Download 13th International Conference on Cochlear Implants and Other

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts

Auditory system wikipedia, lookup

Audiology and hearing health professionals in developed and developing countries wikipedia, lookup

Earplug wikipedia, lookup

Olivocochlear system wikipedia, lookup

Dysprosody wikipedia, lookup

Sensorineural hearing loss wikipedia, lookup

Noise-induced hearing loss wikipedia, lookup

Hearing loss wikipedia, lookup

Speech perception wikipedia, lookup

Telecommunications relay service wikipedia, lookup

Transcript
P2-10-7
Spectral contrast enhancement in CI coding strategies: A real-time implementation
1
2
Rode T. , Büchner A. , Nogueira W.
2
1
HZH GmbH, Hannover, Germany, 2Medical School Hannover, Hearing4all, Department of Otorinolaringology, Hannover, Germany
Hearing performance in difficult situations as in noisy or reverberant environments has been shown to correlate
with the ability to resolve spectral information. In hearing impaired listeners this ability is often poor since spectral
sharpening mechanisms of the inner ear are affected by the hearing loss. Previous studies observed an
improvement of the quality and intelligibility of speech in noise in hearing impaired listeners when enhancing the
spectral contrast of the speech signals and thus compensating the effect of spectral smearing. In a recent study
identification of spectrally smeared vowels and consonants was improved by spectral contrast enhancement
(SCE) in a group of 166 normal hearing listeners [Alexander et al.]. Spectral resolution for CI users is degraded
because of the limited number of stimulation electrodes and overlapping electric fields activating the nervous
system through the bony structure of the cochlear. Loizou et al. showed that CI users need a higher spectral
contrast than normal hearing listeners in vowel identification tasks. Also, SCE can improve phoneme and
sentence perception in noise.
We implemented a modified version of the algorithm introduced by Loizou et al. within a CI coding strategy using
Matlab Simulink. Our implementation keeps all prominent formant peaks within a speech signal constant while
attenuating valleys in the spectrum, thus increasing the spectral contrast. The amount of SCE depends on the
original contrast found in the signal and can be controlled by a single parameter which we call the SCE factor.
For a clinical evaluation of the algorithm, the Simulink model was compiled on an XPC real-time target to test
subjects in a sound attenuated chamber. Speech and noise were presented in free field simultaneously through
the same loudspeaker in front of the subject at a 1 m distance. The SNR was adapted to the performance of
each CI user such that they scored between 25% and 75% on the HSM sentence test. Two lists were presented
for each condition (SCE enabled and disabled) and the amount of correct words was counted and averaged for
each study participant.
Preliminary results from 7 CI users show an average of 6% improvement of correct words in CCITT noise using
an SCE factor of 1. 6 out of 7 subjects obtained an improvement with SCE enabled and only 1 out of 7
decreased in performance. Generally, bad CI performers obtained a higher improvement than better CI
performers.
References:
Alexander, J. M., Jenison, R. L., & Kluender, K. R. (2011). Real-time contrast enhancement to improve speech
recognition. PloS one, 6(9), e24630. doi:10.1371/journal.pone.0024630
Loizou, P. C., & Poroy, O. (2001). Minimum spectral contrast needed for vowel identification by normal hearing
and cochlear implant listeners. The Journal of the Acoustical Society of America, 110(3), 1619.
doi:10.1121/1.1388004
1151