Noise: Quantification and Perception

1
Noise Quantification and Perception

• Architectural Acoustics II
• February 11, 2008

2
Symphony Hall, Boston
3
Symphony Hall, Boston
http//www.nytimes.com/2007/06/03/arts/music/03kra
m.html
http//www.allposters.com/-sp/Symphony-Hall-Boston
-MA-Posters_i1119076_.htm
4
Symphony Hall, Boston
http//upload.wikimedia.org/wikipedia/commons/thum
b/5/57/Symphony_hall_boston.jpg/800px-Symphony_hal
l_boston.jpg
5
Symphony Hall, Boston
From Beranek, Concert and Opera Halls How They
Sound
6
Outline

• Measuring noise
• Sound-level meters
• Noise metrics
• Speech intelligibility metrics using noise levels
• Basic noise control concepts
• Intensity measurements

7
Sound Level Meters

• Time constants
• Given a sound raised instantaneously to an SPL of
  L, the meter should display (L 2) dB within one
  time constant.
• Why 2 dB? If SPL L has energy E, the meter
  registers (1 1/e)E in one time constant.
• e 2.718, 10log10(1-1/e) -2
• Time Response
• Slow Time constant 1 sec
• Fast Time constant 125 ms
• Impact Time constant 35 ms rising, 1.5
  sec falling

Image from www.bk.dk, BK 2260 Investigator
8
Sound Level Meters

• Frequency Response
• Linear, A-weighted,C-weighted
• Full bandwidth, 1/1-octave, 1/3-octave
• Classes (ANSI S1.4-1983)
• 0 (Laboratory) 0.2 dB, 22.4 11200 Hz
• 1 (Precision) 0.5 dB, 22.4 11200 Hz
• 2 (General Purp.) 0.5 dB, 63.0 2000 Hz
  1.0 dB, 22.4
  11200 Hz
• Orientation
• For free-field measurements, point the meter at
  the noise source (normal incidence)
• For diffuse-field measurements, the meter
  orientation is not too important (random
  incidence)

Image from www.bk.dk, BK 2260 Investigator
9
A-Weighting Review
Octave-Band Center Frequency (Hz) 31.5 63 125 250 500 1k 2k 4k 8k 16k
A-Weighting Adjustment (dB) -39 -26 -16 -9 -3 0 1 1 -1 -7
10
C-Weighting
Octave-Band Center Frequency (Hz) 31.5 63 125 250 500 1k 2k 4k 8k 16k
C-Weighting Adjustment (dB) -3 -1 0 0 0 0 0 -1 -3 -8
11
From dB(A) to NC/RC

• dB(A) is typically insufficient to describe
  interior noise conditions (no spectral
  information)
• NC (Noise Criterion) and RC (Room Criterion)
  metrics were developed to better describe
  interior noise, specifically that generated by
  mechanical systems
• These metrics better approximate the human
  response to various noise spectra and provide us
  with more detailed analysis information

From Paul Henderson
12
Noise Criterion (NC)

• Single number rating based on octave band levels
• 63 Hz to 8,000 Hz frequency range
• Compare measured spectra with NC curves (tangent
  basis)
• 5 point resolution (NC-15 to NC-65)

From Paul Henderson and MJR Fig. 8.2
13
Calculating the NC Rating

• The NC Rating is the lowest NC curve that lies
  entirely above all measured data points
• In this example, the noise is NC-40, and it is
  limited by the 500 Hz octave band

From Paul Henderson
14
Limitations of the NC Rating

• Provides no limits to low frequency noise below
  the 63 Hz octave band
• Permits excessive high frequency noise above
  2,000 Hz
• Provides no information on spectrum balance or
  sound quality

From Paul Henderson
15
Room Criterion

• Introduced in 1981, approved by ASHRAE in 1995
• Two-parameter rating based on octave band levels
• 16 Hz to 4,000 Hz octave band range
• First parameter is the SIL(3) (arithmetic average
  of noise levels in the 500, 1k, and 2k Hz octave
  bands)
• Second parameter is a sound quality rating
  (Hissy, Neutral , Rumbly, Tonal, Vibration)

