Graeme Clark Collection
Permanent URI for this collection
Search Results
1 - 10 of 12
-
ItemLoudness growth characteristics of cochlear implantees using the Spectral Maxima Sound Processor [Abstract]( 1994)The study of perceptual characteristics of subjects with cochlear implants can lead to improvements in the design of speech processors. One important aspect of speech processing which has received little attention in the past is the conversion acoustic signal amplitudes into appropriate levels of electrical stimulation. The optimum conversion would provide implantees with loudness growth characteristics that mimic those of normal hearing. To investigate how implantees using the Spectral Maxima Sound Processor (SMSP) perceive changes in loudness, an experiment involving production of fixed loudness ratios was conducted. Ten subjects participated: five users of the Mini System 22 cochlear implant, and five normally-hearing subjects. In the experiment, the subjects were required to adjust the loudness of two stimuli (white noise and speech-weighted noise) to equal half or twice that of a reference. The reference was presented at various levels over a range of 25 to 75 dBA. The results for three of the implantees were similar to those of all the normally-hearing subjects, who produced an average level change of 10.8 dB for the task. The remaining subjects, who had the largest electrical dynamic ranges, produced larger level changes (up to 20 dB) which were constrained by the limited electrical dynamic range of the processor (46 dB). The SMSP utilises an amplitude conversion function by which the stimulus level (in dB) is directly proportional to the input sound level (in dB). The experimental results suggest that the shape of this function is satisfactory, though not necessarily optimum, for these implantees.
-
ItemPitch percepts associated with amplitude-modulated current pulse trains in cochlear implantees( 1994)Abstract not available due to copyright.
-
ItemComparison of current speech coding strategies( 1993)This paper reports on two studies carried out at the University of Melbourne jointly with Cochlear Pty Ltd. The studies demonstrated substantial speech perception improvements over the current Multipeak strategy in background noise.
-
ItemThe pitch of amplitude-modulated electrical stimuli in cochlear implantees [Abstract]( 1993)The ability of cochlear implantees to detect amplitude modulation of pulsatile electrical stimulation, suggests that some speech feature information may be conveyed effectively by this means. For example, modulations at the fundamental frequency of speech may provide a voice pitch percept to implantees, particularly in speech processing strategies which generate constant-rate stimulation. The pitch evoked by sinusoidally modulated current pulse trains on a single electrodes has been studied. Modulation frequencies of 100, 150 and 200Hz, and carrier pulse rates varying from 200 to 1200Hz, were used. The results showed that the pitch of the stimulation was related to the modulation frequency, provided that either the carrier rate was a multiple of the modulation frequency, or the carrier rate was sufficiently high (at least four times the modulation frequency for the stimuli studied here). Furthermore, when the modulated stimuli were matched in pitch to non-modulated pulse trains, it was. found that the rate of the matched non-modulated stimuli was close to but somewhat higher than the modulation frequency. This difference depended on the carrier rate and varied among subjects.
-
ItemA new portable sound processor for the University of Melbourne/ Nucleus Limited multielectrode cochlear implant( 1992)Abstract not available due to copyright.
-
ItemPlace pitch perception with multiple electrode cochlear implants: the use of concurrent activation of nearby electrodes to produce additional ptich percepts [Abstract]( 1992)In multiple electrode cochlear implants, each electrode produces a pitch percept which is usually related monotonically to its distance from the round window. The number of these pitch percepts is limited by the number of usable electrodes and their discriminability, varying up to a maximum of 22 for the mini-system 22 implant but sometimes significantly less. A study on two implanted subjects in which the pitch of pulse trains on two concurrently activated nearby electrodes was compared with the pitch produced when each of the electrodes was activated on its own, showed that the pitch of the concurrently activated electrodes was different from each component electrode and was placed in an intermediate position. Furthermore the pitch of the concurrent stimulation could be altered by adjusting the relative current levels on the two component electrodes. This may partly explain the improvements, particularly in vowel discrimination, obtained with the SMSP strategy described in the accompanying paper.
-
ItemPreliminary results with a six spectral maxima speech processor for The University of Melbourne/Nucleus multiple electrode cochlear implant( 1991)An improved sound processor for a multiple-channel cochlear implant hearing prosthesis has been developed. The spectral maxima sound processor (SMSP) extracts the six largest frequency components of speech and presents stimuli at a rate of 250 Hz to electrodes at positions selected on the basis of the spectral frequencies. It was designed for use initially with the advanced cochlear implant recently developed at the University of Melbourne, which is capable of high rate and quasi-simultaneous stimulation. The present study, however, was carried out with two subjects who have the more widely used 22-electrode implant produced commercially by Cochlear Pty Limited (formerly Nucleus Limited). Preliminary results comparing the performance of the SMSP with that of previous speech processing techniques (F0/F1/F2 strategy) are presented. The results indicate that the SMSP is capable of providing implanters with significantly greater information about speech.
-
ItemSpeech processing for cochlear implants: variations of the spectral maxima sound processor( 1993)The Spectral Maxima Sound Processor (SMSP) incorporates a bank of sixteen band-pass filters which are assigned to sixteen intracochlear electrodes. In each stimulation period six electrodes are activated, based on the outputs of the filters with the largest amplitudes. The SMSP has been compared with the present MSP(MULTIPEAK) processor and found to improve speech comprehension results. The SMSP speech processing scheme has recently been implemented successfully in a new speech processor, also developed at the University of Melbourne, which utilises digital signal processing techniques. The programming flexibility of the new processor has facilitated the investigation of variations of the SMSP strategy which might provide further enhancement of speech perception. Three variations have been investigated: firstly, increasing the constant pulse rate from the usual 250Hz to 400Hz; secondly, changing the number of electrodes selected in each stimulation period from six to numbers between four and eight; thirdly, sharpening the spectral peaks prior to selection of the active electrodes. The results of these studies showed that all three variations had minimal effect on speech perception in quiet, but that increasing the number of electrodes selected for stimulation to eight, or increasing the rate of stimulation, may have advantages when listening in background noise.
-
Item
-
ItemAn improved speech processor for a 22-electrode cochelar implant [Abstract]( 1992)A spectral maxima speech processing strategy ( SMSP) has been developed as a result of research to improve the speech perception performance of a multiple-channel cochlear implant. With this speech processing strategy. the six spectral maxima from the outputs of 16 band pass filters are used to stimulate the cochlea on a place basis at a constant rate. This SMSP strategy has been compared with the MSP-MULTIPEAK strategy, the present speech processor provided by Cochlear Pty. Limited, on four postlinguistically deaf adults. The study showed that the SMSP strategy was significantly better than the MSP-MULTJPEAK for the recognition of closed-set vowels and consonants, and open-set monosyllable words and sentences in background noise.