What is sound

What is sound?
Sound is the quickly varying pressure wave within a medium. We usually mean audible sound, which is the sensation (as detected by the ear) of very small rapid changes in the air pressure above and below a static value. This "static" value is atmospheric pressure (about 100,000 Pascals) which does nevertheless vary slowly, as shown on a barometer. Associated with the sound pressure wave is a flow of energy. Sound is often represented diagrammatically as a sine wave, but physically sound (in air) is a longitudinal wave where the wave motion is in the direction of the movement of energy. The wave crests can be considered as the pressure maxima whilst the troughs represent the pressure minima.

How small and rapid are the changes of air pressure which cause sound? When the rapid variations in pressure occur between about 20 and 20,000 times per second (i.e. at a frequency between 20Hz and 20kHz) sound is potentially audible even though the pressure variation can sometimes be as low as only a few tens of millionths of a Pascal. Movements of the ear drum as small as the diameter of a hydrogen atom can be audible! Louder sounds are caused by greater variation in pressure. A sound wave of one Pascal amplitude, for example, will sound quite loud, provided that most of the acoustic energy is in the mid-frequencies (1kHz - 4kHz) where the human ear is most sensitive. It is commonly accepted that the threshold of human hearing for a 1 kHz sound wave is about 20 micro-Pascals.

What makes sound?
Sound is produced when the air is disturbed in some way, for example by a vibrating object. A speaker cone from a high fidelity system serves as a good illustration. It may be possible to see the movement of a bass speaker cone, providing it is producing very low frequency sound. As the cone moves forward the air immediately in front is compressed causing a slight increase in air pressure, it then moves back past its rest position and causes a reduction in the air pressure (rarefaction). The process continues so that a wave of alternating high and low pressure is radiated away from the speaker cone at the speed of sound.

What is a decibel (dB)?

The decibel is a logarithmic unit which is used in a number of scientific disciplines. Other examples are the Richter scale for earthquake event energy and pH for hydrogen ion concentration in liquids.

In all cases the logarithmic measure is used to compare the quantity of interest with a reference value, often the smallest likely value of the quantity. Sometimes it can be an approximate average value.

In acoustics the decibel is most often used to compare sound pressure, in air, with a reference pressure. References for sound intensity, sound power and sound pressure in water are amongst others which are also commonly in use.

Reference sound pressure (in air) = 0.00002 = 2E-5 Pa (rms)
     "      "   intensity         = 0.000000000001 = 1E-12 W/m^2
     "      "     power           = 0.000000000001 = 1E-12 W
     "      "   pressure (water)  = 0.000001 = 1E-6 Pa

Acousticians use the dB scale for the following reasons:

1) Quantities of interest often exhibit such huge ranges of variation that a dB scale is more convenient than a linear scale. For example, sound pressure radiated by a submarine may vary by eight orders of magnitude depending on direction.

2) The human ear interprets loudness more easily interpreted with a logarithmic scale than with a linear scale.

How is sound measured?

A sound level meter is the principal instrument for general noise measurement. The indication on a sound level meter (aside from weighting considerations) indicates the sound pressure, p, as a level referenced to 0.00002 Pa, calibrated on a decibel scale.

Sound Pressure Level = 20 x lg (p/0.00002) dB

Often, the "maximum" level and sometimes the "peak" level of the sound being measured is quoted. During any given time interval the peak level will be numerically greater than the maximum level and the maximum level will be numerically greater than the (rms) sound pressure level;

peak>max>rms.

What does dB(A) or "A-Weighted" mean? C-Weighted?

A sound level meter that measures the sound pressure level with a "flat" response will indicate the strength of low frequency sound with the same emphasis as higher frequency sounds. Yet our ear perceives low frequency sound to be of less loudness that higher frequency sound. The eardrum- stapes-circular window system behaves like a mechanical transformer with a finite pass band. In EE parlance, the "3 dB" rollover frequencies are approximately 500 Hz on the low end and 8 kHz on the high end. By using an electronic filter of attenuation equal to that apparently offered by the human ear for sound each frequency (the 40-phon response curve), the sound level meter will now report a numerical value proportional to the human perception of the strength of that sound independent of frequency. Section 8.2 shows a table of these weightings.

