Zoom out Search Issue

ManualsBrandsContents Manualsaudio & home theatreZoom in

IEEE SIGNAL PROCESSING MAGAZINE [79] MARCH 2015

thresholds around 50–100 μs and, in the worst case, no ITD

sensitivity at all. However, with commercial sound processors,

subjects hardly use ITDs in ecologically relevant tasks such as

sound source localization. This is at least partly due to poor

coding of temporal cues by current commercial sound process-

ing strategies. The delay and spectral characteristics of the pro-

cessing paths can be very different for the left and right CI

devices, and certainly for bimodal systems where a CI is used in

one ear and an acoustic hearing aid in the other ear. This may

lead to nonsynchronous and noncoordinated (across channels)

left and right auditory stimulation.

One of the first strategies developed to improve ITD coding

with bilateral CIs is the peak derived timing (PDT) strategy [26].

It operates by synchronizing stimulation pulses from the CI

with amplitude peaks in the fine structure of the signals in the

different channels of the filter bank. In this manner, fine pulse

timing cues are transmitted, in contrast with CIS-type sound

processing techniques that provide only envelope information

with fixed stimulation rates. Evaluations are reported in [22].

As bilateral CI strategies like PDT can introduce temporal

patterns that are not synchronized with the acoustic signal, the

modulation enhancement strategy (MEnS) was proposed [27]

for bimodal stimulation. A deeply modulated envelope is

imposed on all frequency channels simultaneously, explicitly

synchronized with peaks in the acoustic signal. Improved ITD

thresholds and improved lateralization (the extent to which the

sound image can be moved to either side of the head by only

changing ITD) were found with MEnS compared to ACE.

While some improvements in ITD perception have been

obtained in laboratory tests with experimental strategies, the

same caveats hold as with the pitch strategies described in the

section “Periodicity Modulation Enhancement”: performance is

much poorer than with NH. It should be noted, however, that

thus far only acute experiments have been performed, while lis-

teners potentially need long-term exposure to the novel stimula-

tion paradigm to learn to use the binaural timing cues provided.

GENERAL DISCUSSION

In this article, a tutorial of CI stimulation strategies was pre-

sented, together with a review of concepts and rationales of differ-

ent standard and experimental processing schemes. Some of the

newer schemes have demonstrated significant improvements in

the understanding of speech and perception of other types of

sound. Although each of these strategies may lead to only a small

benefit, it is plausible that appreciably larger benefits may be

obtained when they are combined. Furthermore, some signal-

processing approaches introduce speech enhancements in noisy

conditions at the cost of significant signal distortions. These dis-

tortions may be detrimental for sound quality when appraised by

listeners with normal or impaired acoustic hearing, but are

hardly noticeable by most CI recipients. This is an opportunity for

further improvements in auditory perception for CI users.

However, the broad neural excitation profiles inherent to

present-day electrode array technology and electrical stimulation

parameters most probably limit the potential for improvement.

The number of independent information transmission chan-

nels is still very small because of both technical and percep-

tual/neural sensitivity limitations. Not all CI users can

discriminate all channels, but even if all actual and virtual

stimulation channels and electrodes may be perceptually dis-

criminated from each other, this does not imply that channels

can be resolved, nor that different channels can effectively con-

vey independent information.

It has become clear that some temporal aspects of the input

sound, such as the speech envelope and partly periodicity can be

transmitted faithfully by CIs. However, the TFS and F0 are not

adequately represented in present-day CI processors and are there-

fore presented to the auditory neural system only imprecisely.

Auditory perception results can be spectacular for many CI

recipients in quiet environments, particularly in early-

implanted deaf children when neural plasticity can fully play

its role and in adults with a largely intact neural periphery.

However, hearing in realistic adverse listening situations, as

well as music perception and sound source localization are

still major challenges for sound coding and electrical stimula-

tion in CIs. Also, a wide variation in outcomes is observed

across CI users. A significant proportion of recipients receive a

limited benefit from their CI, at least in terms of speech

understanding. Some investigations indicate that a better indi-

vidual fitting of the stimulation parameters (the map) may

result in substantial improvement, be it by better selection of

active channels [28] or by development of closed-loop auto-

matic fitting paradigms [29].

Another important factor is the neural survival at the

electrode-neuron interface in the auditory periphery, which

may be improved by application of drugs such as neurotroph-

ins. Future research will include a greater focus on the com-

bination of nonstandard pulse waveforms [30], new

stimulation modes to reduce across-channel interactions,

and improved electrode designs. These approaches may

result in the provision of more independent information

channels in future CI systems.

DECLARATION OF INTEREST

We have been involved in the development of the following strate-

gies: SMSP, ACE, Speak, SpeL, SCORE (Hugh J. McDermott);

F0mod, MEnS, SCORE, (Jan Wouters and Tom Francart); and EE

(Jan Wouters).

ACKNOWLEDGMENTS

We thank Bas van Dijk (Cochlear), Leonid Litvak (Advanced

Bionics), Peter Nopp (Med-El), and Dirk Meister (Med-El) for

providing the data used to generate the electrodogram plots,

information on the strategies, and insightful comments on the

respective parts of the manuscript. We apologize for not being

able to include all significant references that we wanted due to

space constraints. We are grateful to Jonas Vanthornhout for his

assistance with the figures. The Bionics Institute acknowledges

the support it receives from the Victorian Government, Austra-

lia, through its Operational Infrastructure Support Program.

THE WORLD’S NEWSSTAND

THE WORLD’S NEWSSTAND