CDO 750, Lecture 8-- Perception of Loudness & Intensity Discrimination
For 40 phons, the equivalent sones is 1. For 50 phons, what's the equivalent in sones?
2 sones. What about 60 phons? 4 sones. So a 10 phon increase doubles loudness, or sones
What's an equal loudness contour?
A curve depicting the intensity level at which all sounds have the same perceived loudness.
more about the phon lines
As intensity increases, the curves get flatter, this means that there's less variation of loudness across frequency at high intensity--- this is for the really intensity sounds at the top left corner of the graph
What's the standard for loudness?
For loudness measures, the standard is a 1000 hertz tone.
So if you use a regular scale of sones on y axis, what will it look like with phons on x-axis?
It will look logarithmic. Sones grows logarithmically in relation to phones
Loudness is affected by the following parameters
The complexity of the signal, tones versus noise. Also, the frequency, overall level, and duration
What does the bottom contour line indicate?
The minimum audible field, or MAF.
What is the loudness balance procedure and why is it used?
Used to estimate loudness with a change in frequency. Present subject with 1000 hertz tone at a fixed loudness level. Present another frequency tone at a fixed level and ask them to adjust until they perceive it as equally loud as the 1000 hertz tone. I suppose this is simple matching. This creates an equal loudness contour.
so the worst DL is witnessed where?
low frequency, low intensity, The same group where we saw greatest loudness growth. This is confusing...
When you measure loudness in sones, you are assigning an arbitrary value. How are sones measured?
using direct measurement procedures like magnitude or ratio production
As a review, prothetic continua are quantitative sensations that vary in the magnitude of the sensation. e.g., the loudness of sound
...
Basically, delta I is the smallest perceived increment, divided by I which is the original magnitude, equals K, where K is a constant.
...
But, two people could have different perceptions of whether that 100 hertz tone sounds equal in loudness at those levels, but overall, most people will show a similar trend.
...
For noise, weber's fraction is about constant
...
For recruitment, the expected loudness growth eventually returns at some intensity level, but below that, you get more loudness growth than expected
...
I guess 1 phon is a 1000 hertz done at 1 dB SPL
...
What could you call each curve on the equal loudness contour?
A phon line. But this is slightly confusing because I thought equal loudness curves were determined by asking the subject to complete a matching procedure. I guess matching is actually an indirect measurement.
Understanding some terminology. What is a standard?
A signal to which others are compared.
What does Weber's law state?
A stimulus must be changed by a constant percentage of its initial value in order for this change to be perceived.
What's a phon?
A unit of loudness level. Emphasis on level. Specifically, it's the intensity level in dB SPL of a 1000 hertz tone that is perceived as being equally loud as a variable tone.
What's a sone?
A unit of loudness. 1 sone is equal in loudness to a 1000 hertz tone at 40 dB SPL. So 1 sone is equal to 40 phons.
Elaborate
Absolute would just be one intensity minus the other, or change in intensity. The value would not be reported in dB but rather watts per meter squared. Relative difference would be the change in intensity, or absolute difference, divided by the original intensity.
Examples of phon comparisons
All tones that are considered equally as loud as a 60 dB SPL 1000 hertz tone have a loudness of 60 phons. If you adjust a 100 hertz tone to 65 dB SPL and it sounds equally as loud as a 1000 hertz tone at 60 dB SPL, then the 100 hertz tone has a loudness of 60 phons.
The loudness level of any sound in Phons is the intensity level of what?
An equally loud 1000 hertz tone.
Where else does the doubling rule not play?
At very low levels around threshold. Less increase in intensity is needed to produce an increase in loudness than at higher sensation levels.
Regarding very low intensities, where is difference limen best?
Best at mid frequencies, then high frequencies, then worst for low frequencies.
What's one more way to express the relationship between the intensity of a standard and a varying stimulus?
Change in level in dB. This is found by computing 10 log of one plus Since dB is a ratio, then a change in dB is also a Weber fraction. Not sure I need to know this
How is the perception of loudness related to the actual stimulus intensity?
Changes in intensity are highly correlated with changes in loudness.
What do we know about difference limen for different intensities?
