A Novel Method to Assess Contributions of Spectral and Temporal Pitch Cues in Realistic Speech Stimuli

Purpose: Pitch perception is known to be significantly impaired in cochlear implant (CI) users. Previous research has indicated that pitch perception of CI users depends mainly on temporal envelope cues. However these studies used either well-controlled but artificial stimuli or indirect measures of assessing cues. A recent study examining electrodograms obtained with more realistic stimuli of speech concluded that temporal envelope cues were not particularly salient. The authors proposed that some spectral pitch cues may also be available. Using a novel approach, this study aims to directly quantify the relative contributions of spectral and temporal cues to pitch discrimination of CI users in speech stimuli.<br/><br/>Methods: Psychometric functions for fundamental frequency (f0) discrimination were collected in CI listeners, using triplets of consonant-vowel syllables in a three-alternative forced choice task. The preliminary data includes five late-implanted, adult CI users. Two modalities of stimulation were used: acoustic, via a loudspeaker with the participants own processor, and electric, through a research interface. For the latter, electrodograms were recorded in advance with the participants' own program settings, as if they were seated in the soundproof booth. Then, the electrodograms were processed to produce four conditions: unaltered, temporal cues suppressed, spectral cues suppressed, and both cues suppressed. Temporal cues were suppressed by filtering the temporal envelope of each channel with a comb-filter based on the f0. Spectral cues were suppressed by raising or lowering current levels to produce similar spectral envelopes for all three triplets, while mostly preserving the temporal envelopes of each channel. During the experiment, the electrodograms from the four conditions were streamed to the CI user. This recording and restreaming method allowed us to precisely control the available cues while working with natural speech stimuli and incorporating all the bells and whistles of a modern processor.<br/><br/>Results: The preliminary data shows that the acoustic and unaltered electric streaming conditions have similar performance. Suppressing temporal cues decreases pitch discrimination performance, i.e. shifting the psychometric function towards larger f0 differences. Suppressing spectral cues decreased performance of one CI user, although the effect was smaller than suppressing temporal cues.<br/><br/>Conclusions: The similar performance between acoustic and unaltered electric conditions verifies that the streaming of prerecorded stimuli can give a near-normal electric hearing percept. Therefore, this novel method is capable of producing realistic speech stimuli while giving the researcher precise control over the stimulation. The preliminary data on suppressing pitch cues suggests that CI users are mostly relying on temporal cues for pitch discrimination, while spectral cues are utilized less and only by some. Although these findings mostly agree with the previous literature, it seems that some users might be utilizing spectral cues in their pitch perception. Which CI users are utilizing what cues remains to be seen.<br/><br/>Acknowledgements: This work was funded by the Faculty of Medical Science, University of Groningen / University Medical Center Groningen. This work was also conducted in the framework of the LabEx CeLyA ("Centre Lyonnais d'Acoustique", ANR-10-LABX-0060/ANR-11-IDEX-0007) operated by the French National Research Agency