures is hearing aid performance, especially in noise. Advancements in hearing aid performance have the potential to improve quality of life for more than 10 percent of the American population as well as productivity the average hearing- impaired worker. The challenge in advancing hearing aid performance is to overcome the blurring, or distortion, of frequency information important for understanding speech caused by damage to the inner ear. Distortion in the inner ear is the reason why making speech louder does not always make it clearer. In fact, amplification can contribute further to the distortion. Current hearing aid strategies compress, or squeeze, the amplitude and/or frequency range of speech in order to provide listeners with greater access to information, particularly the softer parts and higher frequency parts of speech. The full ability to improve speech understanding with these strategies is limited by the fact that they, too, increase blurring of the speech signal. The signal processing strategy proposed here, the Contrast Enhancement (CE) algorithm, sharpens the contrast in the speech signal in order to minimize effects of the blur caused by the damaged inner ear. The CE algorithm is efficient and operates in real time, so it is ideal for hearing aid applications. Innovative aspects of the CE algorithm include simulation of biological processes that are important for understanding connected speech, especially processes that operate across successive speech sounds to enhance signature changes in their frequency composition. The CE algorithm is currently very successful in restoring speech understanding in normal-hearing individuals listening to speech that is blurred to simulate inner ear damage. Further improvements to the CE algorithm will be investigated using normal-hearing listeners with simulated inner ear damage. The CE algorithm also will be tested for improving speech understanding for people who have a wide range of different hearing losses. These tests with hearing impaired listeners include an innovative approach to customizing the algorithm for individual hearing losses. While the CE algorithm might work well as a standalone processing strategy, there are strong reasons to believe that its greatest and most immediate impact will occur when used together with current signal processing strategies already used in hearing aids. As noted above, some of these methods tend to increase the blur in the speech signal. Therefore, using a wide range of listeners with different types of hearing loss, the effectiveness of the CE algorithm will be tested for reducing the negative consequences of current processing strategies so that overall speech understanding is improved. Finally, the CE algorithm has the potential to help cochlear implant users who experience a severe form of blurring in the speech signal. Like hearing aid users, blurring with electrical hearing is attributable both to impaired inner ear functioning and to device processing that is necessary to accommodate the impairment. As a result of the blurring, cochlear implant users cannot make full use of information in the electrical patterns representing the speech signal. The objective is to exploit the CE algorithm so that current and future cochlear implant users will be able to take greater advantage of the information in their implants. Effectiveness of the CE algorithm will first be tested using normal-hearing listeners with simulated cochlear implant processing. Next, benefits of contrast-enhanced speech on word and sentence recognition will be tested using actual cochlear implant recipients. Proposed research is to refine and implement signal processing strategies that improve performance of hearing aids and cochlear implants in order to provide better hearing and quality of life to millions of Americans with hearing impairments.
|
1 |