Michael A Akeroyd

1
Two methods for synthesizing distant sounds in
the laboratory
Michael A Akeroyd Stuart Gatehouse David
McShefferty Julia Blaschke
maa_at_ihr.gla.ac.uk www.ihr.mrc.ac.uk
MRC Institute of Hearing Research (Scottish
Section), Royal Infirmary, Glasgow.
Co-funded by the MRC and by the Chief Scientist
Office of the Scottish Executive Health Dept.
2

• Our questionnaires with elderly
  hearing-impaired patients
• suggest that self-reported difficulties in
  distance perception
• are a component to auditory disability and
  handicap.

3

• Our questionnaires with elderly
  hearing-impaired patients
• suggest that self-reported difficulties in
  distance perception
• are a component to auditory disability and
  handicap.
• We wanted to study experimentally distance
  perception in
• hearing-impaired listeners.

4

• Our questionnaires with elderly
  hearing-impaired patients
• suggest that self-reported difficulties in
  distance perception
• are a component to auditory disability and
  handicap.
• We wanted to study experimentally distance
  perception in
• hearing-impaired listeners.
• But to test distance in the laboratory, one
  needs a big room
• or a synthetic method for recreating it in a
  smaller room.

5
Method 1 room-image synthesis
Poster presentation Akeroyd MA, Blaschke J and
Gatehouse S (2005). Perception of the cues that
can signal distance in elderly hearing-impaired
individuals, BSA Short-Papers Meeting,
Cardiff. Journal publication Akeroyd MA,
Gatehouse S, and Blaschke J (2007). The
detection of differences in the cues to distance
by elderly hearing-impaired listeners, J.
Acoust. Soc. Am., 121, 1077-1089.
6
4 m
24 loudspeakers 1-m radius 15º separation.
T60 lt 100 ms
24-channel digital interface
2.5 m
7
Thanks to Jay Desloge and Barbara
Shinn-Cunningham (Boston University)
15 ms -14 dB
5 m
As distance 5 m and the speed-of-sound is 345
ms-1, then travel time
15ms
inverse-square level-reduction 10log10 (52)
14 dB.
8
18 ms -16 - 3 dB
6.5 m
3.2 m
3.2 m
distance to 1st reflection
6.5 m, therefore travel time
18 ms
inverse-square level-reduction 10log10 (52)
16 dB. plus the absorbance by the wall
3 dB (arbitrary!)
9
20 ms -17 - 3 dB
10
21 ms -17 - 3 dB
11
22 ms -18 - 3 dB
12
23 ms -18 - 6 dB
Reflection off 2 walls absorbance 2 x 3 6 dB
13
26 ms -19 - 6 dB
14
27 ms -19 - 6 dB
15
28 ms -20 - 6 dB
16
-120º
-135º
-105º
-150º
-90º
-165º
-75º
180º
-60º
-45º
165º
from Akeroyd MA, Gatehouse S, and Blaschke J
(2007). The detection of differences in the cues
to distance by elderly hearing-impaired
listeners, J. Acoust. Soc. Am., 121, 1077-1089.
-30º
150º
-15º
135º
120º
0º
15º
105º
90º
30º
75º
45º
60º
17
from Akeroyd MA, Gatehouse S, and Blaschke J
(2007). The detection of differences in the cues
to distance by elderly hearing-impaired
listeners, J. Acoust. Soc. Am., 121, 1077-1089.
Impulse response measured in ring to a click _at_
a real 1 m
Impulse response measured in ring to a click _at_
a virtual 2 or 5 m
18
Virtual Room
Glasgow Lab Room
2.5 m
2 m
2 m
7 m
5 m
5 m
4 m
8 m
a 0.5
8 m
9 m
As training for a distance experiment, 80 elderly
listeners (some normal, some hearing-impaired)
were asked to estimate the perceptual distance of
short sentences with virtual distances of 2, 5
and 8 m
19
Virtual Room
Glasgow Lab Room
2.5 m
2 m
2 m
7 m
5 m
5 m
4 m
8 m
a 0.5
8 m
9 m
As training for a distance experiment, 80 elderly
listeners (some normal, some hearing-impaired)
were asked to estimate the perceptual distance of
short sentences with virtual distances of 2, 5
and 8 m

• target 2 m mean 1.5 m s 1.0 m N 68
• target 5 m mean 3.2 m s 1.7 m N 70
• target 8 m mean 4.1 m s 2.7 m N
  61.

20
Virtual Room
Glasgow Lab Room
2.5 m
2 m
2 m
7 m
5 m
5 m
4 m
8 m
a 0.5
8 m
9 m
As training for a distance experiment, 80 elderly
listeners (some normal, some hearing-impaired)
were asked to estimate the perceptual distance of
short sentences with virtual distances of 2, 5
and 8 m

• target 2 m mean 1.5 m s 1.0 m N 68
• target 5 m mean 3.2 m s 1.7 m N 70
• target 8 m mean 4.1 m s 2.7 m N
  61.
• 81 of listeners reported 5-m condition as being
  further than the loudspeakers
• and 24 of listeners reported 5-m condition as
  outside the lab room.