Unfortunately, human perception of loudness vis-à-vis frequency changes with loudness. When sound is very loud - 100 dB or more, the perception of loudness is more consistent across the audible frequency band. "B" and "C" Weightings reflect this trend. "B" Weighting is now little-used, but C-Weighting has achieved prominence in evaluating annoying community noises such as low frequency sound emitted by artillery fire and outdoor rock concerts. C-Weighting is also tabulated in 8.2.

The first electrical sound meter was reported by George W Pierce in Proceedings of the American Academy of Arts and Sciences, v 43 (1907-8) A couple of decades later the switch from horse-drawn vehicles to automobiles in cities led to large changes in the background noise climate. The advent of "talkies" - film sound - was a big stimulus to sound meter patents of the time, but there was still no standard method of sound measurement. "Noise" (unwanted sound) became a public issue.

The first tentative standard for sound level meters (Z24.3) was published by the American Standards Association in 1936, sponsored by the Acoustical Society of America. The tentative standard shows two frequency weighting curves "A" and "B" which were modeled on the response of the human ear to low and high levels of sound respectively.

With the coming of the Walsh-Healy act in 1969, the A-Weighting of sound was defacto presumed to be the "appropriate" weighting to represent sound level as a single number (rather than as a spectrum). With the advent of US FAA and US EPA interests in the '70's, the dBA metric was also adapted by them, and with the associated shortfall in precision.

[Editor's Note: A single number metric such as dBA is more easily understood by legal and administrative officials, so that promulgation, enforcement and administrative criteria and actions are understandable by more parties, often at the expense of a more precise comprehension and engineering action capability. For instance, enforcement may be on a dBA basis, but noise control design demands the octave-band or even third-octave band spectral data metric.]

The most commonly referenced weighting is "A-Weighting" dB(A), which is similar to that originally defined as Curve "A" in the 1936 standard. "C-Weighting" dB(C), which is used occasionally, has a relatively flat response. ""U-Weighting"" is a recent weighting which is used for measuring audible sound in the presence of ultrasound, and can be combined with A-Weighting to give AU-Weighting. The A-Weighting formula is given in section 8 of this FAQ file.

In addition to frequency weighting, sound pressure can be weighted in time with fast, slow or impulse response. Measurements of sound pressure level with A-Weighting and fast response are also known as the "sound level".

Many modern sound level meters can measure the average sound energy over a given time. this metric is called the "equivalent continuous sound level" (L sub eq). More recently, it has become customary to presume that this sound measurement was A-Weighted if no weighting descriptor is listed.

How are decibel sound levels added?

If there are two uncorrelated sound sources in a room - for example a radio producing an average sound level of 62.0 dB, and a television producing a sound level of 73.0 dB - then the total decibel sound level is a logarithmic sum i.e.

Combined sound level = 10 x lg ( 10^(62/10) + 10^(73/10) )

= 73.3 dB

Note: for two different sounds, the combined level cannot be more than 3 dB above the higher of the two sound levels. However, if the sounds are phase related ("correlated") there can be up to a 6dB increase in SPL.

How does the ear work?

The eardrum is connected by three small jointed bones in the air-filled middle ear to the oval window of the inner ear or cochlea, a fluid- filled spiral shell about one and a half inches in length. Over 10,000 hair cells on the basilar membrane along the cochlea convert minuscule movements to nerve impulses, which are transmitted by the auditory nerve to the hearing center of the brain.

The basilar membrane is wider at its apex than at its base near the oval window; the cochlea tapers towards its apex. Groups of the delicate hair sensors on the membrane, which membrane varies in stiffness along its length, respond to different frequencies transmitted down the spiral. The hair sensors are one of the few cell types in the body which do not regenerate. They can therefore be irreparably damaged by large noise doses. Refer to the Tinnitus FAQ for more information on associated hearing disorders.

At what level does sound become unsafe?

It is strongly recommended, to avoid unprotected exposure to sound pressure levels above 100dBA. Use hearing protection when exposed to levels above 85dBA (about the sound level of a lawn mower when you are pushing is over a grassy surface), and especially when prolonged exposure (more than a fraction of an hour) is expected. Damage to hearing from loud noise is cumulative and is irreversible. Exposure to high noise levels is also one of the main causes of tinnitus.

The safety aspects of ultrasound scans are the subject of ongoing investigation.

Health hazards also result from extended exposure to vibration. An example is "white finger", which is found amongst workers who use hand-held machinery such as chain saws.

What is sound intensity?

Sound intensity is expressed in decibels with respect to one pico-watt (10^-12 watts) per square meter. This is very nearly* numerically equal to the sound pressure level in decibels. This presumes no standing waves or reflections where the effective impedance can differ from that of free space air. In its complete form, intensity include the unit vector of the propagation direction, i.e. intensity is a vector quantity.

*For a plane wave, the sound power that passes through a surface of A square meters is defined as the ratio of the pressure squared to the air impedance

I = p^2/(rho*c)

When combined with the propagation unit vector, this defines the rate of sound energy transmitted in a specified direction per unit area normal to the direction. When measured in practical units, we can compute intensity after the relation that

Numerically, the sound intensity is related to the sound power as follows: In free air space, a source emitting Lw dB re 1 picowatt produces the sound pressure level Lp at a distance R feet as

Lp=Lw-20logR-0.6

At a one foot radius, that sound power is distributed over a surface of 4*pi = 12.57 square feet or (*.3048^2=.0920*) 1.17 square meters. 10log1.17=0.7dB. So within 0.1 dB, the coincidence exists that the sound intensity in picowatts per square meter is numerically equal to the sound pressure level in dB!

NOTE: This identity holds true only when the impedance, rho*c is exactly 400 mks rayls. This occurs for sea-level at 39 degrees C. For 22 C, rho*c = 412; a 0.13 dB difference arises. But at higher elevations, air density decreases for a given temperature. At an elevation of 840 feet above sea level, rho*c reduces to 400 at 22 C. (fortunate for much of Midwestern US!). The 0.13 dB difference at sea level is not usually significant for acoustical measurements.

Sound intensity meters are popular for determining the quantity and location of sound energy emission.

How does sound decay with distance?

At distances large compared to the size of the source, sound intensity diminishes according to the inverse square law.

I = Io/D^2

It is relatively simple to reliably calculate provided the source is small and outdoors, but indoor calculations (in a reverberant field) are rather more complex.

If the observation position is at a distance that is small compared to the size of the source, sound level changes very little with location. One should be able to determine the "virtual center" of the whole sound field, whence inverse square law calculations can proceed in reference to that distance.

The surrounding environment, especially close to the ground and in the presence of wind and vertical temperature gradients have a great effect on the sound received at a distant location. Ground reflection affects sound levels more than a few feet away (distances greater than the height of the sound source or the receiver above the ground). Wind and air temperature gradients affect all sound propagation beyond 100 meters over the surface of the earth.

If the noise source is outdoors and its dimensions are small compared with the distance to the monitoring position (ideally a point source), then as the sound energy is radiated it will spread over an area which is proportional to the square of the distance. This is an 'inverse square law' where the sound level will decline by 6dB for each doubling of distance.

Line noise sources such as a long line of moving traffic will radiate noise in cylindrical pattern, so that the area covered by the sound energy spread is directly proportional to the distance and the sound will decline by 3dB per doubling of distance.

Close to a source (the near field) the change in SPL will not follow the above laws because the spread of energy is less, and smaller changes of sound level with distance should be expected.

In addition it is always necessary to take into account attenuation due to the absorption of sound by the air, which may be substantial at higher frequencies. For ultrasound, air absorption may well be the dominant factor in the reduction.

What is the sound power level?

Sound power level, Lw, is often quoted on machinery to indicate the total sound energy radiated per second. It is quoted in decibels with respect to the reference power level. The reference level is 1pico-watt (pW) [1x10^(-12) watts]. One watt of radiated sound power is represented as "Lw=120 dB re one picowatt". If the reported sound power is in terms of A-Weighted spectral weighting, a suffix, A, is applied to form dB(A).

