Effect of Noise Reduction Gain Errors on Simulated Cochlear Implant Speech Intelligibility

Area of Science:

• Auditory Neuroscience
• Signal Processing
• Speech Perception

Background:

• Cochlear implant (CI) users often struggle with speech intelligibility in noisy environments.
• Two key factors for speech intelligibility are preserving low-frequency amplitude modulations and speech transients.
• The relative importance of these factors for noise reduction benefit in CI users is debated.

Purpose of the Study:

• To investigate the impact of noise reduction strategies on speech intelligibility in CI users.
• To determine the relative importance of preserving speech transients versus minimizing noise in speech gaps.
• To evaluate the effectiveness of noise reduction in stationary and modulated noise conditions.

Main Methods:

• A cochlear implant (CI) simulation was used with normal-hearing listeners.
• Noise estimation errors were systematically introduced in speech segments, gaps, and transitions.
• Speech intelligibility was measured in stationary and modulated noise.

Main Results:

• Minimizing noise in speech gaps improved intelligibility, particularly in modulated noise.
• Significantly greater intelligibility improvements were achieved when both gaps were cleared of noise and speech transients were preserved.
• Noise reduction algorithms benefit from accurate identification of speech segment and gap boundaries.

Conclusions:

• Preserving speech transients and reducing noise in speech gaps are both crucial for enhancing CI speech intelligibility.
• Accurate boundary detection is vital for noise reduction algorithms to optimize both gap noise reduction and transient preservation.
• These findings have implications for developing more effective noise reduction strategies for cochlear implant users.