Limitations on temporal information in cochlear implants

Verschuur, C.A. and Platt, A. (2007) Limitations on temporal information in cochlear implants. British Society of Audiology Short Papers Meeting on Experimental Studies of Hearing and Deafness, London, UK, 20-21 September 2007, London, UK. 19 - 20 Sep 2007.

Abstract

The objective of this poster is to demonstrate the limitations on temporal information that can be imposed by a cochlear implant (CI) processor which employs the Fast Fourier Transform (FFT) to decompose the signal. The Nucleus 24 processor increases temporal information with increased stimulation rate by increasing the temporal overlap between FFT analyses (increases FFT rate). Benefits to performance with higher stimulation rates have been shown in some cases (Holden et al., 2002), but there is little evidence to determine whether improved performance (where it occurs) is due to genuine increases in temporal information in the processor output. The study reported here aimed to determine changes in electrode output with increasing FFT rate, and to evaluate the possible perceptual effects of this via an acoustic model. In order to achieve the first aim, sinusoids modulated at varying modulation rates (at 100% modulation depth) were processed through the Nucleus Matlab Toolbox in order to simulate the processing of the Nucleus 24 system. The modulation depth of the resulting electrode output was analysed for each modulation rate. The resulting temporal modulation transfer functions will be reported for three different FFT rates, but preliminary analysis shows only small differences across FFT rates. A corresponding acoustic model study was also undertaken. Processed sentence materials were presented monaurally to 10 normal hearing subjects via an ER-3 insert earphone. The speech materials were processed through the same software as used for the acoustic measurements, in order to the mimic the processing of the Nucleus 24 device implementing the Advanced Combination Encoder speech processing strategy (Cochlear, 2002), at three FFT rates equivalent to stimulation rates of 250, 900 and 2000 pulses per second. Noise bands were modulated according to the fluctuations in the virtual electrode channels of the processor and added together to produce simulated sentence materials. Performance on a sentence recognition task was compared between the three FFT rates in quiet and stationary noise at +10 dB SNR. Results will be presented and discussed in the context of how FFT-based CI processing impacts on temporal information and whether changes in stimulation rate produce real changes in electrode output.

This record has no associated files available for download.

Contributors

Download statistics