Weighting of binaural cues when lateralizing sounds subjected to dynamic-range compression I.M. Wiggins and B.U. Seeber, MRC Institute of Hearing Research, Nottingham, UK When hearing devices worn on each ear operate independently of one another, the binaural cues that are crucial to spatial hearing are not always preserved. Dynamic-range compression, a common feature in modern hearing devices, alters interaural level differences (ILDs) when applied bilaterally. As ILDs are one of the primary cues for lateral position, compression may thus affect the perceived lateralization of sounds. Compression does not directly affect interaural time differences (ITDs), and so, after compression, ILDs and ITDs can suggest conflicting directions. This study investigated the potential consequences of compression on the perceived position of sounds using normal-hearing listeners. Virtual-acoustic stimuli were presented over headphones. A source was simulated at either - 60° or +60° azimuth. The signal at each ear was split into low- and high-frequency channels (2 kHz crossover). Three high-frequency-channel processing conditions were run: ‘unprocessed’, ‘fast-acting 3:1 dynamic-range compression’, and ‘static ILD bias’ (static reduction in the long-term ILD by a factor of three). The high-frequency channel was presented either in isolation or recombined with the unprocessed low-frequency channel. Stimuli included pink-noise bursts with varied onset/offset slopes and types of envelope modulation, and speech. In a lateralization task, listeners indicated the leftmost and rightmost extents of the sound image(s) and selected from three response options according to whether they heard (1) a single, stationary image, (2) a moving image, or (3) a split image. When listening to the high-frequency channel in isolation, both dynamic compression and the static ILD bias significantly shifted the perceived lateral position towards the centre of the head. The static ILD bias generally did not affect the basic nature of the sound image, with a ‘single, stationary image’ being reported on most trials. The effect of dynamic compression was similar for stimuli with relatively abrupt onsets and offsets. In contrast, for stimuli containing more gradual onsets and offsets, including speech, dynamic compression increased the occurrence of moving and split images by up to 57 percentage points, and this was typically accompanied by a wide separation of the leftmost and rightmost extents of the lateralized image(s). The effects of the processing were similar in nature but reduced in magnitude when undisturbed low-frequency information was made available to listeners. The results were analysed in the context of binaural-cue weighting. For high-pass sounds, listeners gave on average roughly equal weight to high-frequency ILDs and envelope ITDs if the ILD bias was constant throughout the sound. When compression caused ILDs to change dynamically at rates low enough to be followed in detail (approximately 2–7 Hz), listeners generally gave greater weight to ILDs. However, some listeners seem to have been relatively insensitive to dynamically changing ILDs. Acknowledgements This work was supported through the Intramural Programme of the Medical Research Council.