Graeme Clark Collection

Permanent URI for this collection

http://hdl.handle.net/11343/375

Filters

1992 2
Reset filters

Settings
Statistics
Citations
Author: Cowan, Robert × End date: 1999 × Start date: 1992 × Type: Conference Item × Subject: cochlear implant × Subject: otolaryngology × Subject: speech processing ×

Search Results

Now showing 1 - 2 of 2
  • Item
    Thumbnail Image
    Preliminary speech perception results for children with the 22-electrode Melbourne / cochlear hearing prosthesis
    Sarant, J. Z. ; Clark, Graeme M. ; Cowan, Robert S. C. ; Dowell, R. C. ; Pyman, B. C. ; Dettman, S. J. ; Dawson, P. W. ; Rance, G. ( 1992)
    The Cochlear 22-electrode cochlear prosthesis was first implanted in a child at the Cochlear Implant Clinic at the University of. Melbourne and Royal Victorian Eye & Ear Hospital in 1985. Since that time 42 children have received the device in Melbourne. Analysis of patient details shows that the majority of these children have a congenital as opposed to on acquired aetiology of hearing loss. In all but 3 cases, the children use 15 or more electrodes. In order to assist with evaluation of, benefits to speech perception across the very heterogeneous group of children, a six level hierarchical classification scheme for speech perception performance levels was created. All of the children achieved a minimum of Category 2 (discrimination of suprasegmental information). In total, 59 % of the children achieved Category 5 or 6 (open set recognition for unfamiliar materials). Analysis showed that the majority of these children had more than one year of experience. In contrast, the majority of children in Category 2 are those with less than one year of experience with the device.
  • Item
    Thumbnail Image
    Improvements in speech processing for the nucleus cochlear implant [Abstract]
    Cowan, Robert S. C. ; Dowell, R. C. ; McDermott, H. D. ; McKay, C. ; Clark, Graeme M. ( 1992)
    The Nucleus Cochlear implant allows ~ variety of multiple channel speech processing strategies to be developed and trialled. The initial strategy first developed by the University of Melbourne presented the second formant frequency as place and voicing as rate of stimulation. The strategy was subsequently improved by presenting the first format on a place basis as well. Since that time the addition of more spectral information coded as place of stimulation and temporal information presented as variations in amplitude have resulted in better open-set CNC word and sentence scores for electrical stimulation alone. One of the improved strategies selects four pairs of electrodes from the 22 electrode array each glottal pulse to present the first and second formants as well as the output from two high Frequency band pass filters. The other strategy stimulates six of the 22 electrodes representing the six maximal outputs from 16 bandpass filters. The clinical results have shown that both the above strategies results in better open-set speech perception for electrical stimulation in quiet and in noise. 80 % scores have been obtained for open-sets of CNC words and 90 % for open-sets of words in sentences for some of the patients using the latter strategy.