Richard L. Freyman - US grants

A series of experiments will investigate adults' perception of an auditory phenomenon known as the precedence effect. The precedence effect occurs when two identical sounds are delivered from two spatial locations (e.g., two loudspeakers), with the onset of one leading the other by a few thousandths of a second. The listener localizes the sound solely at the leading location and does not hear the sound at the lagging location at all. When the time delay between onsets is increased, the echo threshold is reached and the lagging sound is heard as an echo. The precedence effect is due to the nervous system's active suppression of the echoes. These experiments will be directed toward establishing the conditions under which this suppression breaks down. Previous work has shown that a sudden switch in the location of leading and lagging sound will break down the suppression. These experiments will explore whether the presence of the echo before the switch is necessary and whether only the leading (or lagging) sound being changed in location is sufficient to break the suppression. Acoustic characteristics of the sound, such as intensity and frequency, will be varied to determine how they affect suppression. Inhibition of echoes is a critical feature of everyday listening in normal settings. Without echo suppression, it would be very difficult to detect the original source of a sound, leading to confusion of sound sources. Because the precedence effect is an inhibitory process of the central nervous system, brain damage produced by lesions or strokes could impair a person's ability to localize sound. Better understanding of the precedence effect and the stimulus conditions affecting it has relevance for theories of binaural hearing and has practical implications for patients with damage in the auditory areas of the brain.

The proposed program of research will investigate the cues contained within the speech waveform envelope in order to discover how these cues might be preserved or accentuated to benefit speech perception by hearing-impaired listeners. Spectral information is often limited or degraded in listeners with substantial degrees of hearing losses because auditory frequency resolution is poor or because a useable dynamic range exists only in a narrow range of frequencies. Under these conditions of reduced spectral resolution or limited bandwidth, it is widely thought that listeners need to rely more heavily than normal on the amplitude envelope of the speech waveform. However, only a limited knowledge of the kinds of information contained in the waveform envelope is currently available. The proposed project is basic research designed to improve our understanding of the role of amplitude envelope information in speech recognition. The work will focus specifically on the consonant-vowel (C-V) intensity ratio, as natural differences in level between consonants and vowels are largely responsible for the shape of the waveform envelope. Three years of research are proposed to investigate the role of the C-V ratio. The effects of modifying the C-V ratio will be evaluated in normal-hearing listeners for speech degraded in ways relevant to those occurring frequently in hearing loss. These ways are (1) smearing of the spectral details through simulated widening of the critical band, and (2) low-pass filtering. This research will investigate the interactions between modification of the C-V ratio and the amount of spectral information available to the listener at both suprathreshold and near-threshold levels. It will also investigate the extent of improvement in speech recognition that can be obtained by optimum processing of the C-V ratio, and the effects of training on the ability of listeners to make use of optimum amplitude envelope cues. Finally, automatic algorithms for processing the C-V ratio will be developed and tested.

DESCRIPTION: The precedence effect is a phenomenon that enables us to localize sound in a reverberant environment. When sound is produced in enclosed spaces, such as rooms, the original sound is followed by reflected sound off any hard surface. These reflected sounds, or echoes, are not perceived by the listener as independent sound sources, but rather serve to enhance the original sound's loudness and spatial extent. Traditionally the precedence effect has been regarded as a low level mechanism that inhibits later-arriving sounds. A more cognitive approach will be taken in the proposed experiments. The guiding hypothesis is that echoes carry valuable information about room acoustics. Although the echoes are not heard as separate sounds, the room acoustic information is encoded and analyzed by the brain. Specifically, our hypothesis states that during an ongoing sound, properties of the echo that convey room acoustic information should not change abruptly. Sudden changes of this sort cause a drop in echo threshold. The suppression of the echo is released and the listener hears the echo as a separate, localizable sound. One set of experiments will investigate conditions under which the echo threshold is hypothesized to drop and conditions where it should remain stable. A second set of experiments will investigate listeners' sensitivity to below-threshold changes in the echo's intensity and delay relative to the original sound. The methodology used involves having normal-hearing adults sit in an anechoic chamber, facing an array of loudspeakers that feature an initial or leading sound followed by sounds simulating a variety of echoes. Listeners are asked to press a button when they hear an echo from a particular location (subjective task), or to discriminate a left/right shift in location of the echo (objective task). A final set of experiments will investigate the role of the precedence effect in speech recognition in noisy backgrounds. Previous research has shown that a spatial separation of signal from noise improves speech recognition. The proposed experiments will determine if the spatial separation of perceived auditory images of signal and noise serves the same advantage as their physical separation. Since one of the main problems of the hearing-impaired is separating signal information from a noisy background, the proposed research is relevant to finding solutions to this problem. A better understanding of how the precedence effect aids sound localization in reverberant spaces and its possible role in separating signal from noise should help to devise better hearing aids and design acoustic environments that enhance hearing in general.

