Results from a strategy encoding temporal fine structure

Abstract

Current cochlear implant stimulation strategies are widely based on the continuous interleaved sampling (CIS) paradigm. With CIS, temporal fine structure is only crudely represented in the stimulation pattern, and pitch is encoded primarily in the place of stimulation and to a limited extent in the temporal fluctuations of the channel envelopes. However, fine structure information is a crucial ingredient for the perception of music and of tonal languages, as demonstrated in studies with normal hearing subjects. In the fine structure strategy presented, “channel specific sampling sequences” (CSSS) or pulse-packages are applied to the apical electrodes, while the remaining basal electrodes carry CIS-like stimuli. The pulse sequences are triggered by the zero-crossings of the corresponding filter channel outputs and scaled with the channel envelopes. Thus, both fine time structure and envelope information are represented on CSSS channels. The new strategy has been evaluated in two experiments: In a first experiment, pitch discrimination and scaling abilities were compared for the fine structure strategy and CIS in four MED-EL implant recipients. For pure tone stimuli below 300 Hz, pitch discrimination with CIS was limited. Conversely, CSSS seems to add robust temporal cues to pitch at low frequencies, supporting better discrimination. In a second experiment, speech reception measures were taken for MEDEL implant recipients and native speakers of Cantonese Chinese in Hong Kong. Preliminary results from a pilot study showed substantial improvements in the perception of lexical tones with the fine structure strategy over CIS and triggered a more extensive survey. Data from this ongoing study will be presented. Support provided by the Christian Doppler Research Association