From Paul Henderson
16
Room Criterion

• Each line has a -5 dB per octave slope
• The RC-X line crosses X dB at 1000 Hz

MJR, Figure 8.3, p. 165
17
Finding the RC Limit Curve

• Draw an RC line ( ) with slope -5 dB/oct that
  intersects the 1000 Hz band at the SIL(3)
• The limit curve (- - -) is 5 dB above the RC line
  at and below 500 Hz and 3 dB above the RC line at
  and above 1000 Hz

RC-36
From Paul Henderson
18
Determine the RC Quality Rating

• (R) for rumbly if data exceeds limit curve at or
  below 500 Hz
• (H) for hissy if data exceeds limit curve at or
  above 1000 Hz
• (N) for neutral if spectrum is below limit curve
• (T) for tone if audible (any one band is at least
  5 dB above both of its neighboring bands)
• (V) for noise induced vibrations in light-weight
  structures (above 75 dB at 16 or 31 Hz, 80 dB at
  63 Hz)

From Paul Henderson
19
Other Noise Metrics

• Balanced Noise Criterion (NCB)
• Proposed by Beranek in 1989
• Extend lower in frequency than original NC curves
• More stringent at high frequencies than original
  NC curves
• Similar quality ratings (e.g. rumbly and hissy)
  to RC rating system

http//ceae.colorado.edu/muehleis/classes/aren402
0/handouts/lecture24/nc_rc.pdf
20
Other Noise Metrics

• Room Criterion Mark II
• Proposed by Blazier in 1997
• More stringent at low frequencies than the
  original RC curves
• Uses a Quality Assessment Index (deviations from
  RC curve in low, mid, and high frequencies) to
  qualify the numeric rating

http//ceae.colorado.edu/muehleis/classes/aren402
0/handouts/lecture24/nc_rc.pdf
21
Blazier and RC Mark II

• Three factors influence the subjective response
  to HVAC-related background noise
• The loudness of the noise relative to the noise
  created by normal activities in the space
• The potential for task interference e.g. the
  reduction of speech intelligibility
• The quality of the noise, e.g. a
  neutral-sounding noise spectrum will be judged
  mainly by its loudness but a hissy or rumbly
  noise spectrum is inherently more irritating
  regardless of loudness

Blazier, W., "RC Mark II A refined procedure for
rating the noise of heating, ventilating, and
air-conditioning (HVAC) systems in buildings,"
Noise Control Eng. J. Vol. 45, no. 6, pp.
243-150. Nov-Dec 1997.
22
Blazier and RC Mark II

• RC Mark II Rating takes the form RC xx(yy)
• xx is the value of the RC reference curve
  corresponding to the arithmetic average of the
  levels in the 500, 1k, and 2k Hz octave bands
• yy is a qualitative descriptor
• N neutral
• LF low-frequency dominant (rumble)
• LFA substantial sound-induced vibration
• LFB moderate sound-induced vibration
• MF mid-frequency dominant (roar)
• HF high-frequency dominant (hiss)

Blazier, W., "RC Mark II A refined procedure for
rating the noise of heating, ventilating, and
air-conditioning (HVAC) systems in buildings,"
Noise Control Eng. J. Vol. 45, no. 6, pp.
243-150. Nov-Dec 1997.
23
Recommended Background Noise Levels
MJR Table 8.1, pg. 168
24
Recommended Background Noise Levels
MJR Table 8.1, pg. 168
25
Various Levels

• LEQ (Equivalent (Continuous) Sound Level)
• Given a time-variant sound-pressure level
  measured over time T, the LEQ is the constant SPL
  which contains an equal amount of energy over
  time T
• LDN (Day Night Equivalent Sound Level)
• A 24-hour LEQ calculated with a 10 dB penalty for
  levels measured between 1000 PM and 700 AM
• LD daytime LEQ, LN nighttime LEQ
• Ln (Exceedance Level)
• SPL equaled or exceeded n of the time during a
  measurement period. L10 is often used to
  represent the maximum level and L90 is often used
  to represent the ambient level

