Speech Perception With Combined Electric-Acoustic Stimulation: A Simulation and Model Comparison

Area of Science:

• Audiology
• Speech Science
• Biomedical Engineering

Background:

• Electric-acoustic stimulation (EAS) combines electrical stimulation of the cochlea for high frequencies with acoustic stimulation for low frequencies.
• Understanding the benefits of EAS is crucial for improving speech perception in individuals with hearing loss.
• Previous studies have investigated EAS user performance, but direct simulation in normal-hearing (NH) subjects offers a controlled environment for hypothesis testing.

Purpose of the Study:

• To simulate speech perception using EAS in NH subjects and verify the combined stimulation advantage.
• To compare simulated EAS results with previous findings from cochlear implant (CI) and EAS users.
• To develop an automatic speech recognition (ASR) system to analyze the impact of low-frequency information and test hypotheses about EAS benefits.

Main Methods:

• Speech perception was assessed using the Oldenburg Sentence Test in NH subjects (n=43) under simulated CI and EAS conditions.
• EAS was simulated using low-pass filtered speech (cutoff frequencies 200 and 500 Hz) and vocoded high-frequency information.
• An ASR system was employed with similar low-pass filtering (100-500 Hz) to model speech processing and signal-detection-theory (SDT) models were used for performance prediction.

Main Results:

• NH subjects demonstrated a significant speech perception advantage with simulated EAS, which improved with higher low-pass cutoff frequencies.
• Simulated EAS performance in NH subjects was comparable to CI and EAS users in continuous noise, but superior in modulated noise at 500 Hz cutoff.
• The ASR system mirrored NH subject performance, revealing synergistic effects for cutoff frequencies ≥300 Hz, and SDT models predicted continuous noise results but not modulated noise effects.

Conclusions:

• The simulation successfully demonstrated the EAS advantage in NH subjects, highlighting the 'gap listening' effect in modulated noise.
• NH subjects benefited from simulated EAS, particularly in modulated noise, unlike previously studied CI and EAS users.
• ASR systems are viable for evaluating signal processing strategies in simulated hearing loss, though SDT models require refinement for modulated noise scenarios.