Measuring Loudness

Loudness
Loudness is a complex subjective experience related to both the intensity and the frequency of the sound. Much research has been performed over the years to develop loudness indices, two early attempts being the phon and the sone.

Loudness
The phon was developed by experiments which used pure tone sound signals of fixed frequency and amplitude. In each test the participant presented a 1000 Hz pure tone sound as a reference, then the sound frequency was changed and the participant was asked to adjust the amplitude of the new signal until it was of equal loudness. By performing the test many times with different frequencies and different people it was possible to generate a set of equalloudness curves.

Loudness
From the equal loudness curves it can be seen that human perception of loudness varies as a function of frequency. Humans are particularly sensitive to frequencies in the range from 1000 to 6000 Hz.

Loudness
The phon was designated the unit of loudness and was set equal to the decibel level of the 1000 Hz reference tone. For example, all tones judged to be of equal loudness to the 60 dB reference tone are designated as having a loudness of 60 phons.

Loudness
The frequency weighting networks used in sound level meters are based on the phon curves developed by Fletcher and Munson. The A and B frequency weightings are the 40 and 70 phon contours, but with some minor modifications to simply the required electrical filter network.

Loudness
Phon curves provide information about the equivalence of sounds, but not about the absolute level of perceived loudness. We cannot say, for example, how many times louder a 40 phon sound is with respect to a 20 phon sound. Fletcher and Munson therefore performed further tests with a rating scale which was later named the sone. One sone is defined as the loudness of a 1000 Hz tone of 40 dB (40 phons). A sound which is judged to be twice as loud as the 1000 Hz standard reference tone has a loudness value of 2 sones, a sound judged three times as loud is 3 sones, etc..

Loudness

The graph presents the relationship between the level in phons and the perceived loudness in sones for pure tone sounds. The perceived loudness grows rapidly with increasing sound pressure, particularly at lower levels.

Loudness

The graph above gives an approximate indication of the sone values of some typical sounds from everyday life.

Stevens Loudness
The loudness of broadband sounds can be estimated by means of the Stevens Loudness Method (ISO 532A). In this method the sound energy is first divided into octave or 1/3 octave bands. A loudness value for each band is then determined by means of a loudness nomogram. The total loudness is then determined from the individual band values by means of a summation formula. The formula takes acoustic masking into account by weighting the loudness of the band with the greatest value about three times as much as the other bands.

Stevens Loudness

The loudness nomogram used in the Stevens method provides a loudness index which is the relative loudness of a given octave or third octave random noise sound to that of a reference octave or third octave random noise band centred on 1000 Hz.

Stevens Loudness
The summation formula for obtaining the total sone loudness value from the sone values of the individual bands is

Zwicker Loudness
Another method for estimating total loudness is the Zwicker method. Like the Stevens method, the Zwicker method is based on the use of octave or 1/3 octave band analysis of the sound signal.

Zwicker Loudness
The Zwicker loudness procedure is more complex than the Steven loudness procedure because masking effects are evaluated at each stage. Masking occurs when a sound is not heard due to the presence of an intense sound at a nearby frequency. For example, a 90 dB tone at 1200 Hz will completely mask a 50 dB tone at 4000 Hz.

Annoyance

One term that is sometimes used to describe the effects of unwanted sound is annoyance. Annoyance is a subjective quantity associated with the inappropriateness or unwantedness of the sound. It is important to note that the loudness value of a given sound is only weakly correlated with its annoyance.

Annoyance

Annoyance can be caused by sounds that are too intense with respect to the environment in which they occur. Even low intensity sounds can, however, cause annoyance when they are unexpected or unusual. For example, even if low in amplitude, a pure tone sound from a fan or air conditioner can produce great annoyance if it is not expected or not wanted in a specific room.

