Janina Fels

Switching in the cocktail party: exploring intentional control of auditory selective attention.

Using a novel variant of dichotic selective listening, we examined the control of auditory selective attention. In our task, subjects had to respond selectively to one of two simultaneously presented auditory stimuli (number words), always spoken by a female and a male speaker, by performing a numerical size categorization. The gender of the task-relevant speaker could change, as indicated by a visual cue prior to auditory stimulus onset. Three experiments show clear performance costs with instructed attention switches. Experiment 2 varied the cuing interval to examine advance preparation for an attention switch. Experiment 3 additionally isolated auditory switch costs from visual cue priming by using two cues for each gender, so that gender repetition could be indicated by a changed cue. Experiment 2 showed that switch costs decreased with prolonged cuing intervals, but Experiment 3 revealed that preparation did not affect auditory switch costs but only visual cue priming. Moreover, incongruent numerical categories in competing auditory stimuli produced interference and substantially increased error rates, suggesting continued processing of task-relevant information that often leads to responding to the incorrect auditory source. Together, the data show clear limitations in advance preparation of auditory attention switches and suggest a considerable degree of inertia in intentional control of auditory selection criteria.

Sound localization in individualized and non-individualized crosstalk cancellation systems.

The sound-source localization provided by a crosstalk cancellation (CTC) system depends on the head-related transfer functions (HRTFs) used for the CTC filter calculation. In this study, the horizontal- and sagittal-plane localization performance was investigated in humans listening to individualized matched, individualized but mismatched, and non-individualized CTC systems. The systems were simulated via headphones in a binaural virtual environment with two virtual loudspeakers spatialized in front of the listener. The individualized mismatched system was based on two different sets of listener-individual HRTFs. Both sets provided similar binaural localization performance in terms of quadrant, polar, and lateral errors. The individualized matched systems provided performance similar to that from the binaural listening. For the individualized mismatched systems, the performance deteriorated, and for the non-individualized mismatched systems (based on HRTFs from other listeners), the performance deteriorated even more. The direction-dependent analysis showed that mismatch and lack of individualization yielded a substantially degraded performance for targets placed outside of the loudspeaker span and behind the listeners, showing relevance of individualized CTC systems for those targets. Further, channel separation was calculated for different frequency ranges and is discussed in the light of its use as a predictor for the localization performance provided by a CTC system.

Intentional attention switching in dichotic listening: Exploring the efficiency of nonspatial and spatial selection

Using an auditory variant of task switching, we examined the ability to intentionally switch attention in a dichotic-listening task. In our study, participants responded selectively to one of two simultaneously presented auditory number words (spoken by a female and a male, one for each ear) by categorizing its numerical magnitude. The mapping of gender (female vs. male) and ear (left vs. right) was unpredictable. The to-be-attended feature for gender or ear, respectively, was indicated by a visual selection cue prior to auditory stimulus onset. In Experiment 1, explicitly cued switches of the relevant feature dimension (e.g., from gender to ear) and switches of the relevant feature within a dimension (e.g., from male to female) occurred in an unpredictable manner. We found large performance costs when the relevant feature switched, but switches of the relevant feature dimension incurred only small additional costs. The feature-switch costs were larger in ear-relevant than in gender-relevant trials. In Experiment 2, we replicated these findings using a simplified design (i.e., only within-dimension switches with blocked dimensions). In Experiment 3, we examined preparation effects by manipulating the cueing interval and found a preparation benefit only when ear was cued. Together, our data suggest that the large part of attentional switch costs arises from reconfiguration at the level of relevant auditory features (e.g., left vs. right) rather than feature dimensions (ear vs. gender). Additionally, our findings suggest that ear-based target selection benefits more from preparation time (i.e., time to direct attention to one ear) than gender-based target selection.

Anthropometric Parameters Influencing Head-Related Transfer Functions

This study deals with the question of how much influence the head, torso, and pinna dimensions have on spatial hearing, particularly as far as children and their growth are concerned. A statistical analysis of head, torso, and pinna dimensions of subjects including children and adults of all ages has been conducted for the first time to create a detailed database. Spatial cues are derived from CAD 1 models of the head using the Boundary Element Method (BEM) with a basic simplified model. Anthropometric parameters for the head and torso (height, breadth, length, and the distance from the ear to the shoulder) and for the pinna (height, breadth, and rotation angle, and cavum concha height, depth, and breadth) are analyzed and their effects are discussed. Varying each parameter individually makes it possible to isolate the effect of this parameter on the HRTF and on the spatial cues. The same approach is taken for various incidence angles. As far as the future construction of artificial heads for specific groups of population, including various ages, is concerned, the results presented here show where larger tolerances are acceptable regarding the different dimensions and where the dimensions should be chosen more accurately.

From children to adults: How binaural cues and ear canal impedances grow

In this thesis the growth dependency of head-related transfer functions and their resulting parameters (interaural time and level differences) are analyzed as well as ear canal impedances. Differences in the anatomy of children and adults thus result in different binaural cues. The individual anthropometric parameters, however, affect the binaural cues to a varying extent. In this thesis the most important anthropometric parameter with regard to their influence on binaural cues, is determined. Ear canal impedances are undergoing certain changes as well while children are growing up and finally reach adulthood. This thesis presents, for the first time, the age-related development of data that is most decisive for the impedances. These new findings open up new possibilities to develop artificial child heads and couplers for hearing aids that are suitable for children. Thus, improvements in certain fields of applications, such as the development and fitting of hearing aids for children, are now possible, and...

