reproducing spatial sound properties of auditoria
Angelo Farina (1) and Lamberto Tronchin (2)
(1) Dipartimento di Ingegneria Industriale, Università di Parma, Via delle Scienze 181/A Parma, 43100 ITALY
HTTP://pcfarina.eng.unipr.it - mail: angelo.farina@unipr.it
(2) DIENCA - CIARM, Università di Bologna, Viale Risorgimento 2 Bologna, 40136 ITALY
HTTP://www.ciarm.ing.unibo.it - mail: lamberto.tronchin@unibo.it
Topics
• This paper is a tribute to M. Gerzon, who had foreseen 3D impulse response measurements and 3D Auralization obtained by convolution.
• The advantages (and disadavantages) of employing measured IRs
• Comparison between Auralizations based on calculated and measured IRs
(e.g. Theatre “La Fenice”, Venice)
• Different approaches to 3D Auralization
10-Sep-08 Farina, Tronchin, RADS
Concept (Gerzon, 1975)
Portable PC with 4- channels sound board Original Room
Sound Source
SoundField Microphone
B-format 4-channels signal (WXYZ) Measurement of B-format
Impulse Responses
MLS or sweep excitation signal
Convolution of dry signals with the B-format Impulse
Responses Sound Source
Mono Mic.
B-format Imp. Resp.
of the original room
B-format 4-channels signal (WXYZ) Convolver
Ambisonics decoder
Speaker array in the reproduction room
The measurements of IRs:
1. In case something happens to the original
space (e.g.: La Fenice theater) they contain a detailed “acoustical photography” which is
preserved for the posterity
2. They can be used for studio sound processing, as artificial reverb and surround filters for
today’s and tomorrow’s musical productions
3. Several configurations of listening rooms could be developed, by means of multichannel
auralization, for subjective tests
10-Sep-08 Farina, Tronchin, RADS
Theatre la Fenice, Venice
• The first theatre was realised in 1792 by Gian Antonio Selva, after the burning of Teatro San Benedetto
• In December 1836 the theatre burned down again and was rebuilt by G. and T. Meduna the year after
• The theatre was closed in 1995 for maintainance; it had to open again in February 1, 1996, but it
burned two days before (January 29, 1996)
• A few weeks before the fire, L.Tronchin measured binaural impulse responses
Impulse Responses of La Fenice
In 27 positions a series of binaural impulse responses (with gun shots) was recorded
Each recording is consequently a
stereo file at 16 bits, 48 kHz
During
measurements the room was perfectly fitted, whilst the stage was empty (no scenery)
Point n. 12
10-Sep-08 Farina, Tronchin, RADS
Example n. 1 – La Fenice
• Dry music
• Convolution with
experimental I.R. (pt. 12)
• Convolution with simulated IR
Overture alle Nozze di Figaro di Mozart
Preludio al primo atto della Traviata di G.Verdi
• Dry music
• Convolution with
experimental I.R. (pt. 12)
• Convolution with simulated IR
Stop
Example n. 2 – Paris vs ...la petit Paris
Citè de la Musique, Parigi Auditorium di Parma
Stop
10-Sep-08 Farina, Tronchin, RADS
Advanced IR capture and rendering ( project)
• Description of the measurement technique
• Analysis of some acoustical parameters of some theaters measured
• Description of the processing methods to be employed for transforming the measured data in audible reconstructions of the original spaces
• Description of the usage of the measured data for studio processing, musical production and for scientific Auralization tests
Sound propagation in rooms
Direct Sound
Reflected Sound
Receiver Direct Sound
Reflected Sound
Point Source
10-Sep-08 Farina, Tronchin, RADS
Measurement process
• The desidered result is the linear impulse response of the acoustic propagation h(t). It can be recovered by knowing the test signal x(t) and the measured system output y(t). It is necessary to exclude the effect of the not-linear part K and of the background noise n(t).
Not-linear, time variant
system K[x(t)]
Noise n(t)
input x(t)
+ output y(t) linear system
w(t)⊗h(t) distorted signal
w(t)
Test signal: Log Sine Sweep
x(t) is a sine signal, which frequency is varied
exponentially with time, starting at f1 and ending at f2.
⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
⎟⎟
⎟⎟
⎠
⎞
⎜⎜
⎜⎜
⎝
⎛
−
⋅
⎟⎟⎠
⎜⎜ ⎞
⎝
⎛
⋅
⋅ π
= ⋅ ⎟⎟⎠
⎜⎜ ⎞
⎝
⋅ ⎛
1 e
f ln f
T f
sin 2 )
t (
x 1
2
f ln f T
t
1 2 1
10-Sep-08 Farina, Tronchin, RADS
Deconvolution of Log Sine Sweep
The “time reversal mirror” technique is emplyed: the
system’s impulse response is obtained by convolving the measured signal y(t) with the time-reversal of the test signal x(-t). As the log sine sweep does not have a
“white” spectrum, proper equalization is required
Test Signal x(t) Inverse Filter z(t)
Test Signal – x(t)
10-Sep-08 Farina, Tronchin, RADS
Measured signal - y(t)
The not-linear behaviour of the loudspeaker causes many harmonics to appear
Inverse Filter – z(t)
The deconvolution of the IR is obtained convolving the measured signal y(t) with the inverse filter z(t) [equalized, time-reversed x(t)]
10-Sep-08 Farina, Tronchin, RADS
Result of the deconvolution
The last impulse response is the linear one, the preceding are the harmonics distortion products of various orders
2° 1°
5° 3°
Measurement Setup
The measurement method incorporates all the known techniques:
Binaural
B-format (1st order Ambisonics)
WFS (Wave Field Synthesis, circular array)
ITU 5.1 surround (Williams MMA, OCT, INA, etc.) Binaural Room Scanning
M. Poletti high-order virtual microphones
Any multichannel auralization systems nowadays available is supported
10-Sep-08 Farina, Tronchin, RADS
Measurement Parameters
• Test Signal: pre-equalized sweep
Start Frequency 22 Hz End Frequency 22 kHz
Sweep length 15 s
Silence between sweeps 10 s
Type of sweep LOG
z Deconvolution:
Transducers (sound source #1)
• Equalized, omnidirectional sound source:
Dodechaedron for mid-high frequencies One-way Subwoofer (<120 Hz)
Dodech. LookLine D200
60.0 70.0 80.0 90.0 100.0 110.0 120.0
25 31.5 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300 8000 10000
Frequency (Hz)
Sound Power Level (dB)
Unequalized Equalized
Lw,tot = 94.8 dB Lw,tot = 106.9 dB
10-Sep-08 Farina, Tronchin, RADS
Transducers (sound source #2)
• Genelec S30D reference studio monitor:
Three-ways, active multi-amped, AES/EBU Frequency range 37 Hz – 44 kHz (+/- 3 dB)
1000 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
250 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
2000 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
4000 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
8000 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
16000 Hz
-40 -35 -30 -25 -20 -15 -10 -5 0 0
30
60
90
120
150 180
210 240 270
300 330
Transducers (microphones)
• 3 types of microphones:
Binaural dummy head (Neumann KU-100) 2 Cardioids in ORTF placement (Neumann K- 140)
B-Format 4 channels (Soundfield ST-250)
Turntable
Binaural dummy head Cardioids (ORTF) Soundfield Microphone
10-Sep-08 Farina, Tronchin, RADS
Other hardware equipment
• Rotating Table:
Outline ET-1
