A%20tutorial%20on%20acoustic%20measurements%20for%20the%20non-technician

1
A tutorial on acoustic measurements for the
non-technician

• Svante Granqvist
• Royal Institute of Technology (KTH)
• Dept of Speech Music and Hearing (TMH)
• Stockholm, Sweden
• svante.granqvist_at_speech.kth.se

2
Todays topics

• Sound and microphones
• Room acoustics
• Calibration
• Recommendations

3
Conclusions

• Use omni-directional electret or condenser
  microphones whenever possible
• Do not use directed (e.g. cardioid) microphones
  unless you really need the directivity
• Especially not close to the speaker
• Avoid dynamic microphones
• Place the microphone within the reverberation
  radius of the room
• Keep noise level low
• Establish a routine for level calibration

4
What is sound?

• Demo of sound field
• Sound pressure (pascals, Pa)
• Sound pressure level, SPL (decibels, dB)
• Particle velocity (metres per second, m/s)
• Particle velocity level? Rarely!

5
Sound pressure

• Simple relation to sound intensity
• Our ears are mainly pressure sensitive
• Simple relation to distance (1/r)
• Doubled distance gt halved SP ltgt SPL -6dB
• Pressure has no direction
• Pressure sensitive microphones are
  omni-directional (no directivity)
• So how do they make directed microphones?

6
Particle velocity

• Particle velocity has a direction
• So it can be used to create directivity!
• Particle velocity only gt figure of eight
• Mainly sensitive in two directions

7
Cardioid

• Particle velocity and sound pressure combined gt
  cardioid
• Mainly sensitive in one direction

8
Directivity

• Omni-directional (SP only)
• Directed (involves particle velocity)
• Figure of eight
• Cardioid
• Super-cardioid
• Other special directivity patterns
• Great!
• ...or is it?

9
Directed microphones

• We are primarily interested in sound pressure
• ...but also measure particle velocity
• ...then PV and SP have to be proportional to one
  another!
• Are they?

10
NO!
11
Particle velocity

• Particle velocity is proportional to sound
  pressure, but only in the far field (1/r)
• In the close field, it differs! (1/r2)
• The limit between far and close field depends on
  frequency

12
Particle velocity

• Particle velocity exhibits a bass lift in the
  close field
• proximity effect

13
Proximity effect (cardioid mic)
14
Proximity effect

• Demo

15
Manufacturers data sheets
Cardioid
16
Manufacturers data sheets
Cardioid
17
Manufacturers data sheets
Cardioid
18
Manufacturers data sheets
Omni-directional
19
Manufacturers data sheets

• Frequency responses are mostly measured in the
  far field, even for microphones that obviously
  are intended to be mounted in the close field
• You have to add the proximity effect for directed
  microphones to those curves!

20
Proximity effect (cardioid mic)
21
Manufacturers data sheets
Cardioid
22
Manufacturers data sheets
Cardioid
23
Manufacturers data sheets
Cardioid
24
Manufacturers data sheets
Omni-directional
25
Demo

• Proximity effect
• Hear the bass lift from the directed microphone

26
OK, point taken, he doesnt like directed
microphones

• But then, why are there so many directed
  microphones out there?
• Music industry, broadcasting, stage use etc
• A bass boost of a few dBs does not matter much or
  might even be desired (sound better)
• Noise supression may be more important than a
  flat frequency response
• Most recordings do not have a scientific purpose

27
Transducer type

• Electret/condenser
• Can easily be made to have flat response
• Cheap electret microphones (lt 30 ) can be of
  sufficient quality
• Requires battery/power supply
• Sensitivity may decrease towards end of battery
  life
• Dynamic
• Difficult to acheive a flat response
• Good dynamic microphones are expensive
• Rarely purely pressure sensitive (even though
  data-sheet may say so)
• No need for battery/power supply

28
Bottom line...

• Use omni-directional, electret/condenser
  microphones for scientific purposes!
• Make sure batteries are fresh or use some other
  type of power supply

29
Room acoustics

• In a room sound originates from
• the sound source, directly
• or from reflections at the walls

30
Room acoustics
31
Room acoustics

• Reverberation radius, rr
• The distance where reflected and direct sound are
  equally loud
• Less absorbtion gt stronger reflections gt
  smaller rr

32
Room acoustics
33
Room acoustics

• Reverberation radius
• At what distance is the direct sound as loud as
  the sound that has been reflected from the walls
• Typical value 4 0.5 meters
• Reverberation time
• How long does it take for the sound level to drop
  by 60 dB?
• Typical value 0.5 4 seconds

34
Room acoustics

• How to measure Reverberation time/radius
• Several ways, one would be to record a bang and
  see at what rate the sound level drops
• The time for a 60dB drop corresponds to
  reverberation time
• Calculate reverberation radius from this time

35
Bottom line...

• Within the reverberation radius, conditions are
  similar to free field
• Outside, reflections from the walls dominate the
  sound
• So, put the microphone (well) within the
  reverberation radius!

36
Level calibration

• Most common method
• Record a signal with a known level
• i.e. a calibration tone
• By relating the level of the calibration tone to
  the levels of the signals of interest, absolute
  calibration is acheived

37
Calibration file, example
Calibration tone
The level was 89 dB
38
CalibrationCalibrator

• Procedure
• Mount and start the calibrator (2-10 seconds)
• Unmount calibrator and say the level of the
  calibrator
• Advantages
• Stable calibration tone
• No sensitivity to room acoustics or surrounding
  noise
• Disadvantage
• Calibrator that fits the microphone required
• Important that the seal is tight!

39
Calibration Loudspeaker SPL meter

• Procedure
• Beep at 1kHz 80 dB (2-10 seconds)
• say the level as read on the level meter
• Advantages
• Stable calibration tone
• Disadvantage
• Loudspeaker signal source reqiured
• Some sensitivity to room and surrounding noise

40
Calibration Voice SPL meter

• Procedure
• Sustain /a/ 80 dB (5-10 seconds)
• say the level as read on the level meter
• Advantages
• No loudspeaker required
• Calibration signal (voice) has approximately the
  same spectrum as the signals of interest
• Disadvantages
• Hard to keep the level of the /a/ stable
• Some sensitivity to room and surrounding noise

41
Calibration Voice SPL meter

• Procedure
• Sustain /a/ 80 dB (5-10 seconds)
• say the level as read on the level meter
• Advantages
• No loudspeaker required
• Calibration signal (voice) has approximately the
  same spectrum as the signals of interest
• Automatic compensation for microphone distance
• Disadvantage
• Hard to keep the level of the /a/ stable
• Only valid for this particular distance
• Some sensitivity to room and surrounding noise
• dB meter should be within rr

42
Calibration, directed microphones ?

• Only in the far field (gt30 cm), but still within
  rr
• Only for rough estimation of SPL
• Never use SPL calibrators!
• Dont try this at home

43
Distance compensation

• Sound pressure drops as 1/r
• Re-calculate SPL to appear as recorded at a
  different distance, e.g. record at d25 cm, but
  report at d130 cm.
• Only for omni-directional microphones!
• Formula

44
Bottom line...

• Establish a routine for calibrations
• Dont calibrate directed microphones
• Report SPL at 30 cm
• Compensated or actual
• Beware of the mixer on most PC soundcards

45
Recommendationsmicrophone and room acoustics

• Depend on
• the purpose of recording
• the recording environment
• Noise
• Room acoustics

Example of purposes SPL Spectrum F0 Inverse
filtering HNR Perceptual evaluation
46
SPL

• Omni-directional electret/condenser microphone
• If noisy environment
• Try to attenuate the noise
• Shorten microphone distance (10 cm to the side of
  the mouth)
• Avoid directed microphones for this purpose!
• Put the microphone well within the reverberation
  radius of the room (rr/2)
• Re-calculate or calibrate for 30 cm

47
Spectral properties (spectrogram)

• Omni-directional electret/condenser microphone
• If noisy environment
• Try to attenuate the noise
• Shorten microphone distance (5-10cm to the side
  of the mouth)
• If background noise still is a problem a directed
  microphone can be used, but beware of the
  proximity effect and keep microphone distance
  constant!
• Put microphone well within rr

48
Spectral properties (LTAS, H1-H2, line spectra)

• Omni-directional electret/condenser microphone
• If noisy environment
• Try to attenuate the noise
• Shorten microphone distance (5-10cm to the side
  of the mouth)
• Do not use a directed microphone
• Put microphone well within rr
• Pay attention to reflective surfaces such as
  windows, manuscripts etc.

Added proximity effect, cardioid at 5 cm
49
F0, jitter/shimmer

• Any decent microphone is OK, since periodicity is
  independent of frequency response
• If noisy environment
• Try to attenuate the noise
• Shorten microphone distance (5-10 cm)
• Use a directed microphone
• Check if F0 algorithm is affected by a bass lift!

50
Inverse filtering

• Omni-directional electret/condenser microphone
  flower lt 10 Hz
• Reduce background noise as much as possible
• Never use a directed microphone
• Microphone distance 5-10 cm
• Within rr/10
• Pay attention to reflective surfaces such as
  windows, manuscripts etc.
• Anechoic chamber is preferred

Addition of reflection to the direct signal
51
Harmonics-to-noise ratio

• Omni-directional electret/condenser microphone
• Background noise must be lower than voice noise
• Microphone distance 5-50 cm
• Well within rr

52
Perceptual evaluation

• Omni-directional electret/condenser microphone
• Reduce background noise as much as possible
• Microphone distance 5-50 cm
• Well within rr

53
But you never know...

• For example, the first intention may be to only
  extract F0
• It might turn out, after the recordings are made,
  that the recorded material would be suitable for
  some other measurement, like SPL
• Therefore, do it right from the start!

54
Conclusions

• Use omni-directional electret or condenser
  microphones whenever possible
• Do not use directed (e.g. cardioid) microphones
  unless you really need the directivity
• Especially not close to the speaker
• Avoid dynamic microphones
• Place the microphone within the reverberation
  radius of the room
• Keep noise level low
• Establish a routine for level calibration

55
These were my recommendations

• You may find reasons to not follow them
• But they better be good... ?

56
Questions?
This presentation is available on the
webwww.speech.kth.se/svante/pevoc5 svante.granqv
ist_at_speech.kth.se