This thesis suggests a small set of essential
parameters for a personalised
and effective dynamic binaural synthesis
with headphones. An imageguided
procedure with two 2D images
of the head guides the personalisation
of head-related transfer function
(HRTF), combining a spherical
head model with ear displacement
with the HRTF magnitude selected
from a database according to anthropometry.
Room acoustics phenomena
are simplified following the scattering
delay network (SDN) approach
which allows an accurate spatialisation
of first order reflections. The proposed
model is compared to the common
higher-order ambisonics (HOA)
rendering. Improvements in localisation
and externalisation performances
within a virtual reality listening experiment
attest the benefits of HRTF personalisation
compared to the use of
generic HRTFs, and indicates the success
of the proposed customised SDN
model once compared to HOA.