Intentional switching in auditory selective attention: Exploring age-related effects in a spatial setup requiring speech perception

Using a binaural-listening paradigm, age-related differences in the ability to intentionally switch auditory selective attention between two speakers, defined by their spatial location, were examined. Therefore 40 normal-hearing participants (20 young, Ø 24.8years; 20 older Ø 67.8years) were tested. The spatial reproduction of stimuli was provided by headphones using head-related-transfer-functions of an artificial head. Spoken number words of two speakers were presented simultaneously to participants from two out of eight locations on the horizontal plane. Guided by a visual cue indicating the spatial location of the target speaker, the participants were asked to categorize the targets number word into smaller vs. greater than five while ignoring the distractors speech. Results showed significantly higher reaction times and error rates for older participants. The relative influence of the spatial switch of the target-speaker (switch or repetition of speakers direction in space) was identical across age groups. Congruency effects (stimuli spoken by target and distractor may evoke the same answer or different answers) were increased for older participants and depend on the targets position. Results suggest that the ability to intentionally switch auditory attention to a new cued location was unimpaired whereas it was generally harder for older participants to suppress processing the distractors speech.

Intentional Preparation of Auditory Attention-Switches: Explicit Cueing and Sequential Switch-Predictability

In an auditory attention-switching paradigm, participants heard two simultaneously spoken number-words, each presented to one ear, and decided whether the target number was smaller or larger than 5 by pressing a left or right key. An instructional cue in each trial indicated which feature had to be used to identify the target number (e.g., female voice). Auditory attention-switch costs were found when this feature changed compared to when it repeated in two consecutive trials. Earlier studies employing this paradigm showed mixed results when they examined whether such cued auditory attention-switches can be prepared actively during the cue–stimulus interval. This study systematically assessed which preconditions are necessary for the advance preparation of auditory attention-switches. Three experiments were conducted that controlled for cue-repetition benefits, modality switches between cue and stimuli, as well as for predictability of the switch-sequence. Only in the third experiment, in which predictability for an attention-switch was maximal due to a pre-instructed switch-sequence and predictable stimulus onsets, active switch-specific preparation was found. These results suggest that the cognitive system can prepare auditory attention-switches, and this preparation seems to be triggered primarily by the memorised switching-sequence and valid expectations about the time of target onset.

On the influence of continuous subject rotation during HRTF measurements

In recent years, the measurement time of individual Head-Related Transfer Function (HRTF) measurements has been reduced by the use of loudspeaker arrays. The time reduction is achieved by some kind of parallelization of measurement signals. One such fast system was developed at the Institute of Technical Acoustics, RWTH Aachen University and is evaluated in this paper. When measuring HRTFs, the subject is usually rotated by some angle, and stops and waits for the measurement signal to complete before moving to the next measurement angle. It was shown that with this static approach a comparable results to a traditional measurement using a single speaker could be achieved. To further reduce the measurement time, a slow continuous subject rotation can be used instead. While this rotation will violate LTI (linear, time-invariant) requirements of the commonly used signal processing, the influence is assumed to be negligible. As the subject is rotating during the measurement sweep, different azimuth angles are measured per frequency. This frequency dependent offset in the measurement positions has to be corrected during the post processing. To this end, a spherical harmonic decomposition and reconstruction is applied as an interpolation method. To quantify the influence of the rotation and the subsequent post processing, a subjective and objective comparison between statically and continuously measured objects is shown in this paper.In recent years, the measurement time of individual Head-Related Transfer Function (HRTF) measurements has been reduced by the use of loudspeaker arrays. The time reduction is achieved by some kind of parallelization of measurement signals. One such fast system was developed at the Institute of Technical Acoustics, RWTH Aachen University and is evaluated in this paper. When measuring HRTFs, the subject is usually rotated by some angle, and stops and waits for the measurement signal to complete before moving to the next measurement angle. It was shown that with this static approach a comparable results to a traditional measurement using a single speaker could be achieved. To further reduce the measurement time, a slow continuous subject rotation can be used instead. While this rotation will violate LTI (linear, time-invariant) requirements of the commonly used signal processing, the influence is assumed to be negligible. As the subject is rotating during the measurement sweep, different azimuth angles are ...

Examining auditory selective attention in realistic, natural environments with an optimized paradigm

The topic of the collaborative project is the exploration of cognitive control mechanisms underlying auditory selective attention. The aim is to examine the influence of variables that increase the complexity of the auditory scene with respect to technical aspects (dynamic binaural hearing with consideration of room acoustics and head movements) and that influence the efficiency of cognitive processing. Using a binaural-listening paradigm, the ability to intentionally shift auditory attention in various anechoic setups was tested. An anechoic reproduction fails to represent realistic listening experiences. The original paradigm is extended to use longer stimuli to offer more opportunities. Spoken phrases by two speakers were presented simultaneously to subjects from two of eight positions. The stimuli were phrases that consisted of a single number word followed by either the German direction “UP” or “DOWN”. Guided by a visual cue prior to auditory stimulus onset, subjects were asked to identify whether th...

A high-resolution head-related transfer function and three-dimensional ear model database

Nowadays immersive three-dimensional virtual scenes have become very popular. This paper presents and provides a database of 48 head-related transfer function (HRTF) datasets and the corresponding three-dimensional ear mesh models of individual subjects, in order to select or model an HRTF dataset for binaural synthesis and auditory reproduction. For easy access, the database of the Institute of Technical Acoustics (RWTH Aachen University, Germany) is accessible online. The HRTF measurement setup and the generation of the three-dimensional models of the ear geometry, reconstructed from magnetic resonance imaging scans, are specifically described in depth.