Audition 1 Exam #4

What is psychoacoustics?

-Psychoacoustics is the psychology of hearing science

-Definition: behaviour consequences of acoustic stimuli

-Psychophysics: sound properties and behaviour

-We investigate the relationship (F): how the behaviours are resulted from the stimuli

Analytical Studies

-Roles of individual acoustic parameter: intensity, frequency, and temporal patterns

-How are the acoustic parameters detected, discriminated, identified and scaled (quantified)?

Integrative Studies

-Information is embedded in the individual acoustic parameters

-The responses to the individual parameters are integrated for better perceptions of the information

-Integration is the process of neurons across different channels (different location in one nuclei), different types of neurons, and different nuclei, excitation vs inhibition, afferent vs efferent

Examples of Integration in Psychoacoustics

-Timbre: quality of a complex sound; integration of frequency components, intensity and temporal patterns

-Sound localization: use different cues in intensity, time and spectrum

-Auditory perception:

·Acoustic image in complex sound field

·Meaning: involving top-down process (experience and knowledge already stored)

Why care about analytical vs integrative?

-Analytical versus integrative are also two ways we treat information in our brains: we have two categories of skills in perception

-Sometimes, we pay attention to individual details

-Sometimes, we focus more on overall meaning

-We use two types of skills unconsciously

-Training changes the way we deal with signals

Dimensions of Auditory Ability

-Absolute limen: lower boundary of detection (i.e., minimum or sensitivity)

-Terminal limen: upper boundary of detection (i.e., maximum or limitation)

-Difference limen: the smallest change in some aspect of the stimulus which can be detected

-Limen also means threshold

What is Sensitivity?

-Absolute thresholds to sound intensity/pressure—basic target of hearing evaluation

What are Minimum Audibility Curves?

-Sensitivity across frequency range, or audiogram

-Frequency range of hearing: limits at the two ends

What is a Terminal Threshold (upper limit)?

-Threshold of discomfort; threshold of pain

Dynamic Range of Hearing

-Two thresholds define the dynamic range: terminal threshold and minimum audibility curve

·Above the terminal threshold we assume there is no auditory sensation

-What is dynamic—output changes with input?

·Loudness changes as a result of sound level change

-Dynamic range varies with frequency: largest at middle frequency

Hearing Area (defined by level and frequency)

-Terminal thresholds can be defined in different ways

-Less clear (accurate), less investigated (ethical reasons)

Threshold is determined statistically

-The threshold is statistically obtained, based upon a criterion as % of correct response that is selected

·Point at which subjects make 50% of correct responses for yes/no method (which is significantly higher than random)

-Corresponding sound level is determined as threshold

-The % criterion depends on the methods; the principle is that the criterion can be reached not only by guessing

Variability and Reliability

-Many factors (other than real ability) impact performance—that is why statistics is used

-Variability (individual variation): threshold can vary 20 dB across individuals with normal hearing (i.e., normal hearing range from -10 to 10 dB HL)

-Test-retest reliability (variation with individuals): the variation across different tests on the same groups of subjects

-It is often given as a range (i.e., 10-15 dB in hearing threshold)

The meanings of the variations

-The normality is defined as a range

-The range depends on the accuracy of the method and the skill level of the performer

-In our clinical setting, 10-15 dB is normal for test-retest reliability

·If you do poorer than that, you need more training

-Criteria are established based upon the reliability to judge if one really has a hearing loss

How is the configuration of the audibility curve (threshold curve) formed?

-Contribution from external and middle ears

-Recall the frequency response properties of external and middle ear (most gain at middle frequencies)

-Audibility curve is the same as absolute threshold

Contribution of cochlea to audibility curve

-Bone conduction by-passes external and middle ear; therefore, can exam the cochlear response

Minimum Audible Field (MAF) vs Minimum Audible Pressure (MAP)

-MAF: threshold measured with speaker in open field, calibration in open field, SPLs measured with subject absent

-MAP: threshold in closed field, monaural, SPLs measured with coupler

-The difference is 6-10 dB better (lower) for MAF

3 proposed reasons for the 6 dB difference

-MAP method eliminates external ear resonance, resulting in an amplification loss

-Binaural summation in MAF

-Physiological noises heard in MAP

Current opinion of MAP vs MAF 6 dB difference

-The current opinion is that the "missing 6 dB problem" does not exist.

