Speech soundSpeech productionMeasuring speech
Today's kick-o
Midterm last year:
1
True/False
2
Multiple choice
3
Short answer: Groups, tests, tree forms, reasons, etc.
Lecture
In-class exercise
2 / 99
Speech soundSpeech productionMeasuring speech
Today's kick-o
Midterm last year:
1
True/False
2
Multiple choice
3
Short answer: Groups, tests, tree forms, reasons, etc.
Lecture
In-class exercise
3 / 99
Speech soundSpeech productionMeasuring speech
Today's kick-o
Midterm last year:
1
True/False
2
Multiple choice
3
Short answer: Groups, tests, tree forms, reasons, etc.
Lecture
In-class exercise
4 / 99
Speech soundSpeech productionMeasuring speech
I.
Speech sound
5 / 99
Speech soundSpeech productionMeasuring speech
Consonants
Last week we talked aboutconsonantsand their IPA symbols.
There are a few criteria/features to describe consonants:
1
Place of articulation(labial, alveolar, velar, ... etc.)
2
Manner of articulation(stops, fricatives, aricates, ... etc.)
3
Voicing(voiced vs. voiceless)
4
Aspiration(aspirated vs. unaspirated)
Note that Mandarin only has aspiration contrast not voicing
contrast.
)[p];[p
h
]
In comparison, Taiwan Southern Min contrasts both aspiration
and voicing.
)8[p];S[p
h
];‰[b]
6 / 99
Speech soundSpeech productionMeasuring speech
Consonants
Last week we talked aboutconsonantsand their IPA symbols.
There are a few criteria/features to describe consonants:
1
Place of articulation(labial, alveolar, velar, ... etc.)
2
Manner of articulation(stops, fricatives, aricates, ... etc.)
3
Voicing(voiced vs. voiceless)
4
Aspiration(aspirated vs. unaspirated)
Note that Mandarin only has aspiration contrast not voicing
contrast.
)[p];[p
h
]
In comparison, Taiwan Southern Min contrasts both aspiration
and voicing.
)8[p];S[p
h
];‰[b]
7 / 99
Speech soundSpeech productionMeasuring speech
Consonants
Last week we talked aboutconsonantsand their IPA symbols.
There are a few criteria/features to describe consonants:
1
Place of articulation(labial, alveolar, velar, ... etc.)
2
Manner of articulation(stops, fricatives, aricates, ... etc.)
3
Voicing(voiced vs. voiceless)
4
Aspiration(aspirated vs. unaspirated)
Note that Mandarin only has aspiration contrast not voicing
contrast.
)[p];[p
h
]
In comparison, Taiwan Southern Min contrasts both aspiration
and voicing.
)8[p];S[p
h
];‰[b]
8 / 99
Speech soundSpeech productionMeasuring speech
Vowels What are the (N. American) English vowels?[i, e,E, , a,@, u,U, o,2,O,A] What are vowels? Vowels: Sounds (with vocal folds vibrating) produced with at
most only a slight narrowing somewhere in the vocal tract,
allowing air to ow freely through the oral cavity.
9 / 99
Speech soundSpeech productionMeasuring speech
Vowels What are the (N. American) English vowels?[i, e,E, , a,@, u,U, o,2,O,A] What are vowels? Vowels: Sounds (with vocal folds vibrating) produced with at
most only a slight narrowing somewhere in the vocal tract,
allowing air to ow freely through the oral cavity.
10 / 99
Speech soundSpeech productionMeasuring speech
Vowels What are the (N. American) English vowels?[i, e,E, , a,@, u,U, o,2,O,A] What are vowels? Vowels: Sounds (with vocal folds vibrating) produced with at
most only a slight narrowing somewhere in the vocal tract,
allowing air to ow freely through the oral cavity.
11 / 99
Speech soundSpeech productionMeasuring speech
Vowels What are the (N. American) English vowels?[i, e,E, , a,@, u,U, o,2,O,A] What are vowels? Vowels: Sounds (with vocal folds vibrating) produced with at
most only a slight narrowing somewhere in the vocal tract,
allowing air to ow freely through the oral cavity.
12 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Traditional view of vowels
1
Height
2
Backness
13 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Traditional view of vowels
1
Height
2
Backness
14 / 99
Speech soundSpeech productionMeasuring speech
Formants
Formants: