Effect of Vocoder Type to Mandarin Speech Recognition in Cochlear Implant Simulation

Vocoder simulation is an effective tool in predicting the general pattern of speech recognition performance in cochlear implant users. The present study investigated the effect of vocoder type (i. e., noise-vocoder and tone-vocoder) to Mandarin speech recognition in vocoder simulation. Mandarin speech materials (sentences and single vowels) were processed by tone-vocoder and noise-vocoder, and presented to normal-hearing listeners to recognize. Experimental results showed that Mandarin sentences synthesized with a tone vocoder were more intelligible than those synthesized with a noise vocoder; however, this intelligibility advantage was not found in the lexical tone identification task. Consistent with previous outcomes, the present work showed that tone-vocoder was able to yield a better sentence recognition performance. It is still arguable that this improved recognition performance of tonevocoded sentences could be attributed to the better lexical tone identification in tone-vocoder based study.