z Computer and sound card:
– Signum Data Futureclient P-IV 1.8 GHz
– Aardvark Pro Q-10 (8 ch., 96 kHz, 24 bits)
Measurement procedure
• A single measurement session play backs 36 times the test signal, and simultaneusly record the 8 microphonic channels
10-Sep-08 Farina, Tronchin, RADS
Theatres measured
N. Theatre N. sources/receivers
1 Uhara Hall, Kobe, Japan 2/2
2 Noh Drama Theater, Kobe, Japan 2/2
3 Kirishima Concert Hall, Kirishima, Japan 3/3 4 Greek Theater in Siracusa, Italy 2/1 5 Greek-Roman Theater in Taormina, Italy 3/2
6 Auditorium of Parma, Italy 3/3
7 Auditorium of Rome (Sala 700), Italy 3/2 8 Auditorium of Rome (Sala 1200), Italy 3/3 9 Auditorium of Rome (Sala 2700), Italy 3/5
10 Bergamo Cathedral, Italy 2/1
11 Teatro Valli, Reggio Emilia, Italy 5/1 12 Sydney Opera House, Opera Theatre 4/2 13 Sydney Opera House, Concert Hall 3/3 14 Sydney Opera House, The Studio 3/1
15 Tearo Regio, Parma, Italy 6/1
Reverberation Time T20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
31.5 63 125 250 500 1000 2000 4000 8000 16000
Frequency (Hz)
T20 (s)
Uhara
Noh
Kirishima
Siracusa
Taormina
Audit. Parma
Roma-700
Roma-1200
Roma-2700
Bergamo Cathedral Valli-RE
SOH Concert Hall
SOH-Opera Theatre SOH-The Studio Regio Parma
Uhara Hall, Kobe, Japan
T 20= 1.44 s
10-Sep-08 Farina, Tronchin, RADS
Noh theater, Kobe, Japan
T 20= 1.14 s
Kirishima Concert Hall, Japan
10-Sep-08 Farina, Tronchin, RADS
Kirishima Concert Hall, Japan
T 20= 1.93 s
Greek Theater in Siracusa
T 20= 0.65 s
10-Sep-08 Farina, Tronchin, RADS
Roman Theater in Taormina
T 20= 1.15 s
Parma Auditorium, Italy
T 20= 2.08 s
10-Sep-08 Farina, Tronchin, RADS
Rome Auditorium, 700 seats
T 20= 2.04 s
Rome Auditorium, 1200 seats
T 20= 2.10 s
10-Sep-08 Farina, Tronchin, RADS
Rome Auditorium, 2700 seats
T 20= 2.56 s
Bergamo’s Cathedral, Italy
T 20= 2.95 s
10-Sep-08 Farina, Tronchin, RADS
Teatro Valli, Reggio Emilia, Italy
T 20= 1.55 s
Sydney Opera House
10-Sep-08 Farina, Tronchin, RADS
Sydney Opera House – opera theatre
T 20= 1.16 s
Sydney Opera House – concert hall
T 20= 2.04 s
10-Sep-08 Farina, Tronchin, RADS
Sydney Opera House – the studio
T 20= 0.76 s
Teatro Regio in Parma (Italy)
T 20= 1.11 s
10-Sep-08 Farina, Tronchin, RADS
Acoustical Parameters (ISO 3382)
z Center Time TS:
( )
∫ ( )
∫
∞
∞
τ
⋅ τ
τ
⋅ τ
⋅ τ
=
0 2 0
2
s
d p
d p T
( ) ( ) ⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
⋅
⋅
⋅
=
∫
∫
∞ ms ms
dτ τ p
dτ τ p C
80 2 80
0 2
80 10 lg
z Clarity C80:
( ) ( )
100 d p
d p D
0 2 ms 50
0 2
⋅ τ
⋅ τ
τ
⋅ τ
=
∫
∫
z Definition D: ∞
20 dB
z Reverberation Time T20:
T20/3
Acoustical Parameters (ISO 3382)
•
Strength: G =SPL−Lw +31 dB( ) ( )
∫ ( )
∫
τ
⋅ τ
τ
⋅ τ
⋅ τ
= 80ms
ms 0
2 W ms 80
ms 5
W Y
d h
d h
h
z LFC: LFC
( )
∫ ( )
∫
⋅
⋅
= ms
ms W ms
ms Y
d h
d h
LF 80
0 2 80
5 2
τ τ
τ
z LF: τ
( )
( ) ( )
( ) ∫ ( )
∫
∫
∞
∞
−
∞
∞
−
∞
∞
−
τ
⋅ + τ
⋅ τ
⋅ τ
τ
⋅ + τ
⋅ τ
= τ ρ
d t h d h
d t h h
2 s 2
d
s d
z IACC:
10-Sep-08 Farina, Tronchin, RADS
Analysis of spatial attributes
Polar diagrams of IACC and (1-LF)
IACC Auditorium Parma - Sorgente a sx
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0
10 20
30 40
50 60
70
80
90
