*4.1.2 Study description: exp-parseihian*

Parseihian and Katz [8], a 2012 study on accommodation to non-individual HRTF. 12 participants trained on auditory localisation, each completing 3 sessions of 12 min each on 3 consecutive days. Each session consisted of an interactive audio localisation game followed by a localisation task evaluation testing participants on 25 positions distributed on a sphere, using a 180 ms sequence of white noise bursts as stimulus. Before training, each participant ranked a set of 7 *perceptually orthogonal* HRTFs [40, 41] from the LISTEN database [42] based on localisation accuracy as perceived during predefined audio trajectories. The best and worst-match HRTF for each participant was extracted from this ranking. Participants were then divided into 3 groups: 2 that trained with their individual HRTF (**grp-parse-indiv**), 5 with the best-match HRTF (**grp-parse-best**), and 5 with the worst-match HRTF (**grpparse-worst**). An additional 2 groups that performed only 1 training session are not considered in the current analysis. The ITDs of all HRTFs were adjusted based on individual participant head circumference, using a model derived from a regression between measured ITDs and morphological parameters. This technique is used as a

practical method, easily carried out by end-users, to maximise initial localisation performance accuracy.
