Vocoder simulations of highly focused cochlear stimulation with limited dynamic range and discriminable steps

Area of Science:

• Auditory Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Recent advancements explore focused cochlear stimulation (e.g., tripolar electrical, infrared neural) to enhance speech and music perception.
• Standard monopolar stimulation is a broad electrical field application in cochlear implants.

Purpose of the Study:

• To investigate speech recognition using vocoder simulations comparing broad (monopolar) versus focused cochlear stimulation.
• To evaluate performance across varying signal-to-noise ratios, dynamic ranges, and loudness quantization steps.

Main Methods:

• Vocoder simulations modeled noise-vocoded speech presented to normal-hearing listeners.
• Nonuniform quantization steps simulated Weber functions typical of cochlear implant users.
• Intelligibility assessed for different filter slopes, SNRs, dynamic ranges, and number of discriminable steps.

Main Results:

• Speech processed with focused stimulation vocoders was substantially more intelligible (up to 60 percentage points) than with monopolar vocoders.
• No significant intelligibility differences were found between two focused stimulation approaches (10 and 17 dB/mm attenuation).
• Highly focused stimulation maintained high performance with a 5 dB dynamic range and ≥8 discriminable steps, but performance dropped below 8 steps.

Conclusions:

• Highly focused cochlear stimulation (tripolar electrical, infrared neural) offers potential for superior performance in noise over monopolar stimulation.
• Further research is needed to fully realize and leverage the strengths of these focused stimulation modalities.