Single-chip digital signal processors (DSPs) allow the flexible implementation of a large variety of speech analysis, synthesis, and processing algorithms for the hearing impaired. A series of experiments was carried out to optimize parameters of the adaptive beamformer noise reduction algorithm and to evaluate its performance in realistic environments with normal-hearing and hearing-impaired subjects. An experimental DSP system has been used to implement a multiband loudness correction (MLC) algorithm for a digital hearing aid. Speech tests in quiet and noise with 13 users of conventional hearing aids demonstrated significant improvements in discrimination scores with the MLC algorithm. Various speech coding strategies for cochlear implants were implemented in real time on a DSP laboratory speech processor. Improved speech discrimination performance was achieved with high-rate stimulation. Hybrid strategies incorporating speech feature detectors and complex decision algorithms are currently being investigated.