frequency-place-transformation
DESCRIPTION
Presentation for the "Models of Auditory Perception Seminar". This presentation is about the auditory system and, in particular, the inner ear. After explaining the mechanics of the cochlea in details, I am focusing on the Frequency-Place-Transformation, which takes place along the basilar membrane. This process is then closely connected to the so-called auditory filters and the masking effect. Finally, I am comparing how all these different scales are related to each other.TRANSCRIPT
Frequency-Place Transformation AIPA: Models of Auditory Perception Seminar
20 November 2014
The speaker
Francesco Bonadiman [email protected]
Computer Science
Bachelor @UniTN (Trento, IT)
2/29
The speaker
Francesco Bonadiman [email protected]
Computer Science
Bachelor @UniTN (Trento, IT)
2/29
The speaker
Francesco Bonadiman [email protected]
Computer Science
Bachelor @UniTN (Trento, IT)
2/29
The outer ear
3/29
The outer ear
3/29
[Raake2014]
The outer ear
3/29
[Raake2014]
catalyzes the sound
The outer ear
3/29
[Raake2014]
catalyzes the sound
The middle ear
[Raake2014]
4/29
The middle ear
[Raake2014]
4/29
The middle ear
[Raake2014]
3 smallest bones in the body
4/29
[Blausen2014] 5/29
2 membrane- covered openings
[Blausen2014] 5/29
2 membrane- covered openings
[Blausen2014] 5/29
2 membrane- covered openings
[Blausen2014] 5/29
2 membrane- covered openings
[Blausen2014] 5/29
2 membrane- covered openings
opposite directions
How it works [Kidsdiscover2014]
6/29
How it works [Kidsdiscover2014]
6/29
difference of sound pressure
How it works [Kidsdiscover2014]
acoustic energy (air) → fluid oscillations
6/29
difference of sound pressure
How it works [Kidsdiscover2014]
impedance matching
acoustic energy (air) → fluid oscillations
6/29
difference of sound pressure
The inner ear
[Amplifon2014] 7/29
The inner ear
[Amplifon2014] 7/29
The inner ear
[Amplifon2014] 7/29
The cochlea: traditional view
8/29
The cochlea: traditional view
[Zwicker,Fastl2007] 8/29
The cochlea: traditional view
[Zwicker,Fastl2007] 8/29
The cochlea
[Wierstorf2014] 9/29
The cochlea
[Wierstorf2014] 9/29
2½ turns for saving space
The cochlea
[Wierstorf2014]
[RobinsonLibrary2014]
9/29
2½ turns for saving space
32 mm “unwound”
The cochlea
[Wierstorf2014]
[RobinsonLibrary2014]
[Wikipedia2014] 9/29
2½ turns for saving space
32 mm “unwound”
The cochlea
[Wierstorf2014]
[RobinsonLibrary2014]
[Wikipedia2014] 9/29
2½ turns for saving space
32 mm “unwound”
The cochlea
[Wierstorf2014]
[RobinsonLibrary2014]
[Wikipedia2014]
helicotrema 9/29
2½ turns for saving space
32 mm “unwound”
Channels and membranes
[Flanagan1972] 10/29
Channels and membranes
[Flanagan1972] 10/29
Channels and membranes
[Flanagan1972] 10/29
hydromechanically → single unit
Channels and membranes
[Flanagan1972] 10/29
Channels and membranes
[Flanagan1972] 10/29
fluids = oscillations
Overview [WhatWhenHow2014]
11/29
Overview [WhatWhenHow2014]
11/29
O H R
Overview [WhatWhenHow2014]
11/29
O H R
Is this enough?
12/29
Organ of Corti
[MixBag2014]
13/29
Organ of Corti
[MixBag2014]
13/29
Organ of Corti
[MixBag2014]
13/29
membranes excitation → nervous impulses (firings)
Organ of Corti
[MixBag2014] [Zwicker,Fastl2007]
13/29
membranes excitation → nervous impulses (firings)
Organ of Corti
[MixBag2014] [Zwicker,Fastl2007]
13/29
membranes excitation → nervous impulses (firings)
Organ of Corti
[MixBag2014] [Zwicker,Fastl2007]
13/29
IHC → brain OHC → tuning
membranes excitation → nervous impulses (firings)
14/29 [GalleryHip2014]
14/29
membranes shearing → HC-cilia → nerve fibers
[GalleryHip2014]
15/29
Frequency-Place-Transformation
[Mandel, Ellis2009] 16/29
Frequency-Place-Transformation
[Mandel, Ellis2009] 16/29
Frequency-Place-Transformation
[Mandel, Ellis2009] 16/29
Frequency-Place-Transformation
[Mandel, Ellis2009] 16/29
BM = kind of Fourier Analysis
Waves and peaks
[Moore2013] 17/29
Waves and peaks
[Moore2013] 17/29
Waves and peaks
[Moore2013] 17/29
Waves and peaks
[Moore2013] 17/29
spatial envelope
Waves and peaks
[Moore2013] 17/29
place + growth ← frequency
spatial envelope
Waves and peaks
[Moore2013] 17/29
place + growth ← frequency
spatial envelope
Waves and peaks
[Moore2013] 17/29
place + growth ← frequency
spatial envelope
Waves and peaks
[Moore2013] 17/29
place + growth ← frequency
spatial envelope
Characteristic Frequencies
