Effects of talker variability on vowel recognition in cochlear implants

Area of Science:

• Auditory Neuroscience
• Speech Perception
• Hearing Technology

Background:

• Cochlear implants (CI) aim to restore hearing but often result in challenges with speech understanding, particularly in noisy or multi-talker situations.
• Talker variability, differences in voice characteristics between speakers, can further degrade speech recognition for individuals with hearing impairments.
• Understanding how talker variability impacts CI users is crucial for improving speech processor design and rehabilitation strategies.

Purpose of the Study:

• To examine the impact of talker variability on vowel recognition in cochlear implant (CI) users.
• To compare CI user performance with normal-hearing (NH) participants using simulated CI processing.

Main Methods:

• CI users were tested with their clinical speech processors.
• Normal-hearing participants listened to 4-channel acoustic CI simulations with varying carrier types (noise band, sine wave) and temporal envelope cutoff frequencies (160 Hz, 20 Hz).
• Vowel recognition was assessed in single-talker and multi-talker contexts.

Main Results:

• CI users demonstrated significantly poorer vowel recognition in multi-talker contexts compared to single-talker contexts.
• For NH participants, talker variability did not significantly impact performance with noise-band carriers.
• Sine-wave carriers in simulations led to significant effects of talker variability on NH performance under both envelope filter conditions.

Conclusions:

• Spectral and temporal cues likely contribute to talker variability effects, especially with sine-wave carriers, due to incomplete preservation of fundamental frequency.
• These findings suggest that both spectral and temporal processing limitations in CI devices may exacerbate the effects of talker variability.
• Further research into optimizing CI signal processing for talker variability is warranted to enhance speech intelligibility.