An Active Headrest

1
An Active Headrest

• M Jones, Prof S J Elliott
• ISVR, University of Southampton

An active headrest has been designed that
reproduces audio in one seat, while cancelling
the resulting soundfield in an adjacent seat,
such that each person seated may experience
personal audio.
In order to achieve personal audio for a variety
of subjects, the concept of acoustic contrast is
used where the ratio of the mean square pressures
at a number of microphones is maximised to ensure
a large enough zone of silence.

• Best performance when secondary source as close
  as possible to primary
• Solution is almost equivalent to cancelling the
  pressure at p6
• 20 to 25 dB of cancellation at some frequencies
• High frequency contrast naturally high due to
  absorption and beaming of loudspeakers

Picture of a pair of airline seats with the
active headrest installed
TOP Diagram of headrest and microphones used in
contrast calculation BOTTOM Results obtained
compared to passive contrast
To allow improvements to be made to the system
and allow the effect of changes in the mechanical
arrangement to be predicted, a BEM model of the
headrest needs to be developed.
The modelling work begins simply and gradually
increases in complexity. A number of models have
been made and compared to measured results
Mesh of KEF loudspeaker source impinging on a
Kemar dummy head

• The pressure on the surface of a sphere due to a
  plane wave, where modelled results are compared
  to analytical results
• The pressure at the ears of a dummy head due to
  an impinging far field point source (HRTF
  modelling)
• The pressure at the ears of a dummy head due to a
  simple closed loudspeaker in the near field,
  compared to measured results
• The passive and active contrast is predicted from
  simulations and compared to the measured
  performance

TOP Kemar mesh for HRTF model, with spherical
point source impinging BOTTOM Results compared
to free field equalised MIT Kemar data set, 30
TOP Mesh of KEF loudspeaker source impinging on
a dummy head BOTTOM Modelled results compared to
measured anechoic results
TOP Model of full seat used in BEM BOTTOM
Modelled contrast compared to measured anechoic
contrast
Both monopole and BEM simulations are carried out
to predict the performance of alternative
scenarios and geometries. One alternative is to
have multiple dark zones present, e.g. an
additional dark zone in a seat in front of the
bright zone.
A bright zone for a mobile phone user is achieved
by defining the dark zone at all angles except
those corresponding to the head position. The
sound power minimisation approach is shown to
achieve around 12dB of contrast and decrease the
sound power output by around 12dB compared to a
monopole generating the same reproduced levels.
LEFT Contour plot of the soundfield calculated
from monopole simulations after control with two
sources in the optimal positions for control in
both dark zones RIGHT Soundfield after control
predicted using BEM model
Sound power minimisation has been shown to
provide a suitable alternative to contrast
maximisation when there are multiple dark zones,
such that as the number of dark zones becomes
large the two methods result in increasingly
similar solutions.
LEFT Soundfield after control for a bright zone
0.5m from the mobile phone speaker RIGHT The
predicted active and passive contrast for this
case
Future Work

• Continue new investigations into new geometries,
  and take realtime measurements from a real seat
  to verify predictions
• Include subjective results on the active control
  performance, and look into the feasibility of a
  headrest for virtual audio reproduction