-It is due to a calibration difference: if calibration with real ear measure (probe mic at eardrum), no 6 dB difference

·However, this difference is still important because real-ear calibration is not practical in many cases

Reference Equivalent Threshold SPL (RETSPL)

-ANSI standard for average thresholds at different azimuths and listening conditions

-Sound pressure is measured in coupler or in space without subjects

-Coupler pressure is measured to quantify voltage-sound pressure relationship to ensure the reference is equal across different labs

RESPLs: data and applications

-Application for:

·References for speaker/earphone calibration

·Establishing hearing level (0 dB HL)

·Establishing allowance of noise in the two conditions

SPL, HL, and SL

-SPL: refers to physical sound pressure (20 uPa)

-HL (hearing level): reference of hearing thresholds of normal subjects using RETSPL

-SL (sensation level): reference of individual hearing thresholds

Weber's Law

-A popular law for sensory discrimination

-Difference threshold = difference limen

-For intensity, it is defined as the minimal intensity difference that can be detected: delta I (also called just detectable difference or JDD)

-Weber's law predicts the relationship between delta I and I: delta I is proportional to I

·Expressed as delta I/I = constant

Converting Weber's fraction to dB

-Delta I/I: Weber's fraction is a ratio measure of intensity discrimination

-Delta I dB = 10log (1 +delta I/I)

-According to Weber's law, both Weber's fraction and JDD in dB will NOT change with base quantity

-The intensity discrimination for pure tones is ~1 dB (or ~0.3 in delta I/I)

Intensity discrimination for pure tones vs white noise

-If Weber's law is correct, we will see a horizontal line

-But we see a slightly sloping line for pure tones

-So, Weber's law is close to the truth, but not 100% for pure tones

-White noise: the intensity discrimination threshold is smaller for white noise than for pure tones (closer to Weber's law)

-Important to note that results vary with method use

Different methods for intensity discrimination

-Continuous tone plus increment: detect change in increment; influenced by adaptation (the decline of a long-lasting signal)

-Gated pulse tone: results in larger noted differences than continuous pedestal. May involve short-term auditory memory

-Continuous pedestal has better intensity discrimination

Loudness Detection Definition

-Loudness is the perception of sound strength, mainly related to sound level, but also changed by many other factors, such as frequency, duration, existence of other signals, etc

-Loudness is a subjective concept and can only be tested behaviourally

Loudness Scale

-10 dB level change results in 2 times of change in loudness

·Not true at low SPL close to threshold

Equal Loudness Contours

-Graph with SPL dB on the y-axis and frequency on the x-axis, showing graph lines connecting points of equal perceived loudness

-At low SPL, such as 10 dB, the SPL required for the same phon (i.e., 10) largely varied with frequency: 30 dB SPL at 100 Hz, 0 dB SPL at 4000 Hz, etc

-At high SPL (i.e.,100 dB), the variation across frequencies is less (the contours are flatter)

Impact of the contour shape on dynamic range and loudness growth

1) The dynamic range measures the SPL difference between the floor and ceiling

2) Ceiling is NOT changed while floor varies

3) 1000-4000 Hz is the most sensitive region (lowest threshold), and therefore the largest dynamic range

4) Smaller dynamic range and faster loudness growth at low frequency; slightly so at high end of hearing frequency

Loudness Contours: Boomy vs Tinny

-Boomy: too much low pitch when the speech/music plays back at a higher level than when it was recorded. This is because some low-frequency sound that was not audible when recording becomes audible when playing back

-Tinny: too much high pitch (crispy) when sound is played back at a lower level than when it was recorded

Filter (input network) setting of sound level meters

-C rating follows contour at very high sound levels, no filtering at all

What is a critical band (CB)?

-The band in which a sound is "heard" by a HC and connected SGNs

The impact of signal bandwidth on loudness

-Keep total sound level (therefore the total power of sound) the same

-Increase the bandwidth of the signal from very narrow to beyond CB

-The loudness won't change within CB, but will increase when the bandwidth is greater than CB

-The signal level is well above threshold. When it spreads to more than one CB, more auditory channels are activated, so the sound is louder

Adaptation vs Fatigue

-Adaptation: decrease of sensitivity during signal presentation—reduced the response to non-novel signal

·Perfume only makes you attractive for so long because of adaptation

-Fatigue: decrease of sensation after signal presentation—reduced the response due to the (partial) failure. Reduced the sensitivity to all stimuli

How to measure adaptation + results

-By loudness matching

·In monaural presentation, let one ear hear a continuous sound

·Match its loudness by presenting a pulsed signal to the other ear

·See the change over time

-Results:

·Large variation across individuals

·More significant at lower sound levels

·More at higher frequency

Effects of masking on loudness growth

-When there is masking, threshold (floor) elevated and ceiling doesn't change: narrower dynamic range

-Result: faster growth with masker

·Also occurs in sensorineural hearing loss (SNHL)

Loudness Recruitment

-Refers to abnormally rapid growth of loudness with stimulus level; the term is reserved to SNHL

-The behaviour of loudness growth in SNHL is similar to what is shown with masking in normal subject, but some differences exist

What is masking?

-Masking: when one sound interferes the sensation of another sound

-Why study masking?

