Figure 4.

Functional block diagram of the MPEAK speech-processing strategy. Input signals from a microphone are analyzed to extract or estimate the parameters of a small number of acoustic features that are important for conveying information about speech. These parameters include voicing, the fundamental frequency (F0), and the frequencies and amplitudes of the first two formants (F1 and F2). In addition, the levels of signals in three higher-frequency bandpass filters (BPFs) are determined. As explained in the text, a subset of these parameters is selected for stimulation in each period. The signal levels are converted into appropriate current levels of electric stimulation. The digital data transmitted to the implant specify the currents to be delivered by the active electrodes. The resulting stimulation pattern comprises groups of four sequential pulses delivered at an overall rate dependent on the estimated F0 whenever the acoustic input signal is judged to contain voiced speech.