Electrodographic analysis and field evaluation of the Speak coding strategy

Area of Science:

• Audiology
• Biomedical Engineering
• Speech Processing

Background:

• Cochlear implants enhance hearing for individuals with severe to profound hearing loss.
• Speech-coding strategies are crucial for optimizing auditory information delivered by cochlear implants.
• The Nucleus Minisystem-22 utilizes the Speak strategy, analyzing speech with band-pass filters and spectral maxima.

Purpose of the Study:

• To evaluate the performance of the Speak speech-coding strategy in quiet and noisy conditions.
• To compare the Speak strategy against the Multipeak (Mpeak) speech-processing strategy.
• To assess auditory perception in cochlear implant users using various speech tasks.

Main Methods:

• Reconstruction of stimulation patterns as electrodograms from radio frequency signals.
• Comparison of electrodograms generated by Speak and Mpeak strategies with input signal spectrograms.
• Field trial study with 20 German-speaking cochlear implant users over 12 weeks using an ABAB paradigm.

Main Results:

• Significant differences in group mean scores for most speech recognition subtests between Speak and Mpeak strategies.
• Largest overall improvements observed for sentence tests in noisy conditions with the Speak strategy.
• Auditory perceptual tasks included vowel, consonant, monosyllable word, and sentence recognition.

Conclusions:

• The Speak speech-coding strategy demonstrates superior performance compared to the Mpeak strategy for cochlear implant users.
• Speech recognition, particularly in noise, is significantly enhanced by the Speak strategy.
• Further research can refine cochlear implant speech processing for improved real-world auditory performance.