Redesign of the Simulated Open Field Environment and it's Application in Audiological Research Bernhard Seeber and Ervin Hafter The Simulated Open Field Environment (SOFE) creates realistic auditory scenes with simulated sources and echos inside an anechoic chamber. By playing sources and echoes from 48 horizontally arranged loudspeakers the directional characteristics of the simulated reverberant environment can be naturally maintained throughout head movements. Moreover, in comparison with virtual environments based on headphone presentation the subject in the SOFE is unencumbered by headphones and listens in the free-field through their own, "perfect" head-related transfer functions. This allows us to directly compare the results from normal hearing subjects against results from subjects with a hearing aid or a cochlear implant. The software of the SOFE is based on Matlab and allows for parallel playback of sound on 48 channels. The frequency response of the loudspeakers is equalized to +/- 0.7 dB within 0.3-10 kHz whereas the phase is correct to a few usec. Echoes can be computed for sources in rectangular rooms with different reflection characteristics of the walls using the mirror-image source method. Synchronized to the sound is a high-resolution video display (1280*768 pixels). Using a trackball visual objects can be moved in a simulated 3-dimensional room to yield the subject's estimates of azimuth, elevation and distance of auditory sources. We currently investigate near speaker panning to give higher horizontal resolution. Preliminary results show that direct comparison of panned vs. direct sources leads to highly detectable differences (d'=2.14). Loudness and timbral cues are one reason for the high detectability. The applied sin/cosine panning law is frequency independent, leading at the fixed head position to energy summation at low frequencies and intensity addition at higher frequencies. Using an HRTF-based model of panning correction filters can be computed to reduce the perceptual impact of these cues. Preliminary tests show that the detectability of panned vs. direct sources can be reduced by applying these filters (d'=1.2). Further cues are changes in perceived source width and source location.