Effects of dynamic-range compression on the spatial attributes of sounds 
Ian M. Wiggins and Bernhard U. Seeber 
MRC Institute of Hearing Research, Nottingham, NG7 2RD, UK  

Compressive amplification in bilateral hearing aid fittings can disrupt binaural cues important for spatial 
hearing. In a simple acoustic environment, naturally occurring interaural time and level differences  
(ITDs and ILDs) are consistent with one another and indicate the true source azimuth. Independent 
dynamic-range compression at each ear acts to reduce ILDs, such that they suggest a more central, 
conflicting, direction than ITDs. The reduction in ILDs is a dynamic effect that depends on the 
characteristics of the compression and the sound.
A semantic differential based procedure was used to assess the effects of compression on the spatial 
attributes of sounds. Normally hearing listeners were asked a set of questions addressing issues of 
diffuseness, movement, image splits, and externalisation. Listeners responded on a seven-point scale 
between two relevant anchoring terms. For example, the question ‘Is the sound stationary or does it 
move?’ was answered on a scale between ‘Stationary’ and ‘Moving’.
Sources at an azimuth of ±60° were simulated using head-related transfer functions. The signal at each 
ear was split into a low-frequency and high-frequency channel (2 kHz crossover). Three high-
frequency channel processing conditions were run: ‘unprocessed’, ‘fast-acting dynamic-range 
compression’, and ‘static ILD bias’ (constant, non-dynamic ILD bias equivalent to that evoked by 
compression).  The high-frequency channel was presented to listeners either in isolation or 
recombined with the unprocessed low-frequency channel. Stimuli included pink noise bursts with 
varied onset slopes and rates of amplitude modulation, and speech. 
Results show that compression can increase the perception of movement and/or diffuseness, lead to 
image splits, and result in less well externalised sounds. Comparison of the dynamic compression and 
static ILD bias conditions reveals that the effects are primarily related to dynamic changes in ILDs 
introduced by the compression, not just to the presence of conflicting binaural cues per se.  Effects 
were greater for high-pass sounds, but occurred for broadband sounds also. Sounds with slow onsets 
were affected more than those with fast onsets. These findings reflect greater salience of ILD as a 
localisation cue at high frequencies and for sounds with gradual onsets. Speech, particularly high-pass 
speech, showed some of the greatest effects, indicating the susceptibility of speech to these adverse 
effects of compression on spatial perception.

Acknowledgements 
This work was supported through the Intramural Programme of the Medical Research Council.