Multi-peak speech procession
US5095904A · kind A · utility
Assignees
Inventors
Key dates
| Filing date | Sep 4, 1990 |
| Grant date | Mar 17, 1992 |
| Priority date | — |
| Expiry date | Sep 4, 2010 |
Classification
- Technology area (CPC A)Human Necessities
- CPC primaryA61N1/36038
- WIPO fieldMedical technology
- WIPO sectorInstruments
Abstract
An improved pulsatile system for a cochlear prosthesis is disclosed. The system employs a multi-spectral peak coding strategy to extract a number, for example five, of spectral peaks from an incoming acoustic signal received by a microphone. It encodes this information into sequential pulses that are sent to selected electrodes of a cochlear implant. The first formant (F1) spectral peak (280-1000 Hz) and the second formant (F2) spectral peak (800-4000 Hz) are encoded and presented to apical and basal electrodes, respectively. F1 and F2 electrode selection follows the tonotopic organization of the cochlea. High-frequency spectral information is sent to more basal electrodes and low-frequency spectral information is sent to more apical electrodes. Spectral energy in the regions of 2000-2800 Hz, 2800-4000 Hz, and above 4000 Hz is encoded and presented to three fixed electrodes. The fundamental or voicing frequency (F0) determines the pulse rate of the stimulation during voiced periods and a pseudo-random aperiodic rate determines the pulse rate of stimulation during unvoiced periods. The amplitude of the acoustic signal in the five bands determines the stimulus intensity.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.