Enhancing temporal cues to voice pitch in continuous interleaved sampling cochlear implants

The limited spectral resolution of cochlear implant systems means that voice pitch perception depends on weak temporal envelope cues. Enhancement of such cues was investigated in implant users and in acoustic simulations. Subjects labeled the pitch movement of processed synthetic diphthongal glides. In standard processing, noise carriers (simulations) or pulse trains (implant users) were modulated by 400 Hz low-pass envelopes. In modified processing, carriers were modulated by two components: (1) Slow-rate (<32 Hz) envelope modulations, conveying dynamic spectral shape changes crucial for speech; (2) a simplified waveform (e.g., a sawtooth) matching the periodicity of the input diphthong. In both normal listeners and implant users performance was better with modified processing, though temporal envelope cues were less effective with higher F0. Factors contributing to the advantage for modified processing may include increased modulation depth and use of a modulation waveform featuring a rapid onset in each period, resulting in a clearer representation of F0 in the neural firing pattern. Eliminating slow-rate spectral dynamics, so that within-channel amplitude changes solely reflected F0, showed that dynamic spectral variation obscured temporal pitch cues. Though significant, advantages for modified processing were small, suggesting that the potential for developing strategies delivering enhanced pitch perception is limited. (C) 2004 Acoustical Society of America.