Microphone Techniques for Stereo and Multichannel

Geoff Martin Bang & Olufsen, Denmark

What well look at

How microphones behave Monitoring environment What different microphone congurations do for you If something doesnt make sense, stop me and ask at any time!

What is sound?
Todays barometric pressure is about 100 kPa Sound is a change in that pressure over time A microphone converts that pressure change into a voltage change

How to make a barometer

Airtight can

Balloon

How to make a barometer

Voltage

pressure mi c

rophone

Voltage

Voltage

How to make a pressure microphone

Pressure Microphone - Sensitivity

S=1
Sensitivity

1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

Sensitivity (gain) Angle of incidence

Angle of Incidence

l a n o i t c e r i d i n Om
S=1

Pressure microphone - Sensitivity

90 120 1 0.8 150 0.6 Sensitivity 0.4 0.2 180 30 60

Angle of incidence
0

210

330

240 270

300

Lets start again

Well build another barometer but this time without the sealed can This means we are measuring the difference in pressure on the front and back of the diaphragm Pressure Difference = Pressure Gradient

Pressure Gradient microphone - Construction

Supporting ring

Diaphragm

Pressure Gradient microphone - Output

Voltage

Voltage

Voltage

Pressure Gradient microphone - Sensitivity

S = cos(angle) 0.8
1 0.6

Sensitivity

0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

Angle of Incidence

Pressure Gradient microphone - Sensitivity

90

l a n o i t Bidirec

S = cos(angle)
120 150

1 0.8 0.6 0.4 0.2

60

30

180

210

330

240 270

300

Lets add them

1 0.8 0.6

Sensitivity

0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

Angle of Incidence

Half-omni, half-bidirectional
1 0.8 0.6

Sensitivity

0.4 0.2 0 -0.2 -0.4 -0.6

S = 0.5 -0.8 + 0.5 cos(angle)

-1 -150 -100 -50 0 50 100 150

Angle of Incidence

Half-omni, half-bidirectional
90 120 1 0.8 150 0.6 0.4 0.2 180 0 30 60

e n o h p o r c i Cardioid m

S = 0.5 + 0.5 cos(angle)

240 270 300

210

330

Make your own Polar Patterns

Pressure Pressure Gradient (Omnidirectional) (Bidirectional) Omnidirectional Subcardioid Cardioid Hypercardioid Bidirectional 100% 75% 50% 25% 0% 0% 25% 50% 75% 100%

Polar Patterns
Omnidirectional Subcardioid Cardioid Hypercardioid Bidirectional
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

Omnidirectional
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 120

1 0.8 0.6

60

150

0.4 0.2

30

180

210

330

240 270

300

Subcardioid
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 120

1 0.8 0.6

60

150

0.4 0.2

30

180

210

330

240 270

300

Cardioid
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 120

1 0.8 0.6

60

150

0.4 0.2

30

180

210

330

240 270

300

Hypercardioid
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 120

1 0.8 0.6

60

150

0.4 0.2

30

180

210

330

240 270

300

Bidirectional
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 120

1 0.8 0.6

60

150

0.4 0.2

30

180

210

330

240 270

300

Localisation in the real world

Left-right localisation relies on:
Interaural time of arrival differences (ITD) Interaural amplitude differences (IAD)

Front-back localisation relies on:

Head rotation cues Reections off the pinnae Shoulder reections

Localization in the reproduced world Phantom image localization relies on:

Interchannel amplitude differences Interchannel time differences

One version
Image position 0 10 20 30 ! Amp. 0.0 dB 2.5 dB 5.5 dB 15 dB or or or or or ! Time 0.0 ms 0.2 ms 0.44 ms 1.12 ms
Gert Simonsen 1983

Other versions
Interchannel Amplitude vs. Phantom Image Angle
20 18 16

Amplitude Difference (dB)

14

12

10

Simonsen (1984) Webers (1968) Theile (1988) de Boer (1940) Sengpiel (1992)

0 0 10 20 30 40 50 60 70 80 90 100 Percentage of distance to loudspeaker

Other versions
Interchannel Time Difference vs. Phantom Image Angle
1.6 1.4

Interchannel Time Difference (ms)

1.2

1 Simonsen (1984) Sengpiel

0.8

0.6

0.4

0.2

0 0 10 20 30 40 50 60 70 80 90 100 Percentage of distance to loudspeaker

Coincident Microphones
We assume identical time of arrival Sensitivity differences produce interchannel amplitude difference Sn = Pn + Gn cos (! + "n)
!

