Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Surround & 3D
Sound Systems
Digital Audio Processing (20023)
Sound and Image in Telecommunication Engineering
Course 2015/2016
Sergio Bleda Prez
Department of Physics, Engineering Systems and Signal Theory
Introduction
Introduction
n
Spatial Perception
n
Why two?
Because our sound localization mechanism needs two
different signals
Using only one, localization is very very difficult
Duplex Theory
n
ITD
n
ITD
n
1.5 kHz
This frequency has a wavelength similar to the head size
But its much less relevant than ITD for lower frequencies
IID
n
Unless the sound source is in front of (or at the back of) the head
Left ear is in
shadow zone
IID
n
No shadow
due to diffraction
10
Duplex Theory
n
11
Duplex Theory
n
Example of ambiguity:
12
Cone of Confusion
n
Ambiguity
13
Improving localization
n
Head
Shoulders
Pinna
14
Improving localization
n
Sound
source
Sound
source
Improving localization
n
Improving localization
n
17
Additional factors:
Kind of sound
Length of the sound
n
Onset
n
Phantom Image
n
20
21
22
Haas Effect
n
Distance Perception
n
It depends on reverberation
We appreciate a drop of 6 dB
n
Distance Perception
n
25
Movement Perception
n
c
f a =
c + vr
f s
26
Distance + Movement
n
Motion Parallax:
n
n
When is far you didnt notice it (it has a constant low level)
When is near it seems like a plane (it produces large sound
level differences)
And, of course: doppler
27
Distance + Movement
n
28
29
Stereophonic Systems
Binaural Systems
Sound Field Reconstruction Systems
30
Monoaural Sound
n
31
Stereophonic Systems
32
Stereo
n
The movement is only based on IID (it does not use ITD)
Producing phantom sources
Stereo
n
Maximum aperture
must be 60
= o
Otherwise phantom
gets unstable
SWEET SPOT
Stereo Panning
n
Linear law
Sine law
Tangent law
35
Stereo Panning
n
Linear law:
This is the easiest law, but it is not advisable
Since it does not maintain the apparent level of the source
during the movement
Phantom Source
100 %
R
GR
Gain applied
to each channel
GL
0%
gL = 1 gR
36
Stereo Panning
n
Sine law:
sin p
gL gR
=
sin o g L + g R
Maximum
aperture
37
Stereo Panning
n
Tangent law:
tan p
gL gR
=
tan o g L + g R
Maximum
aperture
Stereo Panning
n
GL + GR = 1
For Free Field
GL2 + GR2 = 1
For Diffuse Field
39
Stereo Panning
n
GR = sin( P )
GL = cos( P )
40
Quadraphonic
n
41
Quadraphonic
n
42
43
Dolby Stereo
n
Dolby Surround
Dolby Surround ProLogic
44
Dolby Stereo
n
Maximum aperture is 60
0
-30
30
Phantom images
are only possible
with L & R speakers
45
Dolby Stereo
n
46
Dolby Stereo
n
The Dolby Surround version does not have the center channel
47
3/2 System
n
It uses 5 channels:
L, R, C, LS & RS
3/2 System
n
Diagram:
Surround
channels
are not fixed
100-120
100-120
3/2 System
n
6.1
7.1
50
10.2 System
n
2 Channels at 55
n
n
10.2 System
n
n
52
22.2 System
n
53
22.2 System
n
54
VBAP
n
VBAP
n
g = [ pn
Gain
(1x3)
pm
pk ]
Speakers
coordinates
ln1 ln 2 ln3
l
l
l
m1 m 2 m3
lk1 lk 2 lk 3
56
Binaural Systems
57
Binaural Systems
n
58
Binaural Systems
n
59
HRTF
n
A HRTF is a filter
One for the left ear & one for the right ear
HRTF Measurement
n
HRTF Measurement
n
Anechoic chamber
gets rid of
room reflections
62
HRTF Measurement
n
Impulse
63
HRTF Measurement
n
This example
is for the
right ear
64
HRTF Measurement
n
65
HRTF
n
But you will achieve much more accuracy with your own
HRTFs
66
Binaural Systems
n
Source
HRTFL
HRTFR
R
67
Binaural Systems
n
68
Binaural Systems
n
We
Transaural Systems
n
Transaural Systems
n
We need left speaker to left ear, and right speaker to right ear
But we also have crosstalk
Crosstalk:
dashed lines
71
Transaural Systems
n
How do we do this?
