17.05.2014 Views

FRIDAY MORNING, 20 MAY 2005 REGENCY E, 8:30 A.M. TO 12:00 ...

FRIDAY MORNING, 20 MAY 2005 REGENCY E, 8:30 A.M. TO 12:00 ...

FRIDAY MORNING, 20 MAY 2005 REGENCY E, 8:30 A.M. TO 12:00 ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

5aSC14. Spectral contributions to intelligibility of sentences with<br />

flattened fundamental frequency. Peter J. Watson and Robert S.<br />

Schlauch Dept. of Speech-Lang.-Hearing Sci., Univ. of Minnesota, 164<br />

Pillsbury Dr. S.E., Shevlin 115, Minneapolis, MN 55455, pjwatson@<br />

umn.edu<br />

Recently, the contribution of fundamental frequency (F0) variation to<br />

speech intelligibility in background noise has been studied J. Laures and<br />

G. Weismer, JSLHR 42, 1148–1156 1999. The process used for this<br />

analysis is to flatten the frequency contour at the average F0. Results<br />

show that sentences with flattened F0 are less intelligible than those with<br />

natural F0 variation. However, this technique may reduce the prominence<br />

of formant peaks because it excludes F0s that are below the average<br />

flattened frequency. As noted by Laures and Weismer 1999, eliminating<br />

lower F0s from the analysis results in wider spacing between the harmonics<br />

and the available energy to excite the vocal tract resonances is diminished.<br />

This study examined the effect of flattening at the average and<br />

lowest F0 of each sentence to intelligibility. Twenty subjects listened to<br />

sentences in a continuous noise background. Sentences were equated for<br />

root-mean square energy. Results showed that the low-flattened condition<br />

was significantly more intelligible than the average-flattened condition.<br />

However, a condition with natural F0 variation was more intelligible than<br />

the two flattened conditions. These findings show that that the frequency<br />

of flattening does contribute to intelligibility but that natural F0 variation<br />

appears to contribute to a greater degree.<br />

5aSC15. Comparison of cues in auditory selective and divided<br />

attention. Jae hee Lee and Larry E. Humes Indiana Univ. at<br />

Bloomington, Bloomington, IN 47405<br />

This study investigated auditory selective and divided attention in ten<br />

normal-hearing listeners. Subjects listened to two simultaneous sentencelike<br />

utterances from the Coordinate Response Measure CRM speech<br />

corpus at 90 dB SPL in dichotic or monotic listening conditions. This<br />

corpus has the following format, Ready call sign, gotocolor number,<br />

now, with all possible combinations of call signs 8, colors 4 and numbers<br />

8. In all conditions, the listener identified the color-number coordinates<br />

spoken by the target talker. The target talker was identified by a cue<br />

provided either prior to selective attention or following divided attention<br />

the presentation of the two simultaneous utterances. In the first experiment,<br />

either the ear or the call sign served as the cue for the target<br />

talker in monotic or dichotic listening conditions. In the second experiment,<br />

talker gender served as the cue in monotic listening conditions.<br />

Across both experiments, performance for selective attention was superior<br />

to that for divided attention. Performance for ear or gender cueing was<br />

better than for call sign. Analysis showed that 80% of errors were due to<br />

intrusions from the competing utterance, supporting an informational attention,<br />

rather than energetic, form of masking. Work supported, in part,<br />

by NIH R01 AG08293.<br />

5aSC16. Effects of high intensity on recognition of low- and highfrequency<br />

speech in noise. Van Summers and Mary Cord Army<br />

Audiol. & Speech Ctr., Walter Reed Army Medical Ctr., Washington, DC<br />

<strong>20</strong><strong>30</strong>7-5<strong>00</strong>1<br />

For listeners with normal hearing NH, speech recognition scores often<br />

decrease when intensities exceed moderate levels rollover is observed.<br />

It is currently unclear whether the factors leading to rollover in<br />

NH listeners also limit performance for hearing-impaired HI listeners at<br />

high sound levels. This study aimed at clarifying the stimulus conditions<br />

most clearly associated with rollover and whether rollover effects are similar<br />

for NH listeners and listeners with mild hearing impairment. In Stage<br />

1, NH and HI listeners heard digitally-filtered sentences and adaptive procedures<br />

were used to determine high- and low-frequency bandwidths<br />

yielding 70%-correct word recognition in quiet at moderate levels. In<br />

Stage 2, broadband and band-limited stimuli based on the high-, and<br />

low-frequency passbands measured in quiet in Stage 1 were tested at<br />

moderate levels in background noise. Noise levels were varied adaptively<br />

to determine signal-to-noise levels supporting <strong>30</strong>%-correct recognition.<br />