The sound pressure level (SPL) resulting from sound power (Lw) being radiated into free space, e.g. over a paved surface, is computed from

SPL = Lw - 20*log(R) - 11 dB re 20 uPa   (R in meters)
SPL = Lw - 20*log(r) - 0.7 dB re 20 uPa   (r in feet)

If instead the sound is emitted over a reflecting plane such as a hard surface, three (3) decibels are added to the SPL.

For example, a lawn mower with sound power level 100 dB(A) will produce at a sound pressure level (SPL) of about 89dB(A) at the operator (you) position over grass and 92 dB(A) when the mower is operated over a hard surface such as your driveway. At your neighbor's yard 50 feet (15m) away, the SPL will be is 65 dBA.

How is sound power measured?

Sound power is usually measured indirectly as the sound pressure level found at a specific distance, and in every direction that sound can be radiated. The sound power emitted by Items that can be carried to a laboratory is usually measured in a hemi-anechoic room or a reverberation room.

Either the "comparison" or the "direct" method is used.

In the comparison method, the SPL that the item causes in that room is compared the SPL created by a standard "Reference Sound Source" (see the 'Acculab' portion of this web page) to determine the sound power emitted by the item. This is the most common and economical method.

In the direct method two processes may apply. For the hemianechoic method, the SPL is measured in every direction on a surface encompassing the test item, then combined to compute the emitted sound power. For the reverberation room, the SPL is measured at several locations in the that room, then the sound power is computed from

PWL = SPL + 10Log(A)-C.

A = absorption in the reverberation room, sabins or square meters.
C = 16.3 for A as sabins (square feet)
C = 6.2 for A in square meters.

What is meant by loudness?

Loudness is the human impression of the strength of a sound. The loudness of a noise does not necessarily correlate with its sound level. Loudness level of any sound, in phons, is the decibel level of an equally loud 1kHz tone, heard binaurally by an otologically normal listener. Historically, it was with a little reluctance that a simple frequency weighting "sound level meter" was accepted as giving a satisfactory approximation to loudness. The ear senses noise on a different basis than simple energy summation, and this can lead to discrepancy between the loudness of certain repetitive sounds and their sound level.

A 10dB sound level increase is perceived to be about "twice as loud" in many cases. The sone is a unit of comparative loudness with

0.5 sone = 30 phons,
  1 sone = 40 phons,
 2 sones = 50 phons,
 4 sones = 60 phons etc.

The sone is inappropriate at very low and high sound levels where human subjective perception does not follow the 10dB rule.

Loudness level calculations take account of "masking" - the process by which the audibility of one sound is reduced due to the presence of another at a close frequency. The redundancy principles of masking are applied in digital audio broadcasting (DAB), leading to a considerable saving in bandwidth with no perceptible loss in quality.

What makes sound?
Sound is produced when the air is disturbed in some way, for example by a vibrating object. A speaker cone from a high fidelity system serves as a good illustration. It may be possible to see the movement of a bass speaker cone, providing it is producing very low frequency sound. As the cone moves forward the air immediately in front is compressed causing a slight increase in air pressure, it then moves back past its rest position and causes a reduction in the air pressure (rarefaction). The process continues so that a wave of alternating high and low pressure is radiated away from the speaker cone at the speed of sound.

What is vibration?

When something moves periodically about a static position it can be said to vibrate. Examples of unwanted vibration are the movement of a building near a railway line when a train passes, or the vibration of the floor caused by a washing machine or spin dryer. Floor vibration can be reduced with vibration isolators, sometimes at the risk of increased machinery vibration and subsequent deterioration.

How is vibration measured?

Vibration is often measured with an accelerometer. This is a device that is securely attached to the surface under investigation. The accelerometer produces an electrical charge proportional to the surface acceleration, which is then amplified by a charge amplifier and recorded or observed with a meter. The frequencies of interest are generally lower than sound, and range from below 1 Hz to about 1 kHz.

