week 6 february 19, 2013 human psychoacousticsese250/week6/week6_s13.pdf · the physical ear •...

Week 6 – Psychoacoustics ESE 250 S’13 DeHon Kadric Kod Wilson-Shah 1

ESE250:

Digital Audio Basics

Week 6

February 19, 2013

Human Psychoacoustics

2

Course Map

Week 6 – Psychoacoustics ESE 250 S’13 DeHon Kadric Kod Wilson-Shah

Where are we? • Week 2 Received signal is sampled &

quantized

q = PCM[ r ]

• Week 4 Sampled signal first

transformed into frequency domain

Q = DFT[ q ]

• Week 3 Quantized Signal is Coded

c =code[ q ]

• Week 5 signal oversampled & low

pass filtered

Q = LPF[ DFT(q+n) ]

• Week 6 Transformed signal analyzed

Using human psychoacoustic models

• Week 7 Acoustically Interesting signal

is “perceptually coded”

C = MP3[ Q]

Over

Sample DFT LPF

Decode Produce

r(t)

p(t)

q + n

C Perceptual

Coding

Store /

Transmit

Q + N Q

Week 4

Week 6

Week 5 Week 3

[Painter & Spanias. Proc.IEEE, 88(4):451–512, 2000]

3 Week 6 – Psychoacoustics ESE 250 S’13 DeHon Kadric Kod Wilson-Shah

4

The Physical Ear

• External Sound Waves

Guided by outer ear

into auditory canal

• Excite Inner Ear

Through mechanical linkage

connecting ear drum

to cochlea

[R. Munkong and B.-H. Juang. IEEE Sig. Proc. Mag., 25(3):98–117, 2008]


5

The Physical Ear

• Initiates signal processing

frequency domain analysis

Via analog computation

Video: Cochlea

• What part of the Cochlea vibrates for an 800 Hz square wave?



http://www.youtube.com/watch?v=dyenMluFaUw

6

The Cognitive Ear • Modern Psychoacoustics

Benefits greatly from o decades of neural recording o contemporary brain imaging technology



7

Power Spectrum Model of Hearing

• Rough Picture (main content of today’s lecture):

Critical Bands: Auditory system contains finite array

of adaptively tunable, overlapping bandpass filters

Frequency Bins: humans process a signal’s

component (against noisy background) in the one

filter with closest center frequency

Masking: certain signal components in a given band

are “favored” and others are filtered out

• Established through decades of psychoacoustic

experiments

B.C.J. Moore. Int.Rev.Neurobiol., 70:49–86, 2005.


8

Auditory Thresholds

• In the lab, you varied the frequency, amplitude and

phase of signals

• What was the effect of each, if any, on the sound you

heard?

Frequency

Amplitude

Phase


)2sin()( ftAts

Auditory Thresholds

• Harvey Fletcher (1940)

Played pure tones varying

o frequency, f [ Hz]

o Intensity,

I [Dyn ¢ cm-2]

= 10-5 [N ¢ cm-2]

= 0.1 Pa

o phase changes tend to be inaudible

Large listener population

o Young

o Acute

• Recorded extreme thresholds faintest audible

greatest tolerable


(http://www.et.byu.edu/)

10

Auditory Thresholds • Results:

pain-free hearing range extends at most over 20 Hz – 20 KHz

with sensitivity » 2 ¢ 10-4 ¢ 0.1 Pa = 20 Pa


0.1 Pa

[H. Fletcher. Rev. Mod. Phys., 12(1):47–65, 1940].

11

The decibel unit • Define standard pressure: p0 = 0.0002 ¢ 0.1 Pa = 20 Pa

• Threshold of human hearing

• Compute Sound Pressure Level as: LSPL = 20 log10(p/p0) dB

• LSPL for p1 = 20 Pa , for p2 = 200 Pa , for p3 = 20 mPa


Compare to

Ambient sea-level pressure:

1 Atmosphere

= 105 Pascal

• Q: why use log-log

scale?

• A1: dynamic range

• A2: “loudness” is a

power function

0.1 Pa

12

The decibel unit – Hearing intensity

Week 6 – Psychoacoustics

(http://www.dspguide.com/ch22/1.htm)

13

Let’s try to reproduce these results!

Week 6 – Psychoacoustics

(http://www.dspguide.com/ch22/1.htm)

• We will listen to single sine tones starting at a frequency of 10KHz, all the

way up to 20KHz, so each student can figure out their cut-off frequency

• Suggestions to improve this experiment?

14

Animal hearing ranges

• Dogs: Greater hearing range: 40Hz to 60KHz

Ultrasonic dog whistles

• Mice: Large ears in comparison to their bodies

Hearing range: 1KHz to 70KHz

Can’t hear low frequency noises

Communicate with high frequency

Distress call (40KHz), alert of predator

[Pictures from Wikipedia] Week 6 – Psychoacoustics ESE 250 S’13 DeHon Kadric Kod Wilson-Shah

15

Why Sinusoids?

• Why not some other harmonic series?

Fourier’s analysis shows

harmonic analysis could be based on

arbitrary smooth periodic fundamental

• Why does the animal receiver use

sinusoids?

• Hamiltonian Mechanics

Simplest physical model of vibrating

masses

Coupled spring-mass-damper mechanics

Produce sinusoidal harmonics • Video: Cochlea

m

x

b k

…. all sound

is produced

by vibrating

masses ….


http://www.youtube.com/watch?v=dyenMluFaUw

16

Masking - Spatial

• Masking Paradigms “Masker” masking “maskee”

Tone Masking Noise o pure tone

of 80 SPL

at 1 kHz

o just masks “critical band” noise of 56 SPL

centered at 1 kHz

Masker-to-Maskee ratio o Constant for fixed relative frequency and varying amplitude

o Changes with varying relative frequency

[T. Painter and A. Spanias. Proc. IEEE, 88(4):451–512, 2000.]