26
Various Levels

• TNI (Traffic Noise Index)
• TNI 4(L10 L90) L90 30 (dBA)
• NPL or LNP (Noise Pollution Level)
• NPL LEQ sk
• s standard deviation of the time varying level
• k 2.56 (found from studies of subjective
  response to time-varying noise levels)
• Uses A-weighted LEQ

27
Various Levels

• SEL (Sound Exposure Level)
• Li level for a given one-second period
• N number of seconds in the measurement period
• SENEL (Single Event Noise Exposure Level)
• SEL of a single sound event calculated over a
  period in which the level is within 10 dB of the
  maximum level. Often used to quantify noise for
  individual aircraft fly-overs
• CNEL (Community Noise Equivalent Level)
• CNEL SENEL 10log10(ND 3NE 10NN) 49.4
  (dB)
• ND number of daytime flights (7 AM to 7 PM)
• NE number of evening flights (7 PM to 10 PM)
• NN number of nighttime flights (10 PM to 7 AM)

28
CNEL Corrections
Type of Correction Description Correction (dB)
Seasonal Summer (windows open) Winter (windows closed) 0 5
Outdoor Noise Level Quiet suburban or rural community Normal suburban community Urban residential community Noisy urban residential community Very noisy urban res. community 10 5 0 -5 -10
Previous Exposure No prior exposure to intruding noise Some previous exposure Considerable previous exposure 5 0 -5
Pure Tone or Impulse Pure tone or impulsive character 5
http//www.sfu.ca/sonic-studio/handbook/Community_
Noise_Equivalent.html
29
A Few More
Long Figure 4.22, p. 143
30
Noise Source Directivity

• Q (directivity) of a source is
• For a source against a wall (for example)

Total power (W) is radiated uniformly over a
hemispherical surface.
31
Noise Source Location
MJR, p. 174
32
OSHA and Noise Exposure

• OSHA is the Occupational Safety and Health
  Administration
• They provide guidelines (legal limits) for
  workplace noise exposure or noise dose
• Noise Dose
• where Ci is the total daily exposure time to a
  specific noise level (e.g. 90 dBA) and Ti is the
  maximum permissible exposure time for that level
• D gt 1is illegal

33
OSHA and Noise Exposure
Noise dose is measured with a noise dosimeter.
http//www.nonoise.org/hearing/hcp/25.gif
MJR Table 8.2, pg. 169
34
Speech Intelligibility

• Statistical Measures Human Listeners
• Modified Rhyme Test Listeners are given lists of
  6 rhyming or similar-sounding words (e.g. went
  sent bent dent tent rent OR cane case cape cake
  came cave) and are asked to choose which has been
  spoken
• Diagnostic Rhyme Test Listeners are given pairs
  of rhyming words and are asked to choose which
  has been spoken
• Machine Measures
• Percentage Articulation Loss of Consonants
  (ALCons)
• Calculated using RT, speaker-to-listener
  distance, room volume, and speaker directivity
• Speech Transmission Index (STI)
• Changes in the modulation of speech intensity are
  measured for listener positions
• Articulation Index
• Speech Interference Level

35
Articulation Index

• Combines the effects of source level, background
  noise, and hearing sensitivity
• Given the source level and the background-noise
  level (in octave bands), calculate the
  signal-to-noise ratio in each band
  SNR LSource LNoise (dB)
• If SNR gt 30, SNR 30
• If SNR lt 0, SNR 0
• Then

36
Articulation Index

• Use this table of weighting factors to calculate
  AI S SNR weighting factor
• AI 0.7 is desired, lt 0.5 is unacceptable

Octave-Band Center Frequency (Hz) Weighting Factor
250 0.0024
500 0.0048
1000 0.0074
2000 0.0109
4000 0.0078
37
Speech Interference Level

