Assessment of High-Variability Speech Recognition in Adult Cochlear Implant Users using PRESTO

Area of Science:

• Audiology
• Neuroscience
• Speech and Hearing Sciences

Background:

• Speech recognition assessment in adult cochlear implant (CI) users often uses low talker variability materials.
• Limited understanding exists regarding the impact of high talker variability on CI user speech recognition and individual differences.
• Clinical utility of high talker variability sentence materials for CI users remains underexplored.

Purpose of the Study:

• To evaluate the effect of talker variability on sentence recognition in adult CI users.
• To investigate the relationship between working memory capacity and speech recognition in high-variability conditions.
• To assess adult CI users using the Perceptually Robust English Sentence Test Open-Set (PRESTO) with varying talker variability.

Main Methods:

• Adult CI users and normal-hearing (NH) adults under CI simulation completed speech recognition tests.
• Tests included Hearing in Noise Test (HINT; low variability), AzBio (moderate variability), and PRESTO (high variability) sentences.
• Testing occurred in quiet and multitalker babble (MTB); working memory capacity was assessed in CI users.

Main Results:

• Speech recognition performance decreased with increased talker variability in both CI and NH groups.
• PRESTO sentences showed significantly lower performance in multitalker babble (MTB) compared to HINT and AzBio.
• Working memory capacity was correlated with sentence recognition accuracy in CI users across all conditions.

Conclusions:

• High talker variability, particularly in PRESTO materials, detrimentally affects speech recognition in CI users and simulated NH listeners, especially with background noise.
• Working memory capacity is a significant factor in the speech recognition abilities of adult CI users.
• High-variability, multitalker sentence tests like PRESTO offer a robust clinical and research tool for assessing speech recognition and individual differences in CI users without ceiling effects.