Auditory Environment

Auditory Environment

When designing for people the acoustic environment should: avoid hearing loss minimise sounds related to annoyance and stress minimise the disruption of speech communications transmit desired sounds reliably and pleasantly to the listeners

Sound Level
The A-weighted Sound Pressure Level LA is defined as

LA

p A (t ) = 10 Log10 preferecne

dB

Where pA(t) is the instantaneous sound pressure measured using the standard A scale frequency weighting shown below.

Sound Level
The Average Sound Level Lav,T is defined as

Lav ,T

= 10 Log10

1 T

p (t ) dt
2 reference

dB

where T is the time over which the measurement is averaged. The Average A-weighted Sound Level LA,T is defined as
1 2 p A (t ) dt T 0 = 10 Log10 dB 2 preference

LA,T

Sound Level
The Day-Night Sound Level Ldn is defined as

22:00

7:00 2 A

Ldn

1 7:00 = 10 Log10 + 22:00 2 2 24 preference preference

p (t ) dt

10

p (t ) dt

2 A

dB

where the first term covers the daytime hours from 7:00 to 22:00 and the second term covers the night time hours from 22:00 to 7:00. The night levels are taken to be 10 dB more than they actually measure.

Sound Level
The A-weighted Sound Exposure EAT is defined as

E AT

p (t ) dt

2 A

[Pa s ]
2

While the A-weighted Noise Exposure Level LEA,T is defined as

LEA,T

= 10 Log10

E A,T E0

dB

Where E0 is a reference value normally taken to be (20Pa)2s which is (4x10-10Pa)2s

Effects of Noise on Human Performance

A little noise in the work environment is beneficial as it serves to increase arousal, which can produce improved task performance. Beyond a certain level of intensity the task performance begins to degrade. Sudden, unexpected, noise can produce a startle response which interrupts concentration and task performance. Continuous noise normally reduces performance on complex tasks such as visual tracking. The reduction is proportional to the noise level. Psychological effects of noise may include anxiety, helplessness, narrowed attention and other adverse effects that degrade task performance.

Effects of Noise on Human Performance

Criteria for Speech Communication

Acoustic design often seeks to achieve environmental noise levels that are low enough to permit acceptable speech communication between individuals. The design methods used for speech can also be applied to music since the frequencies involved are similar. For example, the tonal range of the violin is from 200 to 8000 Hz.

Articulation Index
Speech intelligibility requires that the listener receive sound that is above the threshold of hearing and below the region of hearing overload. Speech intelligibility also requires that the speech signal be stronger than the background noise present in the environment.

Articulation Index
The hearing threshold, the overload region and the typical speech region for a male raised voice at 1 metre distance are presented below as a function of frequency.

Articulation Index
An articulation index of 100 percent corresponds to a situation where the spectrum levels of speech at the listeners ear lie above the threshold of hearing and below the overload line. The speech spectrum levels must also be above the spectrum of the background noise. If on the other hand the noise spectrum covers part of the shaded speech region, or if part of the speech region falls below the threshold curve, the articulation index is less than 100 percent.

Articulation Index
For an articulation index of 0.6 or more conversation will be satisfactory, while for an AI of 0.3 or less the speech communication will be unsatisfactory. There are, however, other factors influencing word intelligibility therefore the AI cannot be considered the absolute measure of the acoustic environment. An example of the relationship between AI and speech intelligibility is provided by the graph below developed by NASA.

Speech Interference Level

The Speech Interference Level (SIL) is a simple means of quantifying the effect of background noise measured at the location of the listener. The American National Standard definition of the SIL is Speech Interference Level (SIL) in decibels is the arithmetic average of the sound pressure levels of the interfering noise re 20 Pa in the four octave bands centred on the frequencies 500, 1000, 2000 and 4000 Hz.

Speech Interference Level

A NASA document provides a table for interpreting Speech Interference Levels.

Speech Interference Level

A NASA document provides a graph for interpreting Speech Interference Levels.

