1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
The role of high frequencies in speech localization
Rent:
Rent this article for
USD
10.1121/1.1926107
/content/asa/journal/jasa/118/1/10.1121/1.1926107
http://aip.metastore.ingenta.com/content/asa/journal/jasa/118/1/10.1121/1.1926107

Figures

Image of FIG. 1.
FIG. 1.

Spectrograms of (a) a silent period, (b) the recorded word “sludge,” and (c) the same word after low-pass filtering at . The spectrograms show that the recordings were made with high signal-to-noise ratio and that substantial energy exists above in the recorded speech.

Image of FIG. 2.
FIG. 2.

The lateral/polar coordinate system consists of a single pole passing through the two ears. (a) The lateral angle is the horizontal angle away from the midline, with 0° defining the median sagittal plane and and 90° defining the left- and right-most extremities of the sphere. (b) The polar angle describes the angle around the circle described by a particular lateral angle. A polar angle of 0° describes the front-most location on this circle, with polar angle increasing to 90° at the top and at the bottom, with 180° and coinciding at the back.

Image of FIG. 3.
FIG. 3.

Scatter plots showing lateral angle data (pooled across all subjects) for Experiment 1. The three panels contain data for the three stimulus conditions: broadband noise (left), broadband speech (middle), and low-pass filtered speech (right). Target lateral angle (abscissa) is plotted against response lateral angle (ordinate) and the size of the dots represents the number of responses clustered at a point.

Image of FIG. 4.
FIG. 4.

Scatter plots showing polar angle data for Experiment 1. Each row shows data for a different subject as labeled. The three panels contain data for the three stimulus conditions: broadband noise (left), broadband speech (middle), and low-pass filtered speech (right). Target polar angle (abscissa) is plotted against response polar angle (ordinate) and the size of the dots represents the number of responses clustered at a point.

Image of FIG. 5.
FIG. 5.

(a) Mean absolute lateral angle errors from Experiment 1. The six clusters of bars show results for the five subjects as well as the mean across subjects. Mean errors are shown for broadband noise (black bars), broadband speech (gray bars), and low-pass filtered speech (white bars). Error bars show standard error of the mean. (b) Mean absolute polar angle errors from Experiment 1. All other details as for (a). The asterisk indicates that for the mean data all three conditions were significantly different from each other .

Image of FIG. 6.
FIG. 6.

Mean polar angle errors for different groups of words. The numbers 1 to 10 on the abscissa denote 10 groups of words of increasing high-frequency content (see the text for details). Mean errors are shown for broadband speech (closed symbols, solid line) and low-pass filtered speech (open symbols, dashed line). It can be seen that all words were localized more poorly after low-pass filtering.

Image of FIG. 7.
FIG. 7.

An illustration of the stimulus conditions employed in Experiment 2. Shown are power spectral density plots of a speech stimulus after low-pass filtering with varying attenuation in the stopband (0, 20, 40, and down). Note that down was not retested, but is equivalent to the low-pass speech condition of Experiment 1 and is included for comparison.

Image of FIG. 8.
FIG. 8.

Scatter plots showing lateral angle data (pooled across all subjects) for Experiment 2. The three panels contain data for the three stimulus conditions: broadband speech (left), down low-pass speech (middle), and down low-pass speech (right). Target lateral angle (abscissa) is plotted against response lateral angle (ordinate) and the size of the dots represents the number of responses clustered at a point.

Image of FIG. 9.
FIG. 9.

Scatter plots showing polar angle data for Experiment 2. Each row shows data for a different subject as labeled. The three panels contain data for the three stimulus conditions: broadband speech (left), down low-pass speech (middle), and down low-pass speech (right). Target polar angle (abscissa) is plotted against response polar angle (ordinate) and the size of the dots represents the number of responses clustered at a point.

Image of FIG. 10.
FIG. 10.

(a) Mean absolute lateral angle errors from Experiment 2. The six clusters of bars show results for the five subjects as well as the mean across subjects. Mean errors are shown for broadband speech (black bars), down low-pass speech (dark gray bars), and down low-pass speech (light gray bars). Errors for low-pass filtered speech ( down) from Experiment 1 are also shown for comparison (white bars). Error bars show standard error of the mean. (b) Mean absolute polar angle errors from Experiment 2. All other details as for (a). The asterisk indicates that for the mean data the three new conditions were significantly different from each other .

Tables

Generic image for table
TABLE I.

Spherical correlation coefficients (SCCs) for each of the five subjects in Experiment 1 (and the mean). Each of the three rows contains values for the three stimulus conditions. Each SCC is calculated on the basis of five repetitions at each of 76 stimulus locations (i.e., 380 trials in total). See the text for details of this statistic.

Generic image for table
TABLE II.

Percentage of cone of confusion (COC) errors made by each of the five subjects in Experiment 1 (and the mean). Each of the three rows contains values for the three stimulus conditions. Each value represents the percentage of trials (out of the total 380) in which a COC error was made. See the text for details of this statistic.

Generic image for table
TABLE III.

SCCs for each of the five subjects in Experiment 2 (and the mean). The first three rows contain values for the three new stimulus conditions and the fourth row is a reiteration of the data from the low-pass condition of Experiment 1. Each SCC is calculated on the basis of five repetitions at each of 76 stimulus locations (i.e., 380 trials in total). See the text for details of this statistic.

Generic image for table
TABLE IV.

Percentage of COC errors made by each of the five subjects in Experiment 2 (and the mean). The first three rows contain values for the three new stimulus conditions and the fourth row is a reiteration of the data from the low-pass condition of Experiment 1. Each value represents the percentage of trials (out of the total 380) in which a COC error was made. See the text for details of this statistic.

Loading

Article metrics loading...

/content/asa/journal/jasa/118/1/10.1121/1.1926107
2005-07-01
2014-04-20
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: The role of high frequencies in speech localization
http://aip.metastore.ingenta.com/content/asa/journal/jasa/118/1/10.1121/1.1926107
10.1121/1.1926107
SEARCH_EXPAND_ITEM