x1
x2
H12
H21
H11
H22
y1
y2
72
Transaural Systems
n
! y $ ! H
# 1 & = # 11
#" y2 &% #" H 21
H12
H 22
$! x $
&# 1 &
&%#" x2 &%
x1
x2
H12
H21
H11
This is the desired sound
(already known)
H22
y1
y2
73
Transaural Systems
n
Block diagram:
74
Transaural Systems
n
Sound Field
Reconstruction Systems
76
77
Instead of making only one scene for all people at the same
time
78
Example:
DOA depends on
the current position
of the listener
Listener 1
Listener 2
Listener 3
(moving)
Now we are:
Head Independent!
79
Both perform the same task but from different point of views
Ambisonics
Wave-Field Synthesis
80
Ambisonics
n
Ambisonics B-Format
n
W: pressure
X: velocity in x axis
Y: velocity in y axis
Z: velocity in z axis
82
Ambisonics B-Format
n
Ambisonics
n
Example of use:
84
Ambisonics
n
Advantages:
Disadvantages:
It
85
Ambisonics
n
86
B-Format
It uses only 4 channels to encode the sound
n
n
87
88
90
91
Wave-Field Synthesis
n
So lets review it
92
Huygens Principle
n
93
Wave-Field Synthesis
n
Original
source
95
Wave-Field Synthesis
n
96
Wave-Field Synthesis
n
Stereo: Only
Point Sources
Focused
source
97
Wave-Field Synthesis
n
Kirchhoff-Helmholtz integral:
98
Wave-Field Synthesis
n
Kirchhoff-Helmholtz integral:
Sound
Source
Listener
Secondary
sources
99
Wave-Field Synthesis
n
Simplifying:
Lets suppose that the volume is an infinite cube, the surface
is now only a plane that divides inside from outside
Now lets reduce even more, from 3D we change to 2D, the
surface is now a straight line
And last, we discretize the surface in N points
100
Wave-Field Synthesis
n
Delay due
to distance
HP Filter
Pressure
Distance
attenuation
101
WFS Limitations
n
Linear array:
102
WFS Limitations
n
Diffraction:
103
WFS Limitations
n
Spatial Aliasing:
104
WFS Limitations
n
Loudspeakers directivity:
In theory, secondary sources are omnidirectional
In practice, speakers are not omnidirectional at all the
frequencies
But this is not so important, in low frequencies they work ok
105
WFS Examples
n
106
WFS Examples
n
107
WFS Examples
n
Uses 96 channels
108
WFS Examples
n
Prototype at the UA
109
Ambisonics VS WFS
n
Ambisonics:
WFS:
But equations allow it, the problem is the money (& computer power)
Using combinations of linear arrays
111
Sound in Cinema
n
Monoaural Sound
n
n
113
Monoaural Sound
n
Sound
band
114
Multichannel Sound
n
115
Fantasound (1940)
n
116
Cinerama (1952)
n
n
n
117
Cinemascope (1953)
n
Born in 1953
The system used an anamorphic lens to deform the image
and convert a 4:3 film in a 16:9 format
The lens must be used both at recording & reproduction
It used 4 audio channels stored in 4 magnetic tracks over the
film
Lens
Magnetic audio tracks
118
Dolby (1970)
n
The 4 channels
are mixed in
2 optic bands
Later it
became an
ISO standard
119
Sensurround (1974)
n
2 at each side
Each one driven with a 1 kW amplifier
Emitting only infrasounds (bellow 20 Hz)
n
Earthquake
120
DTS
n
SDDS
n
121
Digital Sound
n
Dolby
Stereo
(analog)
SDDS
DTS
122
Digital Sound
n
And DTS
DTS
n
124