DS = 20

S1 log10 S 2

90 Cardioids - Sensitivity Difference

1 0.8 0.6 0.4

Difference in Sensitivity

0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -150 -100 -50 0 50 100 150

90 Cardioids - Sensitivity Difference

60 15

Difference in Sensitivity (dB)

40 10 20 5 0 -20 -5 -40 -10 -60 -15

-150

-100

-50

50

100

150

Angle of Incidence ()

90 Cardioids - Sensitivity Difference

Angle of Elevation (deg)

Angle of Incidence ()

90 Cardioids - Sensitivity Difference

15 dB 10 dB 5 dB 0 dB -5 dB -10 dB -15 dB

90 Bidirectionals - Sensitivity Difference

30

Sensitivity Difference

20 10 0 -10 -20 -30 -150 -100 -50 0 50 100 150

Angle of Incidence ()

90 Bidirectionals - Sensitivity Difference

30

Sensitivity Difference

20 10 0 -10 -20 -30 -150 -100 -50 0 50 100 150

Opposite Polarity (Out of phase)

Same Polarity (in phase)

Angle of Incidence ()

90 Bidirectionals - Sensitivity Difference

15 dB 10 dB 5 dB 0 dB -5 dB -10 dB -15 dB

90 Hypercardioids - Sensitivity Difference

15 dB 10 dB 5 dB 0 dB -5 dB -10 dB -15 dB

Spaced Omnidirectionals - Time Difference We assume identical amplitudes Differences in distance to the source produce interchannel time D difference d D = d sin ! "T = D / c

Time difference for 40 cm spacing

Time difference for 40 cm spacing

2 ms 1 ms 0 ms -1 ms -2 ms

Panning
Left Right

Summed Power

Gain

Coincident Microphones - Power Sum

Sn = Pn + Gn cos (! + "n)
!

S Power = 10 log10 (S12 + S22)

90 Cardioids - Power Sum

10 Summed power (dB) 0 -10 -20 -30 -40 -150 -100 -50 0 50 100 Angle of Incidence () 150

90 Cardioids - Power Sum

+3 dB 0 dB -3 dB -6 dB -9 dB -12 dB -15 dB

90 Bidirectionals - Power Sum

10 0 Summed power (dB) -10 -20 -30 -40 -150 -100 -50 0 50 100 Angle of Incidence () 150

90 Cardioids - Power Sum

0 dB -10 dB -20 dB -30 dB -40 dB

Correlation Coefcient
An indicator of the relationship between two signals Positive correlation: if one goes up, so does the other Negative Correlation: If one goes up, the other goes down Correlation = 0: if one goes up, we dont know what the other will do Geeky denition: the covariance divided by the product of the standard deviations

Correlation Coefcient

Correlation coefcient Coincident directional mics, direct sound Coincident mics have identical time of arrival Only differences are:
Amplitude Polarity

Correlation coefcient must be either 1, 0 or -1

Correlation coefcient Spaced mics, direct sound

Spaced mics have different times of arrival This results in a phase difference
1

Correlation Coefcient

0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 10

2 3 4

10

10

Frequency (Hz)

Correlation coefcient Coincident directional, diffuse eld

Omni Subcardioid

Cardioid

Hypercardioid Bidirectional

Correlation coefcient 30 cm spaced omnis in a diffuse eld

Conclusions
Every recording situation requires a different microphone conguration Treat standard or textbook congurations only as suggestions Understand the behaviour of the microphones and their conguration but Trust your ears! www.tonmeister.ca/research www.hauptmikrofon.de Michael Williamss AES papers

Fukada Tree
C (Card) L / LS (Omni)

b
L (Card)

R / RS (Omni)

c
R (Card)

a=b=c = 1 - 1.5 m d=0-2m L / R angle = 110 - 130 LS / RS angle = 60 - 90

LS (Card)

RS (Card)

OCT Surround
C (Card) L (Hyper)

R (Hyper)

LS (Card)

c d d

RS (Card)

a = 8 cm b = 40 - 90 cm c = 40 cm d = 10 - 100 cm

OCT Frontsystem + Hamasaki Square

L (Hyper) L (Bi) C (Card)

a c

R (Hyper) R (Bi)

a = 8 cm b = 40 - 90 cm c = ~100 cm Cross side = 2 - 3 m

LS (Bi)

RS (Bi)

OCT Frontsystem + IRT Cross

C (Card) L (Hyper)

R (Hyper)

c
L (Card) R (Card)

a = 8 cm b = 40 - 90 cm c = ~100 cm Cross side = 20 - 25 cm

LS (Card)

RS (Card)

Klepko Technique
L (Hyper) C (Hyper) R (Hyper)

a b

LS

RS

a = 17.5 cm b = 125 cm L/R Angle +/- 30

Corey / Martin Tree

L (Subcard) C (Subcard) R (Subcard)

a = 0 - 15 cm b = 60 - 80 cm c = 60 - 90 cm d = 30 cm

LS RS (Card) (Card) Aimed upwards Aimed upwards

Martin Tree
C (Bi - pointing downwards) L (Bi) R (Bi)

LS (Bi)

RS (Bi)

a = < 70 cm LS/RS angle = 45 - 90