Several studies have demonstrated that telephone use presents a challenge for most cochlear implant (CI) users, and this is attributed mainly to the narrow bandwidth (300–3400 Hz) introduced by the telephone network. The present study focuses on answering the question whether telephone speech recognition in noise can be improved by introducing, prior to vocoder processing, low-frequency harmonic information encompassing the missing (due to the telephone network) information residing in the 0–300 Hz band. Experiment 1 regenerates the main harmonics and adjacent partials within the 0–600 Hz range in corrupted (by steady noise) telephone speech which has been vocoded to simulate electric-acoustic stimulation (EAS). Results indicated that introducing the main harmonics alone did not produce any benefits in intelligibility. Substantial benefit (20%) was observed, however, when both main harmonics and adjacent partials were regenerated in the acoustic portion of EAS-vocoded telephone speech. A similar benefit was noted in Experiment 2 when low-frequency harmonic information was introduced prior to processing noise-corrupted telephone speech using an eight-channel vocoder. The gain in telephone speech intelligibility in noise obtained when low-frequency harmonic information was introduced can be attributed to the listeners having more reliable access to a combination of F0, glimpsing and lexical segmentation cues.
展开▼