• SIL (or PSIL) evaluates the impact of background
  noise on speech communication
• SIL(3) is the arithmetic average of the SPL in
  the 500, 1,000, and 2,000 Hz octave bands
• SIL(4) is the arithmetic average of the SPL in
  the 500, 1,000, 2,000 and 4,000 Hz octave bands

From Paul Henderson
38
SIL and Distance
MJR Figure 8.1, pg. 162
39
Speech Interference Level
Foreman, Sound Analysis and Noise Control, Fig.
7.4
40
MTF and STI

• Modulation transfer function (MTF)
• Start with the idea that speech is well
  represented by modulated bands of noise
• Speech is interfered with by reverberation and
  background noise which effectively modify the
  modulation

Long, Fig. 4.28, p. 151
41
MTF and STI

• The effect of background noise is independent of
  the modulation frequency, while the effect of
  reverberation is not
• Skipping a few details, the modulation reduction
  factor is

42
MTF and STI

• m(fm) is calculated for
• fm from 0.63 to 12.5 Hz in 14 1/3-octave steps
• 7 octave bands of noise, from 125 Hz to 8 kHz
• The result is a graph like this with 98 (7 x 14)
  values

Long, Fig. 4.28, p. 151
43
MTF and STI

• Now find the apparent signal-to-noise ratio for
  all 98 values of m
• And the average LSNapp weighted by octave band
• Finally

Long, Fig. 4.28, p. 151
44
STI Comparisons
Long, Fig. 4.29
45
STI Comparisons
Long, Figs. 4.30
46
STI Comparisons
Long, Figs. 4.29
47
RASTI RApid STI
Long, Figs. 4.33
48
Basic Noise Control

• Address the source
• Enclose it
• Modify it to reduce noise production
• Address the path
• Add a barrier between the source and receiver
• Add absorption
• Address the receiver
• Distribute ear plugs or other hearing protection
  and enforce their use

49
Noise Barrier Performance
http//www.ashraeregion7.org/tc26/pastprograms/Out
door_Noise/barriers.pdf
50
Noise Barrier Performance

• Barrier attenuation SPL reduction provided by
  the barrier under free-field conditions (no
  ground absorption considered)
• From MJR
• ?L 10log10(20N 3) where
• N 2d/? (called the Fresnel Number)
• d length of shortest path from S to R over the
  barrier minus the length of the direct path from
  S to R
• ? wavelength
• From every other noise control reference

51
Noise Reduction

• NR achieved by adding absorption in a room
• A1 total room absorption before modifications
• A2 total room absorption after modifications
• NR achieved by a partition between two spaces
• TL transmission loss of the partition
• ARec total absorption in the receiving room
• S surface area of the partition

52
Other Measurement Options

• Thus far, weve only considered noise
  measurements based on sound pressure. Is that all
  we can measure?
• Pressure is a scalar value (as opposed to a
  vector) so it provides no directional
  information.
• Intensity probes are becoming popular as tools to
  locate noise sources/leaks.
• Arrays can be used for this too.

53
Intensity Probe

• Two omni mics are mounted face to face at a known
  separation distance (?x)
• Recall (for a plane wave) I pu
• p pressure, u particle velocity
• Now consider Eulers equation
• Solve for particle velocity

pa pb pressure difference between two mics
54
Intensity Probe

• Use particle velocity and average pressure (pa
  pb)/2 to find intensity
• Orientation of the probe can be changed to find
  the strongest intensity, which (likely) indicates
  the direction toward the noise source

55
Directional Arrays
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
56
Directional Arrays
Original Wall STC 56
(a) With a 5.4-cm hole STC 53
(b) With a 3.8-cm sealed pipe in the hole STC 56
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
57
Directional Arrays
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
58
Directional Arrays
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
59
Directional Arrays
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
60
Directional Arrays
B. Gover and J. Bradley, Identification of Weak
Spots in the Sound Insulation of Walls Using a
Spherical Microphone Array, in Proc. NOISE-CON
2005, Minneapolis, October 2005.
61
Directional Arrays
Open Spherical Array
Rigid Spherical Array
More on these later in the semester