[Zwicker,Fastl2007] 18/29
Characteristic Frequencies
[Zwicker,Fastl2007] 18/29
BM’s maximum displacement
Characteristic Frequencies
[Zwicker,Fastl2007] 18/29
BM’s maximum displacement
Mechanical Reasons [Skidmore.edu2014]
19/29
Mechanical Reasons [Skidmore.edu2014]
19/29
stiff thick narrow < 0.1 mm
Mechanical Reasons [Skidmore.edu2014]
19/29
stiff thick narrow < 0.1 mm
flexible thinner wider ∼0.5 mm
Mechanical Reasons [Skidmore.edu2014]
19/29
stiff thick narrow < 0.1 mm
flexible thinner wider ∼0.5 mm
[Wikipedia2014]
Mechanical Reasons [Skidmore.edu2014]
[Wikipedia2014] 19/29
stiff thick narrow < 0.1 mm
flexible thinner wider ∼0.5 mm
Von Békésy (1947) Pioneering work Light microscope
Human cadaver ear
20/29
Masking
21/29 [Wikipedia2014]
Masking
21/29
shows the limits of frequency selectivity (FPT)
[Wikipedia2014]
Masking
21/29
shows the limits of frequency selectivity (FPT)
auditory system → not distinguish between frequencies
[Wikipedia2014]
Masking
21/29
shows the limits of frequency selectivity (FPT)
depends on both signal and masker
auditory system → not distinguish between frequencies
[Wikipedia2014]
Masking
21/29
Detailed talk about it
shows the limits of frequency selectivity (FPT)
depends on both signal and masker
auditory system → not distinguish between frequencies
[Wikipedia2014]
Fletcher Experiment
22/29 [Wikipedia2014]
Fletcher Experiment
22/29 [Wikipedia2014]
masker = 65 dB
masked = 40 dB
Fletcher Experiment
22/29 [Wikipedia2014]
masker = 65 dB
masked = 40 dB
NO!
Fletcher Experiment
22/29 [Wikipedia2014]
masker = 65 dB not masked = 55 dB
masked = 40 dB
NO!
Fletcher Experiment
22/29 [Wikipedia2014]
masker = 65 dB not masked = 55 dB
masked = 40 dB
NO! YES!
Fletcher Experiment
22/29 [Wikipedia2014]
masker = 65 dB not masked = 55 dB
masked = 40 dB
NO! YES!
detect the shape of the so-called
Auditory Filters
Auditory Filters (Critical Bandwidth) [AcousticLab.org2014]
[Wikipedia2014] 23/29
Auditory Filters (Critical Bandwidth) [AcousticLab.org2014]
[Wikipedia2014] 23/29
USUALLY: sounds located singularly
limited range of frequencies different center frequency
Auditory Filters (Critical Bandwidth) [AcousticLab.org2014]
[Wikipedia2014] 23/29
USUALLY: sounds located singularly
limited range of frequencies different center frequency
MASKING: single, broader maximum signal perceived as one same critical bandwidth
BM frequency selectivity = fail
Critical-band Rate Scale
[Zwicker,Fastl2007]
24/29
Critical-band Rate Scale
[Zwicker,Fastl2007]
24/29
new unit leading to critical-band rate scale
Critical-band Rate Scale
[Zwicker,Fastl2007]
24/29
boundaries of the critical bands
new unit leading to critical-band rate scale
Critical-band Rate Scale
[Zwicker,Fastl2007]
24/29
boundaries of the critical bands
new unit leading to critical-band rate scale
audible frequency range in 24 bands
Critical-band Rate Scale
[Zwicker,Fastl2007]
25/29
Critical-band Rate Scale
[Zwicker,Fastl2007]
25/29
from 0 to 24 unit = Bark
(linear)
Critical-band Rate Scale
[Zwicker,Fastl2007]
25/29
higher frequency = bigger amplitude = broader filters from 0 to 24
unit = Bark (linear)
Critical-band Rate Scale
[Zwicker,Fastl2007]
25/29
higher frequency = bigger amplitude = broader filters
mechanical reasons of the membrane
(rigid = high frequencies)
from 0 to 24 unit = Bark
(linear)
Back to the Cochlea
[Zwicker,Fastl2007]
26/29
Back to the Cochlea
[Zwicker,Fastl2007]
26/29
GOLDEN RULE: FPT → Inside our ear, frequencies are mapped onto certain locations
Back to the Cochlea
[Zwicker,Fastl2007]
26/29
SILVER RULE: Mapping not linear but ∼linear up to 500 Hz ∼logarithmic above
GOLDEN RULE: FPT → Inside our ear, frequencies are mapped onto certain locations
27/29
28/29
References ● Brian C.J. Moore, 2003 - An introduction to the psychology of hearing, 6th Edition
● Zwicker and Fastl, 2007 - Psychoacoustics: facts and models, 3rd edition
● Mandel, Ellis, 2009 - Speech & Audio Processing & Recognition - Lecture 4
● Other slides and scripts from professors Raake and Möller
● Huge help given by www.google.com/imghp to find images and references
29/29
AcousticLab.org: 23 Amplifon: 7 Blausen: 5
Flanagan: 10 GalleryHip: 14 Kidsdiscover: 6
Mandel, Ellis: 16 MixBag: 13 Moore: 17
Raake: 3, 4 RobinsonLibrary: 9 Skidmore.edu: 19
WhatWhenHow: 11 Wikipedia: 9, 19, 21, 22, 23 Wierstorf: 9
Zwicker, Fastl: 8 13 18 24 25 26