100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
300 310
320 330
340 350 Sorgente
IACC Auditorium Roma (Sala 1200) - Sorgente a sx
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0
10 20
30 40
50 60
70 80
90
100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
300 310
320 330
340 350 Sorgente
(1-LF) Auditorium Parma – Sorgente a sx
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0
10 20
30 40
50 60
70
80
90
100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
300 310
320 330
340 350 Sorgente
(1-LF) Auditorium Roma (Sala 1200) – Sorgente a sx
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0
10 20
30 40
50 60
70 80
90
100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290
300 310
320 330
340 350 Sorgente
Auditorium 1-LF IACC Parma 0.725 0.266 Roma 0.676 0.344
10-Sep-08 Farina, Tronchin, RADS
Auralization by convolution
• The basic method consists in convolution of a dry signal with a set of
impulse responses corresponding to the required output format for surround (2 to 24 channels).
• The convolution operation can nowadays be implemented very
efficiently on a modern PC through an ancient algorithm (equally-partitioned FFT processing, Stockam 1966).
Auralization types
• Stereo (ORTF on 2 standard loudspeakers at +/- 30°)
• Rotation-tracking reproduction on headphones (Binaural Room Scanning)
• Full 3D Ambisonics 1st order (decoding the B-format signal)
• ITU 5.1 (from different 5-mikes layouts)
• 2D Ambisonics 3rd order (from Mark Poletti’s circular array microphone)
• Wave Field Synthesis (from the circular array of Soundfield microphones)
• Hybrid methods (Ambiophonics)
10-Sep-08 Farina, Tronchin, RADS
ORTF Stereo
Playback occurs over a pair of loudspeakers, in the standard configuration at angles of +/- 30°, each being fed by the signal of the
corresponding microphone
2 Microphones
60°
2 Loudspeakers
Binaural (Stereo Dipole)
Reproduction occurs over 2 loudspeakers angled at +/- 10°, being fed through a “cross-talk cancellation” digital filtering system
…
2 3
1
Original 2-channels recording of the signals coming from N sources
d1l d1r
d2l d2r
xr
xl
Cross-talk canceller
dNl dNr N
20°
10-Sep-08 Farina, Tronchin, RADS
Binaural (Stereo Dipole#2)
L R
hll hrr
hlr hrl Binaural
stereo signal yl yr
pl pr
xl
xr
convolver
fll
flr
frl
frr
yl yr
Binaural (Stereo Dipole#3)
( )( )
( )
( ) ( )
⎪⎪
⎪
⎩
⎪⎪
⎪
⎨
⎧
⊗
−
⊗
=
⊗
=
⊗
−
=
⊗
−
=
⊗
=
rl lr
rr ll
ll rr
rl rl
lr lr
rr ll
h h
h h
InvFilter InvDen
InvDen h
f
InvDen h
f
InvDen h
f
InvDen h
f C(ω) = FFT(hll)⋅FFT(hrr)− FFT(hlr)⋅FFT(hrl)
( ) [ ( )]
[ ( )ω ] ( ) ( )ωω ε ω
ω = ⋅ +
C C
Conj
C InvDen Conj
hll
hlr
hrl
hrr
fll
flr
frl
frr
10-Sep-08 Farina, Tronchin, RADS
Binaural (Dual Stereo Dipole)
advantages advantages: :
3D sound reproduction3D sound reproduction
RotatingRotating ofof the headthe head
The cross-The cross-talk talk filtersfilters could
could equaliseequalise alsoalso the the loudspeakers
loudspeakers
disadvantages disadvantages: :
Low frequenciesLow frequencies
ColorationColoration outsideoutside the the
““sweetsweet spotspot””
Scheme
Subwoofer
Binaural (Dual Stereo Dipole#2)
Frontal Rear
