Comparative speech recognition results in eight subjects using two different coding strategies with the Nucleus 22...

This study was designed to compare the results of several speech tests administered to eight subjects who used two types of speech-coding strategies with the Nucleus 22 channel cochlear implant. The subjects had an average experience of 40 months with the F0F1F2 coding strategy implemented in their previous wearable speech processor. Three subjects were good performers (showing significant open-set understanding without lip-reading) and five were moderate performers (not able to do speech-tracking by auditory means alone). All subjects were evaluated again after 1 month and 6 months of experience with the MULTIPEAK coding strategy of the miniature speech processor. The test materials included vowel and consonant identification, monosyllabic words, everyday sentences and numbers in noise. All eight subjects showed an improvement on more than three of five measures. The group of moderate performers showed a larger improvement in vowel (+16%) and consonant (+17%) identification scores than the group of good performers. For the open-set sentence test, the better patients were able to increase their score from 52% to 80% correct; two of the moderate performers did not improve. Six subjects achieved significantly higher scores at moderate signal-to-noise ratios (up to 10 dB S/N) in the (Freiburger) number test. Results of information transmission analysis are also discussed.