It is sometimes more useful to know the vibrational velocity or displacement. Often, moving coil transducers are used to directly measure vibrational velocity. A single integration of that signal provides a measure of displacement.

If only an accelerometer is available, it is necessary to integrate the acceleration signal once for velocity and twice for displacement. If the vibration is sinusoidal at a known frequency, f, then an integration is calculated by dividing the original by 2 x pi x f (noting that there is also an associated phase change).

Example: A machine is vibrating sinusoidally at 79.6 Hz with an rms acceleration of 10 m/s^2.
Its rms velocity is therefore 10/(2 x pi x 79.6) = 20 mm/s
Its rms displacement is 10/(4 x pi^2 x 79.6^2) = 0.04 mm

The final result may also be expressed in terms of zero-to-peak, which is found as the square root of two [sqrt(2)] times the rms value. The peak-to-peak value is twice again that.

Thus, one has three measures (acceleration, velocity, displacement) and three scales (rms, 0-p, p-p) totalling nine possible explicit measures of one and the same vibration. Couple that with three possible directions (E-W, N-S, up-down) one faces 27 separate possible values... and then there are inches, mils, microns and millimeters... Needless to say, one must be eternally vigilant and explicit in their vibration measurement and reporting nomenclature!

How is vibration isolated and controlled?

Vibration problems are solved by considering the system as a number of connected springs and masses with damping. The vibration source is included within, e.g. the engine of a motor car, or the environment on which this assembly is mounted is presumed to vibrate, e.g. a scanning electron microscope.

If the vibration is produced by a motor inside a machine, it is necessary that the natural frequency of the supporting system is well below frequency of motor oscillations (the forcing frequency). This is achieved by altering the mass or stiffness of the system as appropriate.

The method of vibration isolation is demonstrated with a weight held from a rubber band. If the band is moved up and down very slowly the suspended weight will move by the same amount. At resonance the weight will move much more and possibly in the opposite direction. But as the frequency of vertical movement is further increased, the weight will become almost stationary. Springs are more often used in compression than intension.

Important:-
Intuitive attempts to reduce vibration from machinery can sometimes instead aggravate the problem. This is especially true when care was originally taken to minimize vibration at the time of design, manufacture and installation.

Another method of vibration control is to cancel the forces involved using a Dynamic Vibration Absorber. Here an additional "tuned" mass-spring combination is added so that it exerts a force equal and opposite to the unwanted vibration. They are only appropriate when the vibration is of a fixed frequency.

Recently, "Active Vibration Control", using techniques akin to Active Noise Control has evolved. This senses the unwanted vibration of a structural member to produce a reversed phase signal to drive a transducer attached to the same member to counter the motion. In that way, for instance, the vibration of rolling wheels of a vehicle is prevented from being transmitted into the body of that vehicle through the chassis.

Architectural & Building Acoustics

What is reverberation time?

The time for sound in a room to decay 60 decibels. Scientific work on room acoustics was pioneered by Wallace Clement Sabine 1868-1919 (see his Collected Papers on Acoustics, 1922). The reverberation time, T, is defined as the time taken for sound energy to decay in a room by a factor of one million (60 dB). It is dependent on the room volume and the total amount of sound absorption contained in the room. In metric units

                              0.161 x room Volume
          T =  ----------------------------------------------
               sum of Surface areas x absorption coefficients

In US English units, dimensions are in feet and the constant is 0.049.

What is the sound absorption coefficient?

The absorption coefficient of a material is ideally the fraction of the randomly incident sound power which is absorbed, or otherwise not reflected. It is standard practice to measure the coefficient at the preferred octave frequencies over the range of at least 125Hz - 4kHz.

It can be determined on small material samples with an "impedance tube" or on large samples in a laboratory "reverberation room". The impedance tube evaluates sound absorption at normal incidence only, and produces absorption values that are sightly lower than those found in the reverberation room where the "Sabine coefficient" is measured over a wide range of incidence angles.

For the purposes of architectural design, the Sabine coefficient is preferred, though the normal incidence absorption may be used in the absence of any other information. Interestingly some absorbent materials are found to have a Sabine coefficient in excess of unity at higher frequencies. This is due to diffraction effects. Where this occurs the value can be taken at face value for small material patches and as 1.0 for very large absorbers (entire walls). The Odeon computer program includes a file of absorption coefficients.