In a typical room, our ears receive not only the original source of sound, but numerous reflections off of surfaces in the room. Fortunately, through the precedence effect we are able to fuse a sound source and its reflections into a single image and locate that image near the original source of sound. The overall goal of the proposed research is to achieve a better understanding of the precedence effect and its importance in understanding speech in the presence of interfering noise or other speech. The primary assumption is that there are two types of masking of speech: (1) traditional energetic masking, and (2) informational masking, in which it is difficult for the listener to separate out the pattern of the target speech within a fluctuating pattern of interference. In non-reflective conditions, the horizontal separation of target and masking sources releases energetic masking, mostly because the head shadows the masker at one ear and because the brain compares the differences in inputs to the two ears. Unfortunately, this type of masking release is dramatically reduced by the reflections in a real room. The hypothesis tested in this research is that because the precedence effect preserves the perceived spatial distinctions between target and masker even in reverberation, release from informational masking is not reduced by room reflections. The studies will investigate the nature of informational masking and how the precedence effect and binaural hearing in general contribute to overcoming this type of masking. The study of binaural hearing is particularly important because many people with bilateral hearing impairment become effectively monaural listeners when they only wear one hearing aid. The results of these studies may well show how important it is to be able preserve binaural hearing and sound localization when prescribing auditory prostheses. The studies will also investigate the phenomena and mechanisms of the precedence effect on a more basic level, adding to our overall understanding of sound perception in typical reverberant spaces.

? DESCRIPTION (provided by applicant): Sound localization is a fundamental auditory ability that is important for safety and for connecting a person to his/her environment. Further, the separate localization of target and interfering sources of sound facilitates better attention to, detection of, and recognition of the source of interest. Due to a phenomenon known as the precedence effect, good sound localization abilities are preserved even when typical room reflections are present that would otherwise be expected to severely disrupt localization. The first of two specific aims of this research is to advance the scientific knowledge of the precedence effect in order to uncover the multiple and currently mysterious mechanisms that underlie it. The project will employ newly developed stimulus conditions as well as physiological experiments in animal models to better understand how spatial hearing is maintained in the face of acoustic reflections. The second aim is to apply this new knowledge to a large but understudied population of listeners with unequal hearing loss in the two ears. The asymmetric inputs these listeners receive present challenges for sound localization. Studies of both objective performance and listener self- report show that these individuals have difficulty in situations where integration across the two ears would normally be of benefit. However, the choices available for people with unbalanced hearing vary widely and the ability to help them through evidence-based clinical decision-making is not nearly sufficient. This project will be the first to investigate the multiple aspects of the precedence effect in listeners with hearing threshold asymmetries, and will not only document the major issues they face, but also discover the bases of these problems through detailed evaluation of the processing of individual localization cues. The results will lead to a better scientific understanding of the extent to whic listeners with asymmetric hearing thresholds cope with their unbalanced hearing, ultimately leading to better-informed professional management of asymmetric hearing loss.