Stimulus conditions leading to criterion performance at moderate presentation<br />

levels in Stage 2 were then retested at higher levels in Stage 3. NH<br />

listeners showed larger and more consistent rollover effects for highfrequency<br />

than for low-frequency or broadband stimuli. The results for HI<br />

listeners showed greater variability but also indicated clear rollover effects<br />

for high-frequency stimuli at high levels.<br />

5aSC17. The effect of auditory feedback alterations on the speech<br />

quality of hearing aid and cochlear implant users. Dragana Barac-<br />

Cikoja and Leslie Klein Gallaudet Univ., 8<strong>00</strong> Florida Ave. NE,<br />

Washington, DC <strong>20</strong><strong>00</strong>2<br />

The speech of seven hearing aid HA users with severe-profound<br />

hearing loss and six cochlear implant CI users was recorded as they read<br />

the Rainbow Passage while their speech feedback was either spectrally<br />

altered in real-time, completely masked by multi-talker babble noise, or<br />

unaltered. Spectral alterations were implemented by filtering the speech<br />

signal into either one or four frequency bands, extracting their respective<br />

amplitude envelopes, and amplitude-modulating the corresponding noise<br />

bands. While the single-band condition provided only coarse information<br />

about the speech rhythmic structure, the four-band noise signal remained<br />

intelligible. Auditory feedback was presented via insert earphones to the<br />

HA users, and via the auxiliary jack with the headpiece microphone silenced<br />

to the CI users, at the participants’ most comfortable listening<br />

level. The quality of the recorded speech separated into individual sentences<br />

was assessed using a 2IFC procedure. For each combination of the<br />

experimental conditions, six judges selected the more natural-sounding<br />

utterance in a pair. Preference scores were calculated for each of the four<br />

feedback conditions and statistically tested. HA and CI group differed in<br />

how feedback intelligibility affected the speech quality. Possible acoustic<br />

correlates of the perceived differences will be discussed.<br />

5aSC18. Articulatory and acoustic measurements of vowels in hearing<br />

impaired speakers following treatment. Penelope Bacsfalvi and<br />

Barbara Bernhardt School of Audiol. and Speech Sci. SASS, UBC,<br />

Vancouver, BC, Canada V6T 1Z3, Penelope@audiospeech.ubc.ca<br />

The purpose of this study was to examine the relationships between<br />

ultrasound tongue shapes, electropalatography EPG contacts, formant<br />

data, and the perceptual data in the vowels of three hard of hearing adolescents.<br />

This pilot study examines the pre- and post-therapy speech of the<br />

participants in a 6 week group therapy programme with ultrasound and<br />

EPG technologies. Before treatment, participants showed particular difficulty<br />

with high vowels and the tense-lax distinction. Recordings were<br />

made of 2 males and 1 female with ultrasound and EPG. Three types of<br />

measurements were made; formant measurements, EPG tongue palate<br />

contacts, and perceptual judgements by experienced speech-language pathology<br />

researchers. Initial analysis values showed a change in the direction<br />

of EPG contact patterns, and perceptual categories. F1 values decreased<br />

and F2 values increased, moving in the direction of typical<br />

formant frequencies found in hearing people. Preliminary results of this<br />

study support a lingual-articulatory approach to treatment.<br />

5aSC19. The role of fundamental frequency „F0… contours in the<br />

perception of speech against interfering speech. Christine Binns and<br />

John F Culling School of Psych., Cardiff Univ., Tower Bldg., Park Pl.<br />

Cardiff, CF10 3AT, UK<br />

A set of experiments investigated the effect of the F0 contour on<br />

speech intelligibility against interfering sounds. Speech Reception Thresholds<br />

SRTs were measured in dB for sentences with different manipulations<br />

of their F0 contours. These manipulations involved either a scaled<br />

reduction in F0 variation, or the complete inversion of the F0 contour.<br />

Against speech-shaped noise, a flattened F0 contour did not have a sig-<br />

2606 J. Acoust. Soc. Am., Vol. 117, No. 4, Pt. 2, April <strong>20</strong>05 149th Meeting: Acoustical Society of America 2606

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!