885095931

6SP9. Speech perception ability of adults with multiple-channel cochlear implanta, using the spectra! maxima sound processor. Colelle M. McKay and Hugh J. McDermott (Dept. of Otolaryngol., Univ. of Melbourne, R.V.E.E.H., 32 Gisbornc St., East Melbourne, Australia, 3002)

The spectral maxima sound processor (SMSP) was designed at the (Jniversity of Melbourne for use with the 22-clectrode cochlear implant manufactured by Cochlear Pty. Ltd. The processor utilizes a bank of 16 bandpass filters that are assigncd to 16 electrode pairs tonotopically. In each stimulation period (typically 4 rns) six electrode pairs are stimu-latcd. The selection of elcctrodcs and stimulation levels are determined by the six filters having the highest amplitudę outputs. Speech perception results with a number of adult implantees were obtained and the SMSP’s performance compared with that of commercially available speech processors designed for this implant, including the most rccent MSP (multi-spectral-peak processor). These results show that, for the subjects investigated, the SMSP improved speech perception ability, in quiet and in noise. This was truć both for vowels and for consonants. Speech tests used in this study includcd open-set sentences, open-set monosylłabic words and closed-set vowel and consonant confusion tests. The tests were administered in the auditory-alone condition.

6SP12. Longitudinal changes in speech production in children with cochlear implants. Arlene E. Camey (Boys Town Natl. Res. Hospital, Omaha, NE 68131), Jilł Firszt (Carle Clinic, Urbana, 1L 618013, Cynthia J. Johnson (Univ. of Illinois, Champaign, IL 61820), and Edward Carncy (Boys Town Natl. Res. Hospital, Omaha. NE 68131)

The purpose of this study is to dcterminc when and how subphonc-mic changes in speech production Iead to changcs in speech intclligibil-ity in threc prelingually deafened children with poor speech production skills, following the rcceipt of multichannel cochlear implants. Mcasure-ments of open- and closed-set speech intclligibility were madę pre-implant, and at 3, 6, 12, and 18 months post-implant. In addition, monosylłabic word tokens were analyzed perccptually with reduced as-pcct feature transcription (RAFT) (A. E. Carncy, J. Acoust. Soc. Am. Suppl. 1 87, S89 (1990)] to determine global arliculatory features. Results indicatcd that relatively large changes in syllable shape, vowel space, and consonant manner preccded smali changes in speech intclligibility. RAFT results were displayed in matrix format for each test period. Matrices were compared to estimate which production changes showed consistent patterns ovcr timc. Results will be discussed in terms of base-line performance and expectcd ratę of change for prelingually deafened children who received cochlear implants.

6SP10. Analytic study of the fundamental speech skills test. Harry Levitt, Frank Eadie, and Karen Youdelman (Ctr. for Res. in Speech and Hearing Sci., Graduate School and Univ. Ctr. of The City Univ. of New York, 33 W. 42nd St., New York, NY 10036)

The fundamental speech skills test (FSST) cvaluates basie speech production skills and is designed for use with hearing-impaired stu-dents. A field study was performed in which 250 hearing-impaired stu-dents (age rangę 6 through 19 yr) were evaluated. A series of factor analyses was performed on the data, the results of which showed dis-tinct patterns among basie speech skills. Relationships with hearing loss, age, and gender were examined. The primary factors were found to be vowel production, consonant articulation, breathstream manage-ment, pitch control, stress and intonation, and syllabification. Similar rates of improvement with inereased residual hearing were observed for almost all factors. Age effects diftered among the factors. Male-female differences were observed primarily with respect to pitch control.

6SP11. Use of spectral and temporal cues in vowel Identification by normal-hearing and hearing-impaired listeners. W. Van Summers and Marjorie R. Leek (Army Audiology and Speech Ctr., Walter Reed Army Med. Ctr., Washington. I3C 20307-5001)

Listeners with sensorineural hearing impairments gcncrally show deficits in frcquency resolution accompanying their decreased auditory sensitivity. Poor frequency resolution may make it difficult for these listeners to distinguish among vowcls with similar formant frequencies. In English, these vowels tend to form tense/lax pairs (c.g., i/i, ae/r) which difler in duration. The present study examined hearing-impaired and normal-hearing listeners’ use of duration and formant frequency Information in the labeling of synthetic CVC stimuli forming a /bit-bit/continuum. Durational and FI frequency cues to vowel iden-tity varied systematically across stimuli. Frequency resolution at 2000 Hz (the FI region for i/i) was measured using a notched-noise masking paradigm. A temporal difference limen for a narrow-band noise cen-tered at 2000 Hz was also measured. Subjects with normal frequcncy and temporal resolution tended to rely primarily on formant frequency Information in the vowel labeling task. However, subjects with abnor-mal frequency resolution but near-normal temporal difTerence limens madę greater use of vowcl duration in vowel identification. (Work sup-ported by NIH.J

1959 J. Acoust. Soc. Am., Vol. 89, No. 4. Pt. 2. April 1991 6SP13. Accounting for normal-hearing listeners* perceptions of syllable-initial /d/*s spoken by hearing-impaired talkers. Brad Rakerd, Lisa Holden-Pitt, and Sally Revoile (Ctr. for Auditory and Speech Sci., Gallaudet Univ., Washington, DC 20002)

A substantial corpus of data on normal-hearing listeners’ perceptions of stop and glide consonants spoken by sensorineural hearing-impaired talkers of widely varying speech production abilitics has been collected. This report conccrns the authors’ first effbrts to measure acoustic cues to consonants in these produclions and to relate cue pres-ence to perceptual judgments. The focus was on /d/, as produced in citation-form /dVk/ syllablcs (V /a/, /x/), by ten severely through profoundly hearing-impaired talkers representing the inidrange of speech production abilities among our hearing-impaired adults. The /dVk/ syllables, along with other syllables contrastive for the initial consonant, wcrc presented to normal-hearing listeners who rnade identification judgments for the syllable-initial consonant and, at the same time, rated their confidence in each judgmenl. Acoustic analysis of the /dVk/ syllables gcncrally revcalcd some form of mcasurable release burst at onset; however, the bursts were often abnormal for /d/ in either their onset characteristics, amplitudę, duration, or spectrum. The strength and quality of the talkers* vocalic cues to /d/ idiosyncratically depended upon the identity of the syllable vowel. A token-by-token analysis of the perception data showed that some constellations of /d/ cues yiclded relativcly high identification accuracy, others yieldcd high confidence, but only a smali subset yieldcd both. (Work supported by NIH and the Gallaudet Research Institute.) 6SP14. Development of phonc communication aid for hearing impaired. D. Kanevsky, P. S. Gopalakrishnan, G. Daggett, C. Danis, E. Epstein, and D. Nahamoo (Speech Recognition Group, IBM T. J. Watson Res. Ctr., P.O. Box 704, Yorktown Heights, NY 10598)

The goal of this paper is to describe work on the use of TANGORA—an Automatic Speech Recognizer (ASR) that was de-vefoped by the Speech Recognition Group at the Watson Research Center—as a communication device that would allow a hearing-impaired person to communicate with hearing individuals over the tele-phone. In this implementation, the speech of the hearing individuał is

dccodcd by the ASR, and the output is displayed on a Computer screen

for the hearing-impaired person. The generał usability of this system could be limited by the degradation in TANGORA’s recognition accuracy duc to (I) usc of public toll-quality tclcphone lines (instead of high bandwidth, low-noise communication lines) and (2) using it as a speaker-independent system (instead of trained to recognize each user).

121 sl Meeting: Acoustical Sociefy of America 1959