·It occurs in our hearing environment and has a great impact on our hearing

·It is a useful tool to study hearing

·It has important clinical applications

Masking changes threshold

-Masked threshold—the threshold of a probe/target sound (tone) in the presence of a masker

-Masked threshold is usually higher than the threshold in quiet

-Maskers: can be any sound: noise, or tones

-Timing relation: simultaneous, before, or after signal

Energetic (Peripheral) Masking

-The masking by the masker excitation to neurons required for sensing target signal

-It is a masking in cochlea in which the excitation by both the signal and masker (partially) overlap

Informational Masking

-The masking that occurs not via energetic interaction between masker and signal in cochlea

-In informational masking, there is no overlap in frequency spectrum between the masker and the signal

Partial Masking

-The hearing of the target when a masker exists

-The masker does not totally cover the target, but reduces loudness of target

Masking tuning curve

-Masking tuning curve (TC) as an indication of frequency resolution

·Masking TC is how the masked threshold by a tone masker changes with signal frequency

·Psychophysical TCs (masking TC obtained in behavioural test)

Psychophysical TC

-When the masker frequency is getting closer to the target tone, the masker threshold is lower

-Asymmetric: the mask tuning curve is broader at lower frequency side of the target tone—consistent with the fact that lower frequency sounds easily mask higher frequency sounds

Why does the psychophysical TC tell us the frequency selectivity?

-The hearing of target tone represents the hearing of one auditory channel

-The change of masking with frequency shows in what frequency range this particular channel is sensitive (therefore can be inferenced)

-This frequency range for effective masking is what this channel can hear (or be selective to)

-Critical band is the concept that is represented by this selectivity

Critical band defined by masking

-Masking TC tells us that if the masker spectrum is too far away from the tone, no masking can be produced

-Therefore, if a broadband noise is used as a masker, only energy in certain band is effective (this is critical band)

-CB can be considered as a psychophysical filter, through which sound can affect the perception of the signal centred in the band; while the sounds outside CB do not work

-CB is proportional to the central frequency (20% or 1/3 octave). Therefore, we use 1/3 octave band of noise as the masker in clinic since only the noise inside this band is effective

Forward, simultaneous, and backward masking

-Backward masking: tone before masker

-Simultaneous masking: tone during masker

-Forward masking: tone after masking

-Masking effect is largest at the onset and offset of the masker

Monotic vs Dichotic Masking

-Monotic: the masker and probe with the signal are in the same ear

-Dichotic: the probe with the signal is in one ear and the masker in the other

·Dichotic is mainly by central masking so weaker than monotic which is by energetic masking

Two ways that frequency is processed (for us to perceive it)

-By place code

·Tuning curve of individual neurons

·Behavioural tuning curve (examined by masking)

·Critical band as it relates to frequency discrimination

-By temporal code

·By phase locking

Frequency discrimination limen

-Behavioural measures of frequency processing ability

-Can be measured in:

·Absolute terms (delta f)

·Relative terms (delta f/f)

·Weber's fraction

Method concerns with frequency discrimination

-Pitch fusion

·Two tones of different frequencies (larger than the delta f) can produce a fused pitch (one pitch) if they are presented simultaneously or close

·Sequential presentation should be used to avoid fusion

-Frequency splitting (splattering)

·When a tone is turned on and off quickly, the frequency will spread to unwanted region

·Shallow ramping and masking help to reduce splattering

Three methods used for intensity discrimination and how they relate to frequency discrimination

-Gated pulse tone: must be used with appropriate methods to reduce splattering

-Continuous tone plus increment: simultaneous presentation, so not suitable for frequency discrimination task due to pitch fusion

-Modulation tone: frequency modulation rather than amplitude modulation

Trends for frequency discrimination by gated pulses

-In the low frequency range, the minimum frequency difference that can be detected does not change much as frequency increases

-However, when the frequency is beyond 2000 Hz, delta f increases quickly with frequency (at higher frequencies we need a larger delta f to notice a difference)

-Better able to discriminate lower frequency sounds

Comparison between gated pulse tone and frequency modulation (FM)

-No difference for frequencies < 2000 Hz

-Better performance using FM when frequencies > 2000 Hz

Effect of stimulus duration (temporal summation) on frequency discrimination

-Delta f improves (gets smaller) up to durations of 100 to 200 ms

-This improvement is also seen in intensity discrimination**

Pitch & Frequency similarities and differences

-Pitch sensation only exists when the sound is heard

-Two tones with different frequencies may evoke the same pitch if close enough in frequency and in time

-Pitch changes with intensity

-Pitch is not linearly related with frequency

Measurement of Pitch

-1000 mels: the pitch produced by a 1000 Hz tone of 40 dB SL

-No direct relationship between frequency and mels

Meaning of mels

-Bark is the unit for critical band

-1 bark covers 100 mels and 1.3 mm on basilar membrane

Analytic vs Synthetic Pitch

-Fundamental frequency determines the pitch

-Harmonic components determine timbre

-Analytic pitch: ability to hear the partials (harmonic components) as well as the fundamental