Difference limen increases for levels below 40 dB SL. So for soft sounds, greater changes in intensity are needed to be noticed. Difference limen is smaller for moderate and loud sounds, and smallest for really intense sounds. So for really loud sounds, smaller changes are needed to tell a difference.
What are other names for this?
Different limen, DL, or just noticeable difference, or JND. So DL and JND are indirect measurement
Does Weber's law do a good job of describing perception of pure tones?
For pure tone intensity discrimination, the law is only a first order approximation. There's lots of variability among studies, and some have found more of a miss than others.
When the neurons are saturated, how do we differentiate different intensity?
For two sounds where one is slightly more intense, the slope representing how much of the high frequency area of the basilar membrane is excited tapers off at a slower rate, meaning a slightly larger area of HF is excited, even though the maximum excitement is no louder than for the slightly lower intensity sound, since both are at neuronal saturation.
Other problems with this?
For very low intensities, the weber fraction increases, and tends to decrease as sensation level increases
How can difference limen be reported?
In absolute or relative terms
Also, that as frequency increases in the low frequencies, there are large changes in perceived loudness.
In the mid frequencies, loudness is pretty consistent across frequencies.
If phons are a measurement of loudness level, are they determined using direct or indirect measures?
Indirect which deals with discrimination and whether things are same, different, or present and not present. This reflects ordinal scaling, meaning an ordinal scale can tell you that the sounds are different or the same but not how different. So, phons can only be ranked and you cannot add, subtract, multiply or divide them.
Intensity is not equivalent to loudness. What are they each measuring?
Intensity is a physical measurement while loudness is a psychological measurement.
How does a sound that is 3 sones compare to a 1000 hertz tone at 40 dB SPL?
It is 3 times as loud as that tone.
What is the dBA scale on a sound level meter based on?
It's based on a filter that is shaped like the 40 phon line.
For mid frequency range, 500 to 4 khz, what's the effect on loudness, or sones, for a every 10 dB increase in intensity level? I suppose a 10 dB intensity level increase is analogous to a 10 phon increase, right?
Loudness doubles.
As a review, what's true about phons and sons and whether they can be added or not?
Phons is a unit of loudness level, while sones is a unit of loudness. Phons, unlike sones, cant be added, subtracted, multiplied, or divided. They can only be ranked. For example 80 phons bay be louder than 40 phons, but 80 phons is not necessarily twice as loud as 40 phons so you know it's louder but not how much louder. We will see this later on with the doubling of loudness, or sones, with only a 10 phon increase
Review of magnitude production
Present with standard sound and variable sound. Subject adjusts the volume of the variable until it reaches a predetermined loudness estimate. In this example, I suppose you would play a sound and tell them it's at a loudness of 20 or so, and then present them with the variable sound and have them adjust it until it's a 40, or a 10 or something.
So explain that.
Prothetic continua are quantitative sensations that vary in the magnitude of the sensation, for example, the loudness of the sound
What is Steven's Power law?
Prothetic sensations follow power functions of the physical stimuli.
R=k x I raised to the beta
R is the magnitude of the sensation, k is a constant that depends on the type of stimulus, I is the magnitude of the stimulation, and B is based on the perception under consideration.
Measuring loudness
Remember, as we mentioned, use scaling, or cross-modality matching or simple matching. You have to pick a standard.
What types of measurement procedures should we use to study loudness?
Scaling procedures, specifically ratio estimation and ratio production, and magnitude estimation and magnitude production, and also cross-modality matching, another type of scaling procedure. Or we can use simple matching.
What's the standard for loudness testing and how's it done?
Set a 1000 hertz tone at a fixed intensity and have the listener adjust different frequency tones to be equally as loud as the standard. This is a matching procedure.
These things are true above about 20 or 30 dB SPL, but what happens around threshold?
Smaller changes in loudness level are needed to perceive a doubling of loudness.
What's more difficult, studying subjective perceptions or measuring absolute sensitivity?
Subjective perceptions. We can determine the limits of the sensitivity of the auditory system by measuring intensity difference limens, but how do we measure the perception of loudness as it relates to intensity?
What can you learn from the equal loudness contours graph?
