On Diwali night one cracker bursts in the street, and then a string of ten crackers bursts together. The string delivers about ten times the intensity to your ears. Yet it does not sound ten times as loud. It sounds perhaps twice as loud. Your ear is not measuring intensity the way a meter does.
The ear hears ratios
Our ears perceive loudness logarithmically, not linearly. A sound 100 times more intense sounds louder by only a few times. What the ear notices is the ratio of two intensities, not their difference. Going from 1 to 10 units of intensity feels like the same step as going from 10 to 100.
This is how the ear copes with an enormous range. Lesson 5 shows that we hear intensities from \(10^{-12}\,\text{W/m}^2\) to about \(1\,\text{W/m}^2\), a range of a million million. An ear that responded in proportion to intensity could not span that range. An ear that responds to ratios needs only twelve steps of ten.
Listen to Fig 4.1 and change the chip. The pressure wave in the figure grows by \(3.16\) at each step, and the intensity by \(10\). Yet each step sounds only about twice as loud as the one before. The step of \(20\,\text{dB}\) multiplies the intensity by a hundred and sounds only about four times as loud.
Gustav Fechner, a German physicist, studied how strongly people feel a stimulus. Earlier, Ernst Weber had found that a weight must grow by a fixed fraction, not a fixed amount, before a person notices the change. In his Elements of Psychophysics of 1860, Fechner turned this into a law. The strength of a sensation grows as the logarithm of the stimulus. The decibel scale of sound is built on the same idea.
The decibel scale
To model the ear, we use the decibel (dB) scale. It measures the sound level \(\beta\) of a sound by comparing its intensity \(I\) with a fixed reference intensity \(I_0\). The reference is \(I_0 = 1.0 \times 10^{-12}\,\text{W/m}^2\), the threshold of hearing of a healthy young ear.
The sound level \(\beta\) of a sound of intensity \(I\) is \(\beta = (10\,\text{dB})\log_{10}(I/I_0)\), where \(I_0 = 1.0 \times 10^{-12}\,\text{W/m}^2\) is the threshold of human hearing. The unit is the decibel, dB.
The logarithm is a ratio in disguise. At the threshold of hearing \(I = I_0\), so \(\beta = 10\log 1 = 0\,\text{dB}\). A level of \(0\,\text{dB}\) is not silence. It is the faintest sound the ear can hear. The name comes from the bel, named after Alexander Graham Bell, the inventor of the telephone. A decibel is a tenth of a bel.
Move the slider in the figure above. On the left column, every step of one power of ten raises the level by the same \(10\,\text{dB}\). On the linear column nothing shows until the level passes about \(120\,\text{dB}\). At \(120\,\text{dB}\) the intensity is \(1\,\text{W/m}^2\), only one hundredth of the column. Every sound we meet in daily life is squeezed into the bottom of a linear scale.
Useful benchmarks
The logarithm turns a ratio of intensities into a difference of levels. So three results are worth remembering.
- An increase of 10 dB means the intensity has multiplied by 10.
- An increase of 20 dB means the intensity has multiplied by 100.
- An increase of 3 dB means the intensity has approximately doubled, since \(10\log_{10}2 = 3.01\).
If the intensity of sound increases by a factor of 100, the sound level increases by
True or false: a sound at 70 dB has twice the intensity of a sound at 35 dB.
A sound wave has an intensity of \(2.0 \times 10^{-6}\,\text{W/m}^2\). What is its sound level in decibels, \(\beta\)?
Use the formula for sound level, \(\beta = (10\,\text{dB})\log_{10}(I/I_0)\), where \(I_0 = 1.0 \times 10^{-12}\,\text{W/m}^2\).
Put in the values: \(\beta = 10\log_{10}\left(\dfrac{2.0 \times 10^{-6}}{1.0 \times 10^{-12}}\right) = 10\log_{10}(2.0 \times 10^6)\).
The logarithm of a product is the sum of the logarithms: \(\beta = 10\left[\log_{10}2.0 + \log_{10}(10^6)\right] = 10\,[0.30 + 6]\).
So \(\beta = 10 \times 6.30 = 63\,\text{dB}\). Final answer: \(63\,\text{dB}\).
Set the slider to \(-5.7\) to check the example: the readouts show \(2.0 \times 10^{-6}\,\text{W/m}^2\) and \(63\,\text{dB}\). The intensity of normal conversation in lesson 2, \(1.0 \times 10^{-6}\,\text{W/m}^2\), is exactly \(60\,\text{dB}\). The example's sound is twice as intense, so its level is \(3\,\text{dB}\) higher.
The whistle of a pressure cooker produces an intensity of \(5.0 \times 10^{-3}\,\text{W/m}^2\) in a small kitchen. What is its sound level, in dB?
Comparing two levels
To compare two sounds, subtract their levels. The reference \(I_0\) cancels, because a difference of logarithms is the logarithm of a ratio. So \(\beta_2 - \beta_1 = (10\,\text{dB})\log_{10}(I_2/I_1)\). Every \(10\,\text{dB}\) of difference is a factor of \(10\) in intensity.
A rock concert has a sound level of 110 dB, while a quiet conversation is 50 dB. How many times more intense is the concert than the conversation?
Adding sources
Two sounds that are not in step add their intensities, not their levels. So two identical sources give twice the intensity of one. The level rises by \(10\log_{10}2 = 3\,\text{dB}\), whatever the level of one source. Ten identical sources give ten times the intensity, \(10\,\text{dB}\) more.
At a traffic signal, one auto-rickshaw idling beside you gives a sound level of \(80\,\text{dB}\). A second, identical auto-rickshaw stops at the same distance. What is the sound level of the two together?
The two engines are not in step, so their intensities add. Each alone gives \(80\,\text{dB}\).
One auto gives \(I_1 = I_0 \times 10^{80/10} = 10^{-12} \times 10^{8} = 1.0 \times 10^{-4}\,\text{W/m}^2\). Two give \(2I_1 = 2.0 \times 10^{-4}\,\text{W/m}^2\).
\(\beta = 10\log_{10}(2.0 \times 10^{-4}/10^{-12}) = 10\log_{10}(2.0 \times 10^8) = 10 \times 8.30 = 83\,\text{dB}\).
The level is \(83\,\text{dB}\): doubling the intensity adds \(3\,\text{dB}\). The ear hears the pair as only a little louder than one.
Which of these statements about sound levels are true?
Now return to the Diwali street. The string of ten crackers delivers ten times the intensity of one. On the decibel scale that is only \(10\,\text{dB}\) more. The ear judges a step of \(10\,\text{dB}\) as about twice as loud. So the ten crackers sound roughly twice as loud as one, just as you noticed.
- Treating decibels as proportional to intensity.The scale is logarithmic. 70 dB is \(10^{3.5} \approx 3162\) times as intense as 35 dB, not twice.
- Adding the levels of two sources.Add the intensities. Two identical sources add only \(3\,\text{dB}\).
- Reading \(0\,\text{dB}\) as silence.\(0\,\text{dB}\) is the threshold of hearing, \(I = I_0 = 10^{-12}\,\text{W/m}^2\).
- Forgetting to divide by \(I_0\) before taking the logarithm.\(\beta = 10\log_{10}(I/I_0)\). The ratio \(I/I_0\) has no unit.