-Synthetic pitch: the single dominant pitch that arises from a complex tone—virtual pitch

-We develop synthetic pitch naturally, whereas it takes a while to develop analytic pitch

Pitch of the missing fundamental

-Listeners hear a pitch of 100 Hz when a series of tones consisting of 400 Hz, 500 Hz, 600 Hz, and 700 Hz are presented

-The fundamental frequency is 100 Hz and is missing

Periodicity Model

-F0 is encoded by the periodical behaviour of neurons in the auditory system in response to the periodical feature of signals in time domain

-That is: phase locking of auditory neurons to the low-frequency envelope of sound produces the perception of residual pitch

Two aspects of auditory temporal processing

-Temporal resolution

-Temporal integration

Temporal summation (integration)

-Hearing improves with elongated stimulation

-System integrates information of long duration to improve hearing

-Improvements in thresholds and discrimination

Hearing threshold improvements with temporal summation

-Threshold improves (reduced) by 3 dB for every doubling of duration when it is less than 200 ms

-Temporal summation is larger at lower frequencies and smaller at higher frequencies

Calculation for Total Energy (E)

-Total Energy (E) is the product between power (P) and time duration (T): E = P*T

-The change in the power can be compensated by the change in the time duration

Why does threshold improve with duration?

-Neural summation

Temporal Resolution & Segregation

-Temporal Resolution: ability to perceive quick changes in signals

-Segregation: ability to verify a series of signals that are presented closely in a time sequence

Methods for testing temporal resolution

-Behavioural tasks:

·Discrimination of pulsed signals

·Gap detection

·Temporal modulation by transfer function (by AM signals)

·Forward masking

·Duration discrimination

-Objective measures

Two aspects that are changed when testing temporal resolution with amplitude modulation

-Frequency and depth

Temporal resolution by click trains

-Clicks presented in different rates

-Higher the rates, smaller the intervals between clicks

-The resolution is ~6 ms, corresponding to 160 clicks per second

-This is much faster than the visual system (16 Hz)

Binaural (spatial) summation

-6 dB gain for binaural hearing compared to monaural hearing

-The signals may not necessarily be presented simultaneously to both ears, rather, they can be separated by a short time (

What are two other aspects that binaural hearing improves?

-Intensity discrimination and frequency discrimination

·Likely causes: reduces masking effect and improves sound quality

Sound localization in binaural hearing

-Binaural hearing allows us to discriminate where each sound source is coming from in our environment

What does do the visual and somatosensory systems have that the auditory system does not?

-They have 2D spatial inputs. Therefore, binaural cues exist to help with sound localization

Two planes for sound localization

-Azimuth (horizontal)

·Only plane that uses binaural cues

·Uses binaural cues in time and level

-Elevation (vertical plane)

What are the two reasons for intensity differences between the two ears?

-Azimuth (more important) and distance differences (less important)

What affects the shadow effect of the head?

-Shadow effect: the intensity of a sound is reduced when sound travels from one ear to the other side of the head (head softens the sound coming from one direction)

-Frequency and wavelength

-More shadow effect at higher frequencies (and more IID)

Where should sound be placed for best sound localization ability?

-In front of the head, at the zero azimuth. Our auditory system is organized so more neurons process location changes in the front

Strongest binaural cues for IID/ILD and ITD

-Binaural cues are largest at 90 degree azimuth for both ITD and IID

When does interaural phase difference (IPD) work?

-When there is a continuous signal. Presenting a sound for a long time means that the time difference at onset can be forgotten, so the phase difference becomes a more important factor for sound localization

Duality theory of localization (Duplex Theory)

-Sound localization in low frequency region depends on ITD/IPD, while in high frequency region on IID

·In reality, sound is often a complex signal. Even a high frequency tone can provide an ITD/IPD if modulated at low frequency

Which frequencies of pure tone localization are the poorest?

-The middle frequencies (even though we have the best hearing there)

Two main methods to measure localization

-Localization Error: refers to the difference between physical target and the identified direction by the subject

-Discrimination: the minimum audible angle (MAA) that can be detected by the subject

·MAA is much smaller than localization error

·MAA also shows the poorest results at middle frequencies

Monaural Spectral Cues (HRTF)

-The vertical plane, filtering from the head and pinna where there is a shadow effect from behind

-The reflection from the pinna creates multiples waves and time differences

How do spectrum cues by HRTF help?

-It helps with the confusion between the front and the back. This helps in cases where the binaural cues are the same on the azimuth plane (i.e., at 0 and 180 there is no difference)