1 “Bark”

frequency

interval


17

Masking


The first graph shows the masking pattern for a 200Hz tone

Mostly masks tones around 200Hz, but also at harmonics

The second graph shows the same plot for different frequencies,

but only the fundamental part

Notice that the band gets wider for increasing frequencies

…masker at fundamental

can somewhat mask maskees

at the harmonics …

… but the “spreading

curve” is traditionally

depicted over the

fundamental only


18

Tone Masking Noise


• Are the following signals masked?

200 Hz tone at 80dB

200 Hz tone at 40dB

300 Hz tone at 40dB

400 Hz tone at 40dB

700 Hz tone at 30dB

19

Masking [H. Fletcher. Rev. Mod. Phys., 12(1):47–65, 1940].

• Tone Masking Noise (Fig 12) value above quiet threshold

such that a signal at the abscissa frequency

can be heard in presence of

top: 200 Hz tone

bottom: various frequencies

• Noise Masking Tone (Fig 13) dots show pure tone magnitude

(in dB)

required to be audible above noise

o Of the magnitude on the middle curve

o centered at that frequency

o with bandwidth

at least wider

than the bars of Fig 12



• Are the

following

signals masked

by the noise?

200Hz at 60dB

1KHz at 60dB

Noise Masking Tone


• Are the

following

signals masked

by the noise?

200Hz at 60dB

o Yes!

1KHz at 60dB

Noise Masking Tone

noise


• Are the

following

signals masked

by the noise?

200Hz at 60dB

o No!

1KHz at 60dB

Noise Masking Tone


• Are the

following

signals masked

by the noise?

200Hz at 60dB

1KHz at 60dB

o No!

Noise Masking Tone

25

Masking - Temporal • Temporal Masking Masker effect persists for tenths of a second

Masker effect is “acausal” o on ~ 2/100 timescales


26

Pitch JND • JND = “just noticeable difference”

change in stimulus that “just” elicits perceptual notice

where “just” means that a smaller variations of stimulus cannot be discerned



• What can you say about the JND: Below 1000 Hz?

o roughly constant

o ~ 3 Hz

Above 1000 Hz? o roughly log-log

linear

o Log[Jnd(f2)] - Log[ Jnd(f1)]

~ n (Log[f2] - Log[f1])

• Suggests that as frequency increases broader frequency

bands

“assigned” to same length of cochlear tissue

Remember cochlea model

What is n?

e.g. f1 =2000 f2 =4000

6 = 10 – 4 ~ n( Log10[2] )

) n ~ 20

27

JND experiment


• The following audio files contain a single

tone playing for 10 seconds. The sine

starts at 200Hz, then changes to a higher

frequency (201, 202, 203, 205, 210).

• This change occurs after a number of

“noises”: 1, 2, 3, 4, 5, 6, 7, 8 or 9.

• Can you notice when the change

happens?

28

Critical Bands

Decades of empirical study • reveal that human audio frequency

perception

• is quantized into < 30 “critical bands”

• of perceptually near-identical pitch classes

• corresponding to ~equal length bands of cochlear tissue (neurons)


29

Critical Bands: Evidence

Tone masking Noise (Fig. a & c)

o noise audibility threshold

o for small bandwidth noise

o remains constant

o until tone frequency locus

o falls away from critical

bandwidth

Noise masking Tone (Fig. b & d)

o same effect

o with masker and maskee

roles reversed

[T. Painter and A. Spanias. Proc. IEEE, 88(4):451–512, 2000.]


30

The Bark Scale

• “Bark” units: Uniform JND scale for frequency

Maps frequency intervals into their respective critical band number

[E. Zwicker. J. Acoust. Soc.Am., 33(2):248, February 1961]


31

The Bark Scale

• Frequency-to-Bark function First Principles vs. Empirical Modeling

[E. Zwicker. J. Acoust. Soc.Am., 33(2):248, February 1961]


))7500/((tan5.3)00076.0(tan13)( 211 fffB

32

Compression opportunities


Consider the following recording

Any ways to improve the compression?

33



Zooming in on a smaller portion


200Hz 205Hz Frequency

195Hz 193 194 196 197 198 199 201 202 203 204 206 207 208

dB

80

0

20

40

60

100

120

Masked

34






195Hz 193 194 196 197 198 199 201 202 203 204 206 207 208

dB

80

0

20

40

60

100

120

JND:

Could only

represent integer

frequency values

35






195Hz 193 194 196 197 198 199 201 202 203 204 206 207 208

dB

80

0

20

40

60

100

120

36

Next Week


• How can we use what we know about

human perception to compress music?

Frequency hearing range

Masking

o Temporal

o Spatial

o JND

o Barks

37

Big Ideas

• Sound is a pressure wave that makes the Cochlea vibrate with frequencies from ~20Hz (at the tip) to ~20KHz (at the base)

• This vibration is sinusoidal (physics) This is why sound harmonics are best represented as sinusoidal signals

• Masking Temporal – A masker tone can mask another tone that is present either

right before or a little after the masker

Spatial – A single tone can mask an entire frequency band (that contains the tone) if its intensity is high enough

There are <30 such bands (Bark scale), and they are wider for higher frequencies


38

Admin

• Lab 5 report due tomorrow

• On Thursday: Lab 6 You will be designing your own experiments

o To measure the range of frequencies you can hear

o To perform spatial masking experiments


39

ESE250:

Digital Audio Basics

End Week 6 Lecture

Human

Psychoacoustics


week 6 february 19, 2013 human psychoacousticsese250/week6/week6_s13.pdf · the physical ear •...

Documents