Intonational cues for speech perception in noise by cochlear implant listeners

The objectives of this study are to evaluate cochlear implant (CI) listeners' ability to infer low frequency (LF) pitch information from temporal fine structure (TFS) cues and to gain insight into its effects on speech perception, especially in the presence of a fluctuating background noise. Pitch perception assessment using linguistic stimuli is believed to better reflect the role of pitch in communicatively realistic situations.

The low-pass-filtered sentence intonation (SI-LPF) test based on linguistic stimuli marked by intonation changes is used to estimate a difference limen for discrimination of LF pitch changes in adult CI listeners (N = 17 ears). Speech perception in the presence of noise is assessed using the sentence test with adaptive randomized roving level (STARR), where everyday sentences are presented at low, medium, and high levels in a fluctuating background noise. SI-LPF correlations with STARR are compared to those with sentence recognition tests presented in quiet (SRQ) and in noise, using fixed signal-to-noise ratio (SNRs at +10 and +5 dB).

SI-LPF findings show significant positive correlations with STARR performance (r_s = 0.63, p = 0.007), whilst the associations with SRQ (r_s = - 0.37, p = 0.149), SNR + 10 (r_s = - 0.24, p = 0.345), and SNR + 5 (r_s = - 0.14, p = 0.587) are not statistically significant.

Present findings reflecting considerably stronger correlations than previous studies using non-linguistic stimuli, in particular for speech perception with roving-level adaptive test method (STARR) highlight the effects of LF pitch perception and TFS sensitivity on challenging everyday situations, where CI users listen to speakers with varying levels in a fluctuating background.