What is the difference between insulation & absorption?

There is often confusion between sound insulation and sound absorption.

Sound insulation prevents sound from traveling from one place to another, such as between apartments in a building, or to reduce unwanted external noise inside a concert hall. Heavy materials like concrete are the most effective materials for sound insulation - doubling the mass per unit area of a wall will improve its insulation by about 6dB. It is possible to achieve good insulation over most of the audio frequency range with much less mass by instead using a double leaf partition (two separated independent walls).

Sound is absorbed when it encounters a material which will convert some or all of it into heat, or which allows it to pass through not to return. For this reason good sound absorbers do not of themselves make good sound insulators. Sound insulators rarely absorb sound. Sound absorbers contribute little to sound insulation. They are treated separately in sound control design.

How is sound insulation measured?

The measurement method depends on the particular situation. There are standards for the measurement of the insulation of materials in the laboratory, and for a number of different field circumstances. Usually

Test procedures (e.g. ASTM E-90 in the lab and E336 in the field) generate a loud and consistent broadband spectrum of steady noise on one side of a partition or specimen of the material under test, then measure the amount of this sound that passes through that material. The ratio of the incident sound to the transmitted sound is the "noise reduction", usually expressed as 10 time the logarithm of this ratio. If the noise reduction is also corrected for the amount of sound absorption to be found in the receiving room, 10 times the logarithm of the corrected ratio is called the "transmission loss. This is performed for 1/3 octave bands of noise from 100 to 4000 Hz.

A single-number rating of that range of noise reductions or transmission losses van be had by fitting them to a segmented curve.

In North America, this procedure is ASTM E413. The fitted range is from 125-4000 Hz. The value of that curve at 500 Hz is called the Noise Isolation Class (NIC) or Sound Transmission Class (STC) respectively. Internationally, ISO140-3 produces the noise reduction and transmission loss data in the same way. But the single number rating is according to ISO 717 which uses data in the 100-3150 Hz range. This single number rating is called "R'" and "R" respectively.

Similar methods are applied to impact ("footfall") noise (a problem in multifamily residential buildings). A standard tapping machine is used to hammer on the floor, lightly and steadily at the rate of 10 taps per second. The sound pressure level in the room below are measured.

How do I improve the noise insulation of my house/dwelling?

This is one of the most commonly asked questions of noise consultants. Firstly you should consider whether it is noise insulation or sound absorption (see 4.3) that is really required. Sound insulation is most often asked for in order to keep out unwanted noise, but is occasionally requested for the purpose of minimizing disturbance to others.

The method of noise insulation will depend on the exact situation; generalities are extremely difficult to devise. Situations are more often than not unique, depending on the nature of the building infrastructure that the architect or his informal successors have devised. More often than not, successful noise isolation improvement requires the advice of a competent and experiences person and at an early stage of the renovation. The following ideas may serve as initial guidelines.

When the noise is from an external source such as a main road it may be possible, if planning authorities permit, to screen with a noise barrier. These can be effective providing that the direct line of sight between traffic and house is concealed by the barrier.

The weak point for sound transmission to and from a building is most often via the windows. Double glazing will usually afford noticeably better protection than single glazing, but in areas of high external noise it might be preferable to have double windows with a large air gap (25 to 100 mm) and acoustic absorbent material on the perimeter reveal around that gap. For a few people, the resultant lower room background noise level can make noise transmitted through party walls more apparent. The fitting of new windows may reduce the level of air ventilation, and it will be vital to compensate for this, if necessary with by improving the noise insulation of certain party walls.

Noise through party walls can be reduced by the addition of a false wall. This is constructed from a layer of sound insulating material, commonly plasterboard, separated from the party wall by a large void containing acoustic quilting. The false wall must not be connected to the party wall because that would allow sound transmission paths. The quality of construction is an important consideration if optimal levels of attenuation are desired. It is advisable to contact an independent noise consultant before allowing any building works to commence.