21

• Distance-discrimination experiment
• Two-interval, two-alternative forced choice.
• Each trial had two sentences
• A reference at 5 m, the other at 2,
  3, 4, 5, 6, 7, 8 m
  (D 1 m)
• B reference at 2 m, the other at 1.00,
  1.33, 1.67, 2.00, 2.33, 2.67, 3.00 m (D 0.33
  m)
• Task was to decide which one was further
  away?
• One sentence spoken by a man, the other by a
  woman.

22

• Distance-discrimination experiment
• Two-interval, two-alternative forced choice.
• Each trial had two sentences
• A reference at 5 m, the other at 2,
  3, 4, 5, 6, 7, 8 m
  (D 1 m)
• B reference at 2 m, the other at 1.00,
  1.33, 1.67, 2.00, 2.33, 2.67, 3.00 m (D 0.33
  m)
• Task was to decide which one was further
  away?
• One sentence spoken by a man, the other by a
  woman.

from Akeroyd MA, Gatehouse S, and Blaschke J
(2007). The detection of differences in the cues
to distance by elderly hearing-impaired
listeners, J. Acoust. Soc. Am., 121, 1077-1089.

• 77 listeners
• age 22 83 years
• hearing-losses -5 to 59 dB.

23
2-Closer 2-Further
5-Closer 5-Further
from Akeroyd MA, Gatehouse S, and Blaschke J
(2007). The detection of differences in the cues
to distance by elderly hearing-impaired
listeners, J. Acoust. Soc. Am., 121, 1077-1089.
24

• Advantages of room-image method
• Seems to work ok
• Relatively simple to implement
• Easy to get data on each echo for presenting
  through the 24-speaker ring
• Disadvantages
• Limited to rooms shaped like rectangular boxes
• Specular reflection only no scattering etc
  (how accurate is it?)

25
Method 2 Odeon synthesis (Odeon is Bruel and
Kjaers acoustic design program)
Poster presentation McShefferty D and Akeroyd MA
(2006) Reproducing the acoustics of the IHR
Nottingham seminar room using a computational
model allied to a loudspeaker array , BSA Short
Papers Meeting, Cambridge
26

• Bruel and Kjaers Odeon program
• Simulates acoustics of intricate rooms.
• Incorporates sophisticated acoustic
• effects, such as scattering and
• directional loudspeakers.
• Has a surround-sound option.

Two views of the full Odeon model of the IHR
Seminar Room.
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
window
window
window
window
Viewpoint
lectern
1-m
Listener
2-m
3-m
4-m
5-m
6-m
7-m
8-m
9-m
4.4 m
door
door
door
8 m
5 m
7 m
3.0 m
3.0 m
2 m
13 m
New seminar room at IHR, Nottingham
31
8 m
8 m
5 m
5 m
2 m
2 m
New seminar room at IHR, Nottingham
32
Direct sound
Clear(ish) echoes
50 dB
300 ms
------ 2-m distance ------
absorption of surfaces 0.2 (20) scatter
coefficient 0.5.
33
------ 5-m distance ------
absorption of surfaces 0.2 (20) scatter
coefficient 0.5.
34
------ 8-m distance ------
absorption of surfaces 0.2 (20) scatter
coefficient 0.5.
35

• Advantages of room-image method
• Seems to work ok
• Relatively simple to implement
• Easy to get data on each echo for presenting
  through the 24-loudspeaker ring
• Disadvantages
• Limited to rooms shaped like rectangular boxes
• Specular reflection only no scattering etc
  (how accurate is it?)

36

• Advantages of room-image method
• Seems to work ok
• Relatively simple to implement
• Easy to get data on each echo for presenting
  through the 24-loudspeaker ring
• Disadvantages
• Limited to rooms shaped like rectangular boxes
• Specular reflection only no scattering etc
  (how accurate is it?)

• Some questions about the Odeon method
• How accurate can we get the acoustics? How
  accurate should we get the acoustics?
• How do we set the surface absorbencies and
  scattering-coefficients?
• On listening to it, it seems too reverberant
  why?

37

• Advantages of room-image method
• Seems to work ok
• Relatively simple to implement
• Easy to get data on each echo for presenting
  through the 24-loudspeaker ring
• Disadvantages
• Limited to rooms shaped like rectangular boxes
• Specular reflection only no scattering etc
  (how accurate is it?)

• Some questions about the Odeon method
• How accurate can we get the acoustics? How
  accurate should we get the acoustics?
• How do we set the surface absorbencies and
  scattering-coefficients?
• On listening to it, it seems too reverberant
  why?