Graeme Clark Collection
Permanent URI for this collection
Search Results
1 - 10 of 22
-
ItemSpeech processing for a multiple-channel cochlear implant( 1980)A laboratory speech processor has been developed for a multiple-channel cochlear implant prosthesis. The speech processor accepts the speech waveform as an input and produces a pattern of electrical stimulus data as output. The electrical stimulus data are transmitted to the implanted receiver-stimulator by a transmitter which is external to the speech processor. Four speech signal parameters were estimated every 20 ms in the parameter estimation section of the speech processor. These parameters included the fundamental frequency (FO), a low frequency energy measure (AO) , the second formant frequency (F2) and its amplitude (A2).
-
ItemPsychophysical studies for a multiple-channel cochlear implant( 1980)This paper describes the results obtained in a series of psychophysical studies conducted with our first multiple-channel cochlear implant patient. The variations of apparent loudness and pitch for single-electrode stimulation were determined by the method of magnitude estimation. The results showed that the loudness growth due to increases in current level was much steeper than the growth for acoustic stimulation in normal hearing subjects. The pitch produced by electrical stimulation was found to increase with pulse rates below 200 pps, while the increase in pitch with pulse rate was less pronounced above 200 pps. For a constant rate of stimulation, the pitch varied with the electrode position. Furthermore, the same pitch estimate could be obtained by driving single electrodes with different pulse rates.
-
ItemSpeech processor design for a multiple-channel cochlear implant( 1980)This paper outlines the strategy adopted for a laboratory-based speech processor used to provide speech information to patients with a multiple-channel cochlear implant It also presents the results of vowel and consonant recognition studies and speech test using open sets of words and sentences.
-
ItemSpeech comprehension with multiple-channel electrical stimulation of human auditory nerve fibres [Abstract]( 1980)The psychophysics of multiple-channel electrical stimulation of residual auditory nerve fibres has been studied in two postlingually deaf patients with total hearing losses, and the information used to design a speech processing strategy that enables them to understand connected speech in everyday situations when used in conjunction with lipreading. A constant rate of stimulation of different groups of auditory nerve fibres produced different pitch sensations which could be scaled from low to high, and this was consistent with the place theory of frequency coding. The me patients were also able to associate the different percepts produced by each electrode with different vowel colours and sharpnesses. In addition, the patient reported that pitch increased with pulse rate and loudness increased with current level. A speech processing strategy was developed whereby the parameters for voicing determine the rate of stimulation for all electrodes, and the parameters for the dominant spectral peak in the mid-frequency range determine the site of electrode stimulation and current level.
-
ItemEvoked responses in humans to tones which are sinusoidally amplitude modulated [Abstract]( 1980)Evoked responses have been recorded from the scalp of normal human subjects to continuous sinusoidally-modulated Amplitude-modulated tones. The responses were periodic in nature, and a Fourier transform was used to quantify the amplitude and the phase of the constituent fundamental frequency and harmonic components. The fundamental frequency of the response equalled the frequency of the modulation envelope and the amplitude of the harmonic components was usually less than the fundamental. The responses were also found to be invariant with time for periods exceeding 30 minutes.
-
ItemThe rehabilitation of patients with a total hearing loss using a multiple-channel cochlear implant [Abstract]( 1980)Two profoundly deaf patients who have multiple-channel cochlear implants receive speech information through a speech processor. The speech processor encodes second formant and fundamental frequency information. Rehabilitation is concerned with evaluating the speech processor and training patients to improve speech perception. The patients have scores up to 37% and 27% on AB word lists, and 36% and 34% on NAL Everyday Sentences using the cochlear implant alone and repetition of the stimuli. However, to gain a better insight into the capabilities of the speech processor and to demonstrate its potential for the patients more investigative procedures have been used. These include the speech tracking method described by DeFillipo and Scott (1978) using the cochlear implant in conjunction with speech reading. Results indicate that patients have gained a two-fold and four-fold improvement in tracking rates (words per minute) using speech reading and the cochlear implant as opposed to speech reading alone.
