CAN COMODULATION OF ENVELOPES IMPROVE SIGNAL DETECTION IN ELECTRIC HEARING?

R. H. Pierzycki and B. U. Seeber

MRC Institute of Hearing Research, University Park, Nottingham NG7 2RD


Abstract: 

Patients with cochlear implants (CIs) experience difficulties with understanding speech in
the presence of modulated noise, which can be also observed in simulated CI-processing (Qin
and Oxenham, 2003). In normal hearing detection thresholds of a tone masked by an
amplitude-modulated on-frequency noise masker can be reduced when presented together with
coherently amplitude modulated flanking bands of noise. This effect seems to rely on
across-channel comparisons of temporal envelopes and has been termed comodulation masking
release (CMR) (Hall et al., 1984). However, within-channel cues which may display a
specific contribution of stem from temporal fine structure (TFS) information are known to
also contribute to CMR. Further, in CMR experiments a tone is usually detected in the
presence of noise, a task that may benefit from the perception of the tone's pitch, a cue
not available in CI-listening. The present studies investigate if CMR is available to
improve signal detection in CI listening without explicitly encoding TFS information and
thus they study the degree to which CMR indeed relies on the comodulation of envelopes
across multiple channels.  
CMR was investigated with a noise-band vocoder in preparation to testing CI-patients. A
noise-band vocoder extracts the envelope of the signal in frequency channels and uses it to
modulate carrier signals, here band-pass noises, thereby replacing the TFS of the signal. A
CMR paradigm was adapted to CI-listening by changing the width of noise bands to roughly
cover equal distances on the basilar membrane similar to electric stimulation. Further,
instead of widening the bandwidth of the noise masker we used a flanking band paradigm
where CMR is observed through a lower detection threshold of a tone in the presence of
co-modulated compared to anti-modulated flanking bands of noise. The flanking bands mimic
stimulation on separate, distant electrodes in electric hearing.  
Results show that CMR can be observed with and without vocoding of the CMR-stimuli and that
CMR is slightly reduced for vocoded stimuli. Removing the TFS in the vocoder thus slightly
reduces CMR, suggesting that TFS somewhat contributes to CMR. However, since strong CMR can
be observed despite vocoding the results demonstrate that CMR is a robust envelope-based
process and it is hoped that it could improve signal detection in CI-listening. 


References: 

Hall J.W., Haggard M.P. and Fernandes M.A. (1984) Detection in noise by spectro-temporal
pattern analysis, J Acoust Soc Am; 76: 50-56. 
Qin M.K. and Oxenham A.J. (2003) Effects of simulated cochlear-implant processing on speech
reception in fluctuating maskers, J Acoust Soc Am; 114: 446-454.