That at low frequencies, a smaller change in intensity leads to a bigger loudness change. Also, very intense sounds equate to low loudness perception, meaning out hearing is less sensitive there. It is most sensitive at mid frequencies and less so at low and high frequencies. Contour lines are pretty evenly spaced apart at high frequencies, but elevated.
But...what happens?
The opposite occurs. Difference limen for intensity is actually bigger... Remember that loudness growth is not the same thing as difference limen
What is loudness?
The psychological correlate of the physical parameter of intensity
The shape of the resulting power function varies depending on what?
The sensation. The loudness of sound plots in a similar way to brightness of light and intensity of smell and taste. It grows and then kind of levels off and grows at a smaller rate. For these, beta is less than 1.
What is difference threshold?
The smallest physical different, either of intensity frequency duration, etc., between two stimuli in which a listener can determine that the two stimuli are perceptually different.
What about when plotted on a log scale?
There is a linear relationship. So a 10 dB increase in phons always results in a doubling of sones.
What's the situation with recruitment?
There's a greater slope on the logarithmic scale, so smaller changes in intensity result in greater loudness growth. The graph shows no loudness perceived up until a certain intensity level and then all of a sudden, boom, it's loud. So there's less dynamic range
What characteristic of fricatives like f, z, and sh help us to discriminate speech?
They are of greater intensity than other fricatives
Review of magnitude estimation
This example does not follow the parameter discussed earlier, the 1000 hertz tone, it says that you present the subject with white noise at 85 dB SPL and tell them it's loudness is a 10. Then, you present them with white noise that is somewhere between 55 and 105 dB SPL and the subject gives the sound a number. A score of 20 would be twice as loud as the original stimulus. 5 would be half as loud.
what else?
This overall pattern is witnessed across frequencies.
What is another name for the relative difference?
Weber fraction.
The weber fraction tends to decrease as sensation level increases. What does this mean?
With higher intensity sounds, the required percent change is smaller to notice a difference. However, this is not in agreement with Weber's law, because it should remain constant.
Is loudness perception affected by other signal parameters?
Yes. Even when the intensity of a signal remains the same, loudness perception can be affected by frequency, duration, and bandwidth.
If you're looking at an equal loudness contour for 30 dB SPL, what will you see?
You will see that sensitivity is best in the mid-frequency range compared to low and high pitches
So, what would you expect for difference limen to be around threshold?
You would expect the different limen to be smaller near threshold.
What happens for very low frequencies?
a 10 dB increase in intensity more than doubles loudness. Recall that sones are a measure of loudness, not loudness level, and that even though a 10 dB increase in SPL correlates with a 10 phon increase, this is not the same as a 10 sone increase.
What is 2 sones?
a sound that is twice as loud as 1 sone. 4 sones? 4 times as loud as 1 sone, 2 times as loud as 2 sones.
Back to intensity discrimination, which is the procedure for determining phons
deals with same/diff, pres/not pres.
Tones from about 1 to 4.5 kHz are what compared to the 1 kHz standard when they are less intense than the 1 kHz tone?
equally as loud as the 1khz standard. this is the area of the graph where it dips down, meaning we're more sensitive in those regions
What's the deal with the loudness growth and DL disparity for low frequency and loud sounds?
loudness growth is greater for low intensity (near threshold) and low frequency sounds, see lecture 8 slide 21 & 26. but it does not mean the DL for intensity is better, see slide 28 & 32 same lecture. That could be because low frequency sounds activate larger area => small change cannot cause enough recognition; low intensity sounds cannot activate enough change for recognition. but both changes may cause relatively larger area activation => greater loudness growth.
What's a large DL reflect?
lower sensitivity
Generally, tones below 1000 hertz need to be what in order to be perceived as equally loud as the 1000 hertz tone?
more intense
Weber's law is a reasonable approximation of human perception, but it is not accurate near threshold and decreases as SL increases. A modified Weber fraction is used when stimuli are what?
presented close to the threshold of sensation
Weber's law underestimates our ability to detect small changes in intensity over a wide range of stimuli, especially at higher intensities
this is why it's a near miss
with increasing intensity, low frequency tones causes a greater increase in area of excitation-- meaning, more neurons are excited
this means low frequency sounds are perceived as creating a larger change in loudness. Low frequency sounds give power and creates more area of excitation.