-
ItemA speech processing strategy for an electro-tactile vocoder [Abstract]( 1980)Past attempts at using the skin for recognition of tactile patterns derived from acoustic speech signals have largely been unsuccessful for perception of running speech. Problems facing researchers in this field include: frequency discrimination, especially for electrical stimulation, temporal and spatial resolution, real time speech processing and tactile pattern configuration strategies. It is considered that recent developments in speech processing which allow real time estimation of formant frequencies and vocal tract area functions will enable a successful speech aid to be developed. Based on results of the Tadoma (or Hofgaard) Method, in which speech is perceived by the deaf-blind using tactile and kinesthetic senses to determine movements of a speaker's articulators, a model is evaluated which enables a tactile display of articulatry information derived from parameters extracted from the speech signal by real time speech processing. Psychophysical measurements of percepts of computer derived patterns were carried out concentrating in particular on patterns more likely to be important for phonemic and speech discrimination. In this way it is hoped to validate the model as a useful speech aid for the profoundly and partially deaf.
-
ItemSpeech processing for a totally deaf patient with a multiple-electrode cochlear implant [Abstract]( 1980)A patient with a total hearing loss had a multiple-electrode cochlear implant performed on 1 August 1978 (Clark et. Al. 1979). Preliminary findings (Tong et al 1979) have shown that low and high pitch are discriminated on a place basis. Stimulating different electrodes produced the vowel colours.......
-
ItemBiophysical design considerations of the cochlear prosthesis [Abstract]( 1980)In order for the cochlear prosthesis to transfer sufficient information to primary auditory nerve fibres for recognition of speech-like sounds, it is necessary to electrically excite discrete fibre populations at several points along the cochlear spiral (Tong et al., 1979). In addition, because of the large number of stimulus electrodes and the small volume of the scala tympani in which the electrodes are usually implanted, current transfer must derive from "passive" electrochemical processes in order to preserve the normal biochemical environment of the nerve fibres. From measurements made in the human cochlea during implantation of the cochlear prosthesis, it has been shown that the longitudinal ground current distribution arising from a multielectrode system incorporating a common intracochlear ground electrode is accurately replicated by measurements made in a saline-filled uniform tube of diameter 3-5 rom. Further measurements in such tubes are described which investigate the possibility of improved current localisation by reduction of the ground electrode impedance and the possibility of producing greater electrochemical "passivity" by the use of voltage sourcing in preference to current sourcing.
-
ItemCritical bandwidth in the cat following the selective destruction of cochlear inner and outer hair cells [Abstract](Australian Physiological and Pharmacological Society, 1978)It is of interest to determine the significance of the inner and outer hair cells of the cochlea in critical bandwidth measurements, as there is evidence for a difference in their roles in frequency selectivity (Dallos et at., 1977) and in frequency discrimination (Nienhuys and Clark, 1978). The present study has been undertaken on-four monauralized cats which were trained to respond by a conditioned suppression technique. Behavioural auditory thresholds were measured by plotting suppression ratios for frequencies of 1kHz, 4kHz, 8kHz, 10kHz, 12kHz and 16kHz. Critical bands were measured at the same frequencies by recording the pure tone thresholds in the presence of a masking noise with six different bandwidths, a constant total power, and geometrically centred on the test tone (Greenwood, 1961). The cats were then given a series of intramuscular injections of kanamycin (200mg kg-l d- l ) for 10 days to selectively destroy the outer hair cells, and behavioural thresholds determined shortly afterwards. Critical bands were again measured at all the previous frequencies except 16kHz, where the auditory behavioural threshold was too high. At the completion of the experiments behavioural thresholds were re-determined, the animals were anaesthetized with pentobarbital sodium (40mg/kg, i.p.) and auditory nerve action potentials recorded.