Header

UZH-Logo

Maintenance Infos

Mechanisms of music perception through cochlear implants


Omran, S A. Mechanisms of music perception through cochlear implants. 2011, University of Zurich, Faculty of Medicine.

Abstract

Cochlear Implants (CIs) are devices aimed at restoring hearing with electrical stimulation of the hearing nerve. The average unrolled length of the cochlea is ~33 mm with a spiral shape of 2½ turns (~ ½ cm Ø) and a tonotopical arrangement. The Nucleus CI’s electrode array typically occupies the basal 1½ turns, corresponding to haracteristic frequencies of about 500-1000 Hz. Deeper insertion of the array to reach lower characteristic frequencies is hindered partly by the array’s dimensions. Additionally, the number of electrodes is limited. As a consequence the standard (Std) Nucleus ACE (Advanced Combination Encoder)frequency mapping presents acoustically relevant information (188-7930 Hz) on upto 22 electrodes, mapping these frequencies to locations with tonotopical characteristic frequencies
between ~500 - 1000 to ~ 6000 Hz. Stimulating an electrode produces an electric field that excites neighbor positions of the auditory nerve. Stimulating two electrodes simultaneously produces a virtual channel (VC) leading to the perception of an intermediate pitch. The VCs concept has been integrated into an acoustic model (AMO) of CI stimulation. New frequency to electrode mappings such as Greenwood and Semitone (Smt) mapping have been proposed
and implemented. Basic parameters of the AMO model were optimized by matching results from a sentence recognition test with normal hearing (NH) subjects to results from CI
recipients. Pitch discrimination tests using pure and complex tones with the Std mapping processed by the AMO were conducted with NH subjects to examine the effect of VCs on the score. The results showed a trend towards enhanced pitch discrimination scores with VCs. In a further step simple synthetic musical tones processed with the Std mapping and resynthesized using the AMO were inspected with a three dimensional (time, critical band and loudness) psychoacoustic spectral-contrast analysis. The output showed deterioration in the harmonic analysis structure of overtones, especially at lower frequencies with the Std
mapping. The spectral-contrast analysis emphasized the results of the pitch discrimination test with pure tones where wider semitone intervals in higher octaves were easier distinguishable whereas complex tones were poorly resolved. This may coincide with the description of
perceived music which many patients characterized as being an unpleasant noise. Technically, music can be described as a series of complex acoustic sounds composed of tones with
fundamentals and overtones that are harmonically related. Music has many aspects such as pitch, melody, harmony, rhythm and timbre. Preserving the harmonic structure enhances melody representation, even if the fundamental frequency component is not explicitly encoded. The two proposed Smt mapping ranges (Smt-LF [130-1502] Hz and Smt-MF [440- 5040] Hz therefore were designed to preserve harmonic structures in musical sounds. Electrodograms of a sequence of synthetic piano tones processed with different mappings (Std, Smt-LF and Smt-MF) were compared using an algorithm employing a newly proposed distance matrix. The results showed that at least one of the Smt mapping ranges preserves the harmonic structure better than the Std mapping. In a further examination, music tones were processed with the three mappings (Std, Smt-LF and Smt-MF) and a proposed harmonicity index was calculated. The analysis showed similar results. The Smt mapping was also
evaluated by CI recipients. Testing was conducted using a melody contour identification (MCI) and instrument recognition (IR) tests by directly streaming offline-processed data to the implant. MCI results showed that the Smt mapping may improve the scores or at least produces similar results to the Std mapping. However, the average scores from the IR test were decreased suggesting a general deficiency in coding of timbre.

Abstract

Cochlear Implants (CIs) are devices aimed at restoring hearing with electrical stimulation of the hearing nerve. The average unrolled length of the cochlea is ~33 mm with a spiral shape of 2½ turns (~ ½ cm Ø) and a tonotopical arrangement. The Nucleus CI’s electrode array typically occupies the basal 1½ turns, corresponding to haracteristic frequencies of about 500-1000 Hz. Deeper insertion of the array to reach lower characteristic frequencies is hindered partly by the array’s dimensions. Additionally, the number of electrodes is limited. As a consequence the standard (Std) Nucleus ACE (Advanced Combination Encoder)frequency mapping presents acoustically relevant information (188-7930 Hz) on upto 22 electrodes, mapping these frequencies to locations with tonotopical characteristic frequencies
between ~500 - 1000 to ~ 6000 Hz. Stimulating an electrode produces an electric field that excites neighbor positions of the auditory nerve. Stimulating two electrodes simultaneously produces a virtual channel (VC) leading to the perception of an intermediate pitch. The VCs concept has been integrated into an acoustic model (AMO) of CI stimulation. New frequency to electrode mappings such as Greenwood and Semitone (Smt) mapping have been proposed
and implemented. Basic parameters of the AMO model were optimized by matching results from a sentence recognition test with normal hearing (NH) subjects to results from CI
recipients. Pitch discrimination tests using pure and complex tones with the Std mapping processed by the AMO were conducted with NH subjects to examine the effect of VCs on the score. The results showed a trend towards enhanced pitch discrimination scores with VCs. In a further step simple synthetic musical tones processed with the Std mapping and resynthesized using the AMO were inspected with a three dimensional (time, critical band and loudness) psychoacoustic spectral-contrast analysis. The output showed deterioration in the harmonic analysis structure of overtones, especially at lower frequencies with the Std
mapping. The spectral-contrast analysis emphasized the results of the pitch discrimination test with pure tones where wider semitone intervals in higher octaves were easier distinguishable whereas complex tones were poorly resolved. This may coincide with the description of
perceived music which many patients characterized as being an unpleasant noise. Technically, music can be described as a series of complex acoustic sounds composed of tones with
fundamentals and overtones that are harmonically related. Music has many aspects such as pitch, melody, harmony, rhythm and timbre. Preserving the harmonic structure enhances melody representation, even if the fundamental frequency component is not explicitly encoded. The two proposed Smt mapping ranges (Smt-LF [130-1502] Hz and Smt-MF [440- 5040] Hz therefore were designed to preserve harmonic structures in musical sounds. Electrodograms of a sequence of synthetic piano tones processed with different mappings (Std, Smt-LF and Smt-MF) were compared using an algorithm employing a newly proposed distance matrix. The results showed that at least one of the Smt mapping ranges preserves the harmonic structure better than the Std mapping. In a further examination, music tones were processed with the three mappings (Std, Smt-LF and Smt-MF) and a proposed harmonicity index was calculated. The analysis showed similar results. The Smt mapping was also
evaluated by CI recipients. Testing was conducted using a melody contour identification (MCI) and instrument recognition (IR) tests by directly streaming offline-processed data to the implant. MCI results showed that the Smt mapping may improve the scores or at least produces similar results to the Std mapping. However, the average scores from the IR test were decreased suggesting a general deficiency in coding of timbre.

Statistics

Downloads

238 downloads since deposited on 18 Jan 2012
16 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Dissertation (monographical)
Referees:Douglas R, Dillier N, Hahnloser R
Communities & Collections:04 Faculty of Medicine > University Hospital Zurich > Clinic for Otorhinolaryngology
Dewey Decimal Classification:610 Medicine & health
Language:English
Date:2011
Deposited On:18 Jan 2012 16:36
Last Modified:17 Feb 2018 14:46
Number of Pages:217
OA Status:Green

Download

Download PDF  'Mechanisms of music perception through cochlear implants'.
Preview
Language: English
Filetype: PDF
Size: 10MB