The effect of frequency allocation on phoneme recognition with the nucleus 22 cochlear implant

Phoneme recognition performance in patients implanted with the Nucleus 22 cochlear implant is affected by the frequency-to-electrode assignment.

Multiple electrodes in modern cochlear implants are intended to deliver frequency-specific information to different tonotopic locations along the cochlea. However, the relation between the electrode locations, distribution of frequency information, and performance has not been explored thoroughly.

Ten listeners were tested on vowel and consonant identification tasks immediately after receiving each of the 15 speech processors. Experimental processors were created with 4, 7, and 20 activated electrodes. Five different frequency allocations were tested with all electrode conditions.

For 7- and 20-electrode maps, best vowel recognition performance was obtained with frequency tables 7 and 9, with subjects showing best performance with the table with which they were most familiar. With 4-electrode maps, no change in vowel recognition performance was observed as a function of the frequency allocation. Consonant scores showed only a small effect of frequency allocation across all processors. Results were similar across listeners with different electrode insertion depths.

The allocation of frequency ranges to electrodes in the Nucleus-22 cochlear implant can affect vowel recognition, when more than four electrodes are used, but is less important for consonant recognition. The allocation of frequency ranges to electrodes is an important factor in multichannel implants with more than four electrodes. The similarity of results across implant listeners with different electrode insertion depths implies that the optimal frequency allocation is one that best matches the allocation to which they've become accustomed, rather than one that matches the original tonotopic location of the electrodes.