Speech Communication Criterion Curves

Criterion curves were developed using the speech interference level (SIL). The curves assume that the talker and listener are located in a free field, and give the permissible SIL at the listeners position as a function of the talker-listener separation r and the talkers voice level. The curves were developed for an AI of 0.5 which corresponds to a monosyllabic word intelligibility of at least 85%, and were developed separately for men and women since womens voices are about 4 dB weaker.

Non-Verbal Auditory Warnings

Non-Verbal Auditory Warnings

Non-Verbal Auditory Warning refers to the transmission of information by means of bells, buzzers and other simple sounds. Examples include the buzzers used on medical monitors and the various warning sounds used to signal emergencies in buildings and in vehicles.

Non-Verbal Auditory Warnings

Although visual displays are able to transmit more information, there are many situations in which an auditory warning is preferable. For example when The operator is mobile and hence not able to see a visual warning given by means of symbols or phrases. The operator is in a high mental workload environment in which visual signals are likely to be missed. An economical means of conveying simple information to a large number of people is sought.

Non-Verbal Auditory Warnings

There are many situations in which verbal auditory warnings are not effective. For example: In a hospital the use of verbal warnings which can be overheard by the patient may be insensitive. In many emergency situations there is not enough time to convey the problem through speech signals. In many work environments there are high levels of background noise which make speech signals difficult to interpret correctly.

Non-Verbal Auditory Warnings

There are two main requirements for an effective non-verbal auditory warning system: It needs to be heard, but should not be too loud. It needs to be psychologically appropriate.

Non-Verbal Auditory Warnings

Setting the sound level of auditory warnings requires knowledge of the background noise levels for the environment. Once the background levels are known the warnings can be set to levels such as 15-25 dB above the background which are high enough to not be masked. As an example the graph below presents data for the flight deck of a Boeing 727 aircraft. The lower lines are measured background noise spectra in various flight conditions. The range of appropriate levels for auditory warnings is given by the shaded region at the top.

Non-Verbal Auditory Warnings

15-25 dB above the background level is a typical design guideline for auditory warning. Several problems occur of warnings are set at too high: There is the potential for hearing loss if the warnings are activated frequently. Loud warnings lead people to switch them off and not turn them back on again. In many environments a loud warning may disturb people. This is especially true in a hospital setting where patients may already have a low tolerance for noise due to illness. Loud warnings tend to be turned off before the situation they are signalling is attended to. In many situations the time required to turn off the alarm may be critical. Loud warning sounds produce startle reactions which hinder concentration at a time when this may be important. Loud warning sounds hinder verbal communication.

Non-Verbal Auditory Warnings

Some sounds are more appropriate than other for indicating a particular situation. If a sound can be chosen which helps the listener to understand the nature of the warning, learning times are minimised and confusion avoided. The table below lists some potential sources of confusion in auditory warnings.

Non-Verbal Auditory Warnings

Some traditional design guidelines are: Use frequencies between 200 and 5000 Hz, especially from 3000 to 5000 Hz since the ear is most sensitive in this middle range. Use frequencies below 1000 Hz when signals have to travel long distances (more than 200 metres) because high frequencies are rapidly dissipated in air. Use frequencies below 500 Hz when signals have to bend around obstacles or pass through partitions. Use modulated signals (1 to 8 beeps per second) since they are different enough from natural sounds to demand attention. Use signals with frequencies different from those of any dominant background noise to minimise masking. If different warning signals representing different conditions are used, each sound should be identifiable from the others. Where possible, use a separate communication system for the warnings. Examples include horns, bells and loudspeakers.

Non-Verbal Auditory Warnings

Several studies have mapped the perceived urgency of warning sounds to their frequency content and temporal pattern. In one study the Stevens Power Law was used to determine urgency exponents for the speed, number of repetitions and frequency of the warning signal. Speed Number of Repetitions Frequency 1.35 0.50 0.38

The higher the value of the exponent, the more quickly the subjective assessment of urgency changed as the parameter increased.