Quested 2108 monitors Quested F11P monitors
10-Sep-08 Farina, Tronchin, RADS
Ambisonics 3D 1st order
Reproduction occurs over an array of 8-24 loudspeakers, through an Ambisonics decoder
Original Room
Sound Source
SoundField Microphone
B-format 4- channels signal
(WXYZ) Ambisonics decoder
Speaker array in the reproduction room
Rooms for Ambisonics 3D 1st order
University of Parma
University of Bologna
10-Sep-08 Farina, Tronchin, RADS
ITU 5.1 surround
• Williams MMA
Schematic of the setup C : Cardioid, 0°
L, R : Cardioid, ± 40°
LS, RS : Cardioid, ± 120°
z INA-5
Schematic of the setup C : Cardioid, 0°
L, R : Cardioid, ± 90°
LS, RS : Cardioid, ± 150°
ITU 5.1 surround
OCT
73 cm
Schematic of the setup C : Cardioid, 0°
L, R : Super Cardioid, ± 90°
LS, RS : Cardioid, ± 180°
10-Sep-08 Farina, Tronchin, RADS
Virtual high-order microphones (M. Poletti)
One of the two ORTF cardioid is
employed, which samples 36 positions along a 110 mm-radius
circumference
1 , 0 6 n
n 3
cos D
1 , 0 4 n
n 2
cos D
1 , 0 2 n
n cos
D 1 D
n , 3
n , 2
n , 1
0
⎟ =
⎠
⎜ ⎞
⎝
⎛ ⋅ϑ+ ⋅π
=
⎟ =
⎠
⎜ ⎞
⎝
⎛ ⋅ϑ+ ⋅π
=
⎟ =
⎠
⎜ ⎞
⎝
⎛ϑ+ ⋅π
=
From these 36 impulse responses it =
is possible to derive the response of cylindrical harmonics microphones (2D Ambisonics) up to 5th order.
0 1 0
90
180
270 0
1 0
90
180
270 0
1 0
90
180
270 0
1 0
90
180 270
Wave Field Synthesis (WFS)
Flow diagram of the process
microphones
loudspeakers
Original space Virtual space
WFS
10-Sep-08 Farina, Tronchin, RADS
Hybrid methods (Ambiophonics)
Ambiophonics 3D (10 loudspeakers):
Hybrid methods (Ambiophonics#2)
10-Sep-08 Farina, Tronchin, RADS
Conclusions
• The spatial informations have to be accurately sampled, making it possible to store, analyze and preserve these
“3D acoustical photographies” of existing musical spaces for the posterity
• Many different kinds of impulse-response measurements are required for different 3D auralization methods: a proper set-up should include all different approaches
• Once the impulse responses are stored in suitable multichannel formats, they become available for surround productions with today technologies (ITU 5.1, 1st order Ambisonics) and future, more advanced methods (high order Ambisonics, WFS, Ambiophonics)
• The only point which requires substantial enhancements:
sound sources used for IR measurements
Future enhancements
Sound source for realistic
emulation of an human singer
10-Sep-08 Farina, Tronchin, RADS
Future enhancements
Omnidirectional sound source with enhanced power frequency response
Dodech. LookLine D300 + Sub Audio Pro B2.27
60 70 80 90 100 110 120
25 31.5 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300 8000 10000
Frequency (Hz)
Sound Power Level (dB)
Unequalized Equalized Lw,tot = 100.5 dB
Lw,tot = 108.4 dB
Acknowledgements
This research was started thanks to the support of Waves, Tel Aviv, Israel (www.waves.com)
For years 2004 and 2005 the research is also supported by the Italian Ministry for the
University and Research (MIUR)
During 2004 a new web site will be started:
www.acoustics.net
This will provide free access to the whole data- base, and it will be possible to add contributions.