Psychoacoustics and CAPD

Question 1

What is psychoacoustics?

Answer 1

A branch of psychophysics that deals with the relationship between and a stimulus and perception of the stimulus by the listener
We all perceive stimulus differently (otherwise we’d all like the same music)
Response from the listener consists of a change in listener’s sensory perception and listeners response criteria (biases) contributing to the response

Question 2

What is the psychoacoustic methodology aimed at?

Answer 2

Controlling or minimizing the effects of listener response bias
Trying to keep everyone at the same level

Question 3

What is sensitivity?

Answer 3

Ability of a test to correct identify those with disease
True response / (true response + false positive) x 100

Question 4

What is specificity?

Answer 4

Ability of a test to correctly identify those without disease
Correct rejection / (correct rejection + false alarm) x 100

Question 5

Does a test with high sensitivity have few false negatives?

Answer 5

Yes

Question 6

Does a test with high specificity have few false positives?

Answer 6

Yes

Question 7

Are sensitivity and specificity reciprocal of each other?

Answer 7

Yes
Ideally, high sensitivity and a high specificity is desired for any kind of test
In reality, if a test has a high sensitivity, then its specificity is lower and vice versa

Question 8

What does a strict criterion do to the sensitivity and specificity?

Answer 8

Increases specificity and lowers sensitivity

Question 9

What does a lax criterion do to the sensitivity and specificity?

Answer 9

Increases sensitivity and lowers specificity

Question 10

What is validity?

Answer 10

The extent to which a test measures what it is supposed to measure; i.e., the accuracy of the test
Measured by sensitivity and specificity

Question 11

What is predictive value?

Answer 11

The likelihood that a positive test result indicates disease or that a negative test result excludes disease
We can use a sensitivity and specificity matrix to calculate the predictive value of a test
The proportion of positive tests that are true positives is the positive predictive value (PPV) of a test (increases with increasing sensitivity and specificity)
PPV increases with increased disease prevalence in the population
If the patient has a +ve test, how likely is she to have the disease?
If the patient has a -ve test, how likely is she not to have the disease?
*to answer these questions, you need to calculate the PPV and NPV

Question 12

What happens to PPV and NPV at prevalence goes up?

Answer 12

The PPV rises and the NPV falls
Only happens if sensitivity and specificity are relatively good (if sensitivity and specificity are bad, no point in doing the test)

Question 13

Do all other methods of testing need to be compared to the gold standard?

Answer 13

Yes
You take the positive mammogram patients to verify with biopsy
If biopsy also is positive, it’s pretty sensitive
If the positive predictive value (PPV) for the test is high (as close to 100 as possible), then it suggests that the new test is as good as the gold standard

Question 14

What is a receiver operating characteristic (ROC) curve?

Answer 14

When sensitivity versus specificity is plotted
The area under the curve is also measured (AUC)
Done because sensitivity and specificity varies across different thresholds of test results
This curve plays a central role in evaluating diagnostic ability of tests to discriminate the true state of subjects (diseased vs. not-diseased)
It helps finding the optimal cut off values for normal vs. abnormal test results

Question 15

Can you compare two alternative diagnostic tasks when each task is performed on the same subject?

Answer 15

Yes
Compare across with the ROC curve
Find out which task gives the best results

Question 16

What is pitch?

Answer 16

A subjective or perceptual attribute that corresponds closely to the physical attribute of frequency
A change in frequency is heard as a change in pitch
Perceptual attribute (cannot be directly measured)
Measured by adjusting a frequency until it matches the pitch in question
Brain recognizes the pitch, not frequency
Pitch is related to the physical repetition rate of the waveform of sound
Increasing repetition rate provides a sensation of increasing pitch

Question 17

What is frequency discrimination?

Answer 17

Ability to detect changes in frequency
It involves discrimination between two sinusoids presented successively with a brief silent interval between them (jnd)

Question 18

What is frequency selectivity?

Answer 18

The ability to resolve a complex sound to its component frequencies

Question 19

What are the two current theories for pitch perception?

Answer 19

Place theory (place of max stimulation along the BM, corresponds to the max displacement in the traveling wave)
Temporal or volley theory (auditory neurons phase lock to vibrations of the BM, pitch assigned to a signal is determined by the timing pattern of the neural impulse - different frequencies produce different patterns of neural spikes)

Question 20

For place theory, are both frequency discrimination and frequency selectivity closely connected?

Answer 20

Yes

Question 21

For the temporal theory, do frequency tones fire at a particular phase of the waveform?

Answer 21

Yes
Neural spikes are at or close to the integer multiples of the period of the tone

Question 22

When do timing cues break down/become inconsistent?

Answer 22

Above 5 kHz

Question 23

Is place theory believed to play a larger role in pitch perception at high frequencies?

Answer 23

Yes
The human auditory system is not very efficient at frequency discrimination at higher frequencies (> 5000 Hz) - can be seen by no norms for EHF
Phase locking breaks down at 5000 Hz
According to Plack (2005), the place theory maybe responsible for representation of frequency but the sensation of pitch is probably derived from the temporal pattern of neuronal firing (picking up on the repetition)

Question 24

What is the pitch of the missing fundamental?

Answer 24

Pitch remains the same, even though the fundamental frequency is missing
The character of each note will change
The brain fills in the blank with the fundamental
This perception is due to the brain interpreting repetition patterns, harmonics-periodicity, that is present

Question 25

What theory explains the phenomenon of the missing fundamental?

Answer 25

Cannot be explained by place theory
Could be explained by temporal theory (temporal pattern of neurological activity is related to the period)
But neither completely explains it

Question 26

Is cochlear loss make speech perception and music appreciation difficult?

Answer 26

Yes
Cochlear hearing loss is associated with reduced frequency selectivity, i.e., broader auditory filters
Spectral analysis of the cochlea is less efficient
Resolution of harmonics of complex tones becomes difficult, especially for moderate number of harmonics
Distorts the signal, they can often misunderstand something

Question 27

Do individuals with cochlear loss depend more on temporal information than spectral information?

Answer 27

Yes
Because of reduced frequency selectivity due to broader auditory filters

Question 28

What is the cause of the variability of results across listeners with similar audiometric thresholds?

Answer 28

Individual differences in auditory filter size
Neural synchrony (well preserved neural synchrony results in good pitch discrimination, poorly preserved neural synchrony results in poor pitch discrimination regardless of the size of the filters)

Question 29

Why is pitch perception important?

Answer 29

Plays an important role in the ability to understand language
To distinguish most important utterances in speech
To indicate structure of sentences of phrases, especially for tonal languages (e.g., Mandarin, Chinese, and Thai)
To convey nonlinguistic information (gender, age, emotional status, inflection, etc.)
To supplement speech reading (the way you move your mouth helps)

Question 30

Why is timing important?

Answer 30

Sounds are dynamic
It is important information that is necessary for speech perception

Question 31

What is temporal resolution?

Answer 31

The ability to detect changes over time between two brief stimuli
Tested with gap detection

Question 32

What are the two main processes that temporal resolution relies on?

Answer 32

Within-channel gap detection threshold (analysis of time patterns within each frequency channel or filter, the minimum time needed to detect a gap between sounds that have the same spectrum, both 1000 Hz)
Across-channels gap detection threshold (minimum time needed to detect a gap between sounds that are spectrally dissimilar (tone and noise), minimum time needed to detect a gap between sounds presented in two ears)

Question 33

What is temporal integration or summation?

Answer 33

The ability of the auditory system to add information over time
The longer the sound is heard the better detection and discrimination
Clicks (OAEs) need to be louder than tones so they are heard because they are shorter in duration
The auditory system appears to integrate the pure tone signal over a 200 to 300 ms period
If the stimulus duration is too long, the threshold may become worse due to adaptation (neurons will not fire)

Question 34

Are sounds characterized by pressure variations over time?

Answer 34

Yes
If intensity and frequency are constant over time, sound is perceived as steady

Question 35

Is broadband white noise perceived as steady?

Answer 35

Yes
Although the waveform of broadband noise fluctuates rapidly from moment-to-moment, the fluctuations are too rapid to be processed by the CANS and cannot be heard distinctly

Question 36

What are the processes that are commonly used to measure temporal resolution?

Answer 36

Gap detection threshold (determine the smallest detectable time-gap between two stimuli)
Amplitude modulation detection threshold (determine the smallest amount of variation needed to detect that a sound is fluctuating in level and not steady)
Temporal masking (non-simultaneous masking, speech or masker comes first and the other comes after)

Question 37

What is the gap detection threshold?

Answer 37

The smallest amount of silence detected between two signals
Signals could be sinusoids, BBN, or NBN
The GDT for clearly audible sound (greater than 25 to 30 dB SL) is about 2 to 3 ms
GDT increases for frequencies < 200 Hz
This test cannot be done at threshold
Can also be done with two or more signal pairs with one signal containing a variable silent gap
Can also be done with a series of BBN segments with 0 to 3 gaps per segment varying in duration (GIN test)

Question 38

What is amplitude detection?

Answer 38

Temporal modulation refers to a recurring change
E.g., frequency or amplitude change in the signal over time
The degree of change determines modulation depth of signal
A minimum modulation depth is required to detect that the stimulus is modulated or changing in some way
The depth of modulation is dependent on the rate of the stimulus, which is the frequency at which the modulation changes over time
Ability to detect modulation worsens at low SLs and higher frequencies as modulation rate increases, greater modulation depth is then needed for detection

Question 39

When are modulation signals used?

Answer 39

When they want to get people’s attention
Could modulate in rate or depth

Question 40

What are temporal or non-simultaneous masking?

Answer 40

Masker and test signal do not overlap in time; separation delay between signal and masker

Question 41

What is forward masking?

Answer 41

Short duration signal masked by louder sound closely preceding it
Masker is terminated before signal onset
Temporal resolution is insufficient to separate signal and masker
Masking of the signal occurs

Question 42

What is the theory behind forward masking?

Answer 42

When the masker is turned off the internal representation of the masker takes time to decay
Neurons have a refractory period before they can refire
At the point of the refractory period, the smaller signal comes in and cannot stimulate the neuron enough to fire an action potential
If a brief signal is presented during this time, forward masking may occur; signal will not be heard for that time
This disappears after 200 ms after the masking noise
If the signal is as large or larger than the masker, it will still be heard

Question 43

What is backward masking?

Answer 43

Short duration signal masked by a sound rapidly following it
Signal is presented near the beginning of the masker
Small signal, large masker
The response to the signal may still be building when the masker follows it
If the masker is sufficiently intense, it may “swamp” the signal
Typically occurs for up to 50 ms (shorter than forward)
For signal detection, separation between masker and signal has to be > 25 to 50 ms
The theory behind this is more controversial

Question 44

What is binaural?

Answer 44

Sound reaches both ears

Question 45

What is diotic?

Answer 45

Identical stimuli presented to both ears

Question 46

What is dichotic?

Answer 46

Different signals presented to two different ears

Question 47

What is localization?

Answer 47

Specifically in the sound field

Question 48

What is lateralization?

Answer 48

Specifically under headphones or inserts

Question 49

Do lateralization and localization rely on the same binaural cues and mechanisms?

Answer 49

Yes

Question 50

What is the duplex theory for sound localization?

Answer 50

There are two sound cues used for localization (ITD and ILD)

Question 51

What are the two sound cues used for localization?

Answer 51

Interaural temporal (or phase) difference (ITD or IPD)
Provides localization information for low frequency stimuli
Interaural level difference (ILD)
Provides localization information for high frequency stimuli
Localization is better for complex stimuli than for pure tones

Question 52

How does ILD work?

Answer 52

Wavelength of lower frequency signals are (greater) longer compared to the head width
They bend very well around the head without creating a “sound shadow” or “head shadow”
Higher frequencies have shorter wavelengths which would produce a sound shadow
It may be as high as 15 to 20 dB
It can occur at low frequencies, they are really only useful at high frequencies

Question 53

How does ITD work?

Answer 53

Low frequency cue
High frequencies do not provide good interaural temporal cues
Ambiguous above 750 Hz and undetectable at ≥ 1500 Hz
When the wavelength of sound wave is smaller than the diameter of the head (> 1500 Hz), an ITD is greater than one period of the wave
Will not be able to give a unique time difference for high frequency sounds

Question 54

Will there be interaural differences when the stimuli is at 0 degrees azimuth?

Answer 54

No

Question 55

What is the minimum audible angle (MAA)?

Answer 55

The detection of small shifts in position of sound source (sinusoids) between reference and comparison signals
Measured using sounds presented through loudspeakers to observe the contributions of both ITD and ILD at once

Question 56

What are the MAA results when ITD cues are used?

Answer 56

MAA is smallest for sounds coming directly from front, i.e., when the reference ITD is at 00 azimuth
MAA is about 3 degrees for frequencies up to 900 Hz using ITD cues
Between 900 and 1500 Hz, the MAA threshold using ITD cues increases dramatically
Above 1500 Hz, ITD cues are undetectable and cannot be used to detect MAA for sinusoids

Question 57

What are the MAA results with ILD cues?

Answer 57

MAA is also smallest for sounds coming directly from front, i.e., when the reference ILD is at 0 degrees azimuth
Changes in ILD can be detected across frequencies when the azimuth is > 0 degrees, but for practical purposes, ILDs are sufficiently large to be useful only at high frequencies
Performance worsens around 1500 to 1800 Hz because wavelengths are smaller than the head diameter

Question 58

What two binaural phenomenon are observed under headphones?

Answer 58

Binaural fusion
Binaural beats
*Both are believed to involve processing within the CANS, probably with the SOC playing a major role

Question 59

What is binaural fusion?

Answer 59

It is a sensation of hearing a fused auditory signal in the midline
Example: Two recordings of a word list presented dichotically
One ear receives a high-pass filtered version of the list and the other a low pass filtered version of the list (same word, filtered differently)
The listener’s word identification will be that of the original unfiltered list of words
Fusion will not happen when different words are presented

Question 60

What are binaural beats?

Answer 60

Beats provide an extra cue for signal detection
When the tones are in-phase, they add
When the tones are out-of-phase, they subtract
When two signals with frequencies very close to each other are presented, beats are produced
Beats are the actual waxing and waning of the signal (2 sounds of different freq combining)
Same intensity for both signals

Question 61

What is the formula for beats?

Answer 61

Number of beats/second = | fS2 - fS1 |
If difference between frequencies = 3 Hz; waxing and waning will occur every 3 times/s
If difference between frequencies = 5 Hz; waxing and waning will occur every 5 times/s

Question 62

What happens when the difference between the signals is greater than 50 to 100 Hz? (beats)

Answer 62

The time difference allows the brain to detect the two signals as separate pure-tones with different frequencies
The two tones will not produce beats

Question 63

What are the different phenomenon for sound source determination?

Answer 63

Perceptual Coherence
The Precedence Effect
Modulation Detection Interference (MDI)
Comodulation Masking
Binaural Masking Level Difference (MLD)

Question 64

What is perceptual coherence?

Answer 64

An advanced but very robust auditory skill by which components of speech are grouped together and perceived as one auditory event
The more common the sound features are, the more likely it is for the brain to group them
When parts of speech cohere, they are set apart from other acoustic events like noise or speech of other talkers

Question 65

What makes perceptual coherence most likely to occur?

Answer 65

A common fundamental frequency (f0)
The same voice onset time (VOT) - When the vocal tract is blocked for a stop consonant and a vowel follows; it is possible to measure the time between release of the stop and start of voicing

Question 66

What is the precedence effect?

Answer 66

Describes an illusion produced when two similar sounds are delivered in quick succession from sound sources at different locations but only a single sound is perceived
Damping out the other sound so you don’t hear an echo
If delays are very short (< 1 to 2 ms), summing localization occurs
A phantom source is perceived; location is toward leading sound
We localize on the basis of which signal reaches our ears first

Question 67

Most of the time, are we unaware of how much (50 to 90%) of the sound heard comes from reflections from environmental surfaces? (precedence effect)

Answer 67

Yes
We only notice these reflections (echoes) when the time delay gets longer than about 30 to 50 ms (the echo threshold)

Question 68

What is modulation detection interference?

Answer 68

The amount of masking or interference caused by the modulating masker
Amplitude modulation is not perceived independently in each of the cochlear auditory filters
Instead, it appears to cause significant interference or crosstalk across auditory filters
When masker and signal modulator frequency are similar, even if not in the same filter, threshold for detection of the amplitude modulated signal increases (gets worse)
When masker and signal modulator frequency are different, the threshold for modulation detection for the signal decreases (improves) - will not mask effectively
Will not occur with pure tones

Question 69

What is co-modulation masking release?

Answer 69

The detection of a tone masked by a modulated noise will improve significantly if another band of noise with the same temporal characteristics is added
Can occur when energy is added in frequency regions remote from the signal

Question 70

Why does co-modulation release occur?

Answer 70

Occurs because the two modulated bands of noise as perceptually grouped by the CANS
The signal is detected as a separate auditory event
Called perceptual coherence

Question 71

What is an auditory scene analysis?

Answer 71

Allows the auditory system to break down sound waves into different frequency components using the spectral analysis power of the cochlea
After separating the sounds, the auditory system can assign different sounds sources to different components
Without this, the auditory system would not be able to separate simultaneous sounds (people with hearing loss have difficulty bc of poor frequency selectivity)

Question 72

How are we able to attend to a single conversation in a noisy background? (cocktail party)

Answer 72

Binaural hearing (using both ears) helps us to separate interesting sounds from a background of irrelevant noise
In a room where several conversations are taking place, one can focus on one of them and ignore the rest
No completely understood
Can occur both when we are paying attention to one of the sounds around us and when it is invoked by a stimulus which grabs our attention suddenly (someone on the other side of the room calling your name)
Auditory version of the figure-ground phenomenon

Question 73

What are some factors that may aid listening in a cocktail environment?

Answer 73

Spatial separation (voices coming from different directions)
Different pitches of different speakers
Different f0 of speakers (male vs. female differences)
Different accents
Different speeds of sound reaching the ears

Question 74

What is the binaural masking level difference?

Answer 74

The masked threshold of a signal can sometimes be significantly lower when listening with two ears
The listeners listens to a tone and noise identical in both ears
The level of the tone is adjusted till it is no longer heard (masked)
If the signal is phase-shifted by an equivalent of 180 degrees, the tone is audible again - release from masking
The difference between the two levels of signal detection is called the masking level difference (MLD)

Question 75

Why does MLD occur?

Answer 75

The more different the interaural relations of a signal, the better its detectability

Question 76

What frequencies are the largest MLD seen for?

Answer 76

Low frequencies
Because interaural phase difference is useful for localization of low frequencies