NOISE-REDUCTION WITH FM SYSTEMS IN FM EM MODE

Commercially available noise reduction hearing aids have failed to del iver improved speech recognition in noise over properly fit convention al devices under typical listening conditions (Fabry & Van Tasell, 199 0; Gabrielsson, Schenkman, & Hagerman, 1988; Tyler Q Kuk, 1989; Van Ta sell, Larsen, & Fabry, 1988). The primary factor related to this findi ng is that speech and noise are mixed at the input of a single microph one; spectral changes are applied equally to both speech and noise, an d the signal-to-noise ratio remains unchanged from unprocessed conditi ons. The purpose of this study was to evaluate a prototype frequency-m odulated (FM) auditory trainer that allowed use of a remote FM microph one and/or an ear level environmental microphone (EM). For each of fiv e subjects with moderate to severe sensorineural hearing loss, the fre quency response of the EM was configured either to match that of the F M response, or to provide a high-pass filter characteristic similar to a noise reduction hearing aid. Speech recognition threshold (SRT) tes ting in noise was measured for five test conditions: 1) FM only; 2) EM only with high-pass filter response (EM-HP); 3) EM only with ''standa rd'' response (EM-S) matched to FM; 4) FM/EM-HP combined mode; and 5) FM/EM-S mode. Results averaged across subjects indicated that SRTs for the FM only condition were 9 to 10 dB better than those for either EM only condition; data from the combined FM/EM-HP mode averaged 4 dB be tter than for FM/EM-S conditions.