The Perception of Contrastive Stress in Vocoded Speech: Implications for Cochlear Implant Users

Abstract

Vocoded speech is used frequently to simulate cochlear implant processing in normalhearing listeners (Loebach, 2007; Oxenham & Kreft, 2014). Speech signals are band-pass filtered at the first stage of processing and outputs of each of these filters are subsequently lowpass filtered. Varying the values of low-pass filter cutoffs influences periodicity cues conveyed through vocoded speech, with higher values passing more periodicity cues and lower levels reducing these cues. We examined the effect that limiting periodicity cues have in the perception of contrastive stress. Previous research showed that both spectral and temporal features of sentence stimuli are varied when speakers use contrastive stress (Cooper, Eady, & Mueller, 1985). We were primarily interested in how variations in spectral, temporal, and intensity features of vocoded speech influenced the perception of contrastive stress. We presented stimuli to participants in four conditions of a contrastive stress experimental test: unprocessed, natural speech and three processed speech conditions with low-pass filter cutoffs at 50 Hz, 160 Hz, and 250 Hz. Participants were assigned to only one filter condition. We also evaluated how talker and syntactic place influenced the perception of contrastive stress. There were four talkers and four syntactic stress conditions. All participants received were sentence stimuli for each talker and syntactic stress condition. We found significant perception effects for filter condition, talker, and place, as well as a significant interaction of talker by place. These outcomes suggest that the perception of contrastive stress in vocoded speech is influenced by a number of factors, including: (1) the availability of periodicity cues primarily conveyed by fundamental frequency, (2) talker variation, and (3) syntactic place of the stressed word in a sentence. The results may partially explain the range of performance experienced by cochlear implant users in tasks that test the perception of suprasegmental cues.