Skip to content

LESSON 82

Masking and the critical band

Why does noise make a sound inaudible? In this lesson we look at psychoacoustic masking, Fletcher's critical band experiment, auditory filters, why low-pitched sounds mask high-pitched sounds more, known as the upward spread of masking, forward and backward masking, and informational masking.

Watch on YouTube

Follow along

Hello, and welcome to the AIHEAR Audiology Academy.
In this lesson we will learn how one sound makes another sound inaudible, that is, psychoacoustic masking.
Step by step, we will see the critical band, auditory filters, the upward spread of masking and temporal masking.
Masking is the rise in the hearing threshold of one sound in the presence of another sound.
The covering sound is called the masker, and the sound that becomes harder to hear is called the signal.
The amount of masking is found by subtracting the signal's threshold in quiet from its threshold measured with the masker present.
For example, if a tone is heard at ten decibels in quiet and at forty decibels in noise, the amount of masking is thirty decibels.
In daily life, this is the main reason we find it hard to follow speech in a noisy café.
In clinical masking, we use the same principle on purpose, keeping the opposite ear busy with noise.
We divide masking into three types by the timing of the masker and the signal.
In simultaneous masking, the signal and the masker are present at the same time; it is the strongest and most studied type.
In forward masking, a signal arriving shortly after the masker ends is harder to hear.
In backward masking, the signal comes shortly before the masker and is still affected.
Forward and backward masking together are called temporal masking.
Let us start with simultaneous masking, because the concept of the critical band was born here.
The concept of the critical band is based on experiments published by Harvey Fletcher in nineteen forty.
In the experiment, the threshold of a tone of fixed frequency, for example one thousand hertz, is measured in noise.
The noise is first given as a narrow band around the tone; the noise level per hertz is kept constant.
As the band widens, the total noise energy increases and the tone's threshold rises too.
But once the band goes beyond a certain width, the threshold no longer rises, even though the noise keeps increasing.
This limiting width is called the critical band; only the noise inside this band masks the tone.
This finding rests on the cochlea: the base of the basilar membrane responds to high frequencies and the apex to low ones.
A one thousand hertz tone stimulates most strongly its own region, where the traveling wave peaks.
Each region behaves like a filter that passes only a certain frequency range; this is called an auditory filter.
The critical band is the effective width of this auditory filter.
When hearing a tone, the ear listens to the output of the filter centered on that tone; noise outside the filter adds little to this output.
So the noise that masks a tone most efficiently is a narrow band around that tone.
Filter width increases with frequency; today it is most often expressed as the equivalent rectangular bandwidth, or ERB.
The table gives approximate values for young normal-hearing adults, from the nineteen ninety formula of Glasberg and Moore.
Although the filter looks narrow at low frequencies, it is wider relative to the center frequency.
At one thousand hertz, the ERB is about one hundred thirty three hertz, about thirteen percent of the center frequency.
At high frequencies the width rises to hundreds of hertz, but the ratio stays around eleven to twelve percent.
Zwicker's classic critical band scale gives somewhat wider values and divides the hearing range into about twenty four bands.
Today the shape of the auditory filter is usually measured with the notched-noise method; the area around the tone is cleared of noise.
As the notch widens, less noise falls into the filter and the tone's threshold drops.
From how the threshold changes with notch width, the width and shape of the filter are calculated.
At moderate levels the filter is roughly symmetrical; at high levels, its low-frequency side becomes broader.
Another method is the psychophysical tuning curve, which plots the masker level needed at each frequency to mask a fixed, soft tone.
In a normal ear this curve is sharp and V-shaped, and it shows the frequency selectivity of the ear.
This is why, in pure-tone audiometry, we mask the opposite ear with narrowband noise centered on the test frequency.
Energy falling outside the critical band does not add to masking; it only makes the sound louder and more annoying.
Narrowband noise gives the same masking effect with a lower overall loudness.
The narrowband noise in audiometers is kept a little wider than the critical band to cover differences between people.
In cochlear damage, auditory filters broaden, so the same noise masks the signal more easily in the damaged ear.
We will return to this in the lesson on how cochlear damage changes perception.
Masking is not symmetrical: a low-pitched sound masks a high-pitched sound much more strongly than the reverse.
The reason is the structure of the traveling wave; the wave always moves from the base of the cochlea toward the apex.
A five hundred hertz sound passes the high-frequency regions at the base before reaching its own region, and sets them vibrating too.
After the wave peaks, it dies out quickly, so it hardly spreads to lower-frequency regions.
As a result, the effect of a low-frequency masker spreads toward high frequencies; this is called the upward spread of masking.
This asymmetry was first shown in nineteen twenty four, in two-tone experiments by Wegel and Lane.
The effect of a low-frequency masker spreads over a wide area above its own frequency.
As the masker level rises, this spread grows disproportionately, because the filter's low-frequency side broadens at high levels.
So sounds rich in low frequencies, such as engine noise, air conditioning hum and crowd noise, are very effective maskers.
The effect of a high-frequency masker, by contrast, stays in a narrow area around its own frequency.
A high-frequency masker changes the threshold of lower sounds very little.
High-pitched sounds such as whistling or hissing can be annoying, but they hardly cover the low-frequency parts of speech.
In speech, strong, low-frequency vowels can partly mask the weak, high-frequency consonants that follow right after them.
In a noisy place, low-frequency noise covers the high-frequency consonants that carry the intelligibility of speech.
This explains why speech in noise breaks down first in the consonants.
In hearing aids, too much low-frequency gain in noisy places can cover high-frequency information by the same mechanism.
So many hearing aids offer programs and settings that reduce low-frequency gain in noise.
In cochlear damage, the filters broaden, so the upward spread becomes even more marked.
Let us think of temporal masking on a timeline; the strongest effect is seen while the masker is on.
The masker's effect does not vanish when it ends; forward masking fades over about one hundred to two hundred milliseconds.
The closer the signal is to the end of the masker, and the louder the masker, the greater the forward masking.
Backward masking occurs over a much shorter interval, only a few tens of milliseconds, and its amount is smaller.
Backward masking decreases clearly with practice, so it is linked more with central processing.
Forward masking, by contrast, partly reflects the recovery time of the cochlea and auditory nerve after a stimulus.
The masking we have described so far comes from energy overlapping in the same auditory filters; this is called energetic masking.
But sometimes a signal cannot be understood even though it is clearly represented in the cochlea.
For example, when two people talk at once, the second talker's words get mixed with the target speech; this is called informational masking.
Informational masking relates to central processes such as attention, segregation and uncertainty.
It decreases when the target and the masker come from different directions, or when their voices are clearly different.
Children and older adults are usually more affected by informational masking.
A tip from the clinic: if only the low-frequency thresholds on an audiogram are unexpectedly poor, think of ambient noise.
Ambient noises such as ventilation, equipment fans and traffic are mostly rich in low frequencies.
This noise masks the test tone simultaneously and raises thresholds, especially at two hundred fifty and five hundred hertz.
Make sure the booth door is closed; if in doubt, insert earphones attenuate outside noise better.
Test room noise should be measured regularly to make sure it does not exceed the maximum permissible levels in the standards.
Masking is the rise in the threshold of one sound in the presence of another; its amount is measured in decibels.
Only noise inside the auditory filter, the critical band, masks a tone; that is why narrowband noise is used in the clinic.
Because of the structure of the traveling wave, low sounds mask high sounds more; this is called the upward spread of masking.
Forward masking lasts about one hundred to two hundred milliseconds; backward masking is much shorter and weaker.
Test yourself: at the same level, in which case is masking greater?
The correct answer is B: five hundred hertz noise masks a four thousand hertz tone more.
The traveling wave of a low sound passes through the high-frequency regions before reaching its own; this is the upward spread of masking.
In the next lesson we will learn how the ear processes time: gap detection and temporal integration.
Subscribe to the channel so you never miss a new lesson. See you next time!

Keywords

Click a keyword to search all lessons for that term.

Prerequisites

LESSON 81 Pitch Perception and Frequency Discrimination Coming soon
LESSON 40 Noise types

Comments

Questions and comments appear after approval.

No comments yet. Be the first to comment.