Elect. Eng. Dept., Eng. IV, 68-113, UCLA, Los Angeles, CA 91403-1594
Sigfrid D. Soli
House Ear Inst., Los Angeles, CA 90057
UCLA, Los Angeles, CA 91403-1594
Speech communication in environments with low signa/noise ratios (SNRs) is a primary complaint of the hearing impaired. Microphone beam formation techniques provide an effective approach to improving SNR in these environments. A novel, fixed microphone array is being developed with user-controlled mainlobe spatial look direction and attenuation band(s), and with a flat frequency response over the speech bandwidth. The array of R microphones and L taps per microphone maximizes energy concentration over a spatial look region and frequency band, subject to spatial and frequency constraints. Constrained maximization of w*Aw/w*Bw is required, where A and B are matrices specifying spatial and frequency factors, and w is the RL dimensional weight vector. The constraining subspace is specified by the array values, derivative values, and spatial directional constraints; w is obtained as the solution of a tractable unconstrained full-rank lower dimensional generalized eigenvalue problem. Numerical and simulation results for different values of R and L and for different bandwidths will be reported, as well as results of preliminary listening tests with normally hearing and hearing impaired individuals. The feasibility of real-time acoustic beamformers with arrays for hearing aids, and the advantages of this scheme over conventional adaptive schemes will also be discussed.