From Spectral Resolution to Speech Perception: A Review of Findings in Postlingually Deafened Adult Cochlear Implant...

Reduced spectral resolution limits speech recognition in cochlear implant (CI) listeners. Although many studies have examined this association, uncertainties remain regarding its strength and contributing methodological and clinical factors. This narrative review synthesizes findings from studies of postlingually deafened adult CI listeners, focusing on psychophysical measures of spectral resolution and their strengths and limitations.

We reviewed studies published through January 2025 that examined the relationship between psychophysical measures of spectral resolution and speech perception outcomes in postlingually deafened adult CI listeners. Twenty-four studies met inclusion criteria and tested this association statistically. Where available, the coefficient of determination (R²) and effect size measures were extracted to quantify the strength of this association.

Several studies found a statistically significant association between psychophysical measures of spectral resolution and speech recognition performance. The predictive power of spectral resolution measures for speech perception outcomes varied substantially (R² = .21-.69), depending on the specific measure used and the type of speech material. Spectral ripple discrimination/detection threshold tasks and their modified versions, on average, show higher predictive capacity for investigating this link. These measures are especially predictive when using speech materials such as consonant-nucleus-consonant words and AzBio or Institute of Electrical and Electronics Engineering sentences presented in quiet or multitalker babble at moderate signal-to-noise ratios. In contrast, measures based on forward-masking paradigms and pitch ranking tend to yield weaker or nonsignificant associations.

The review underscores spectral resolution as a key leverage point in efforts to improve speech recognition outcomes. Variability in observed associations highlights the need for further mechanistic research into the pathways linking spectral resolution to speech recognition outcomes. Future studies are warranted to improve both the predictive accuracy and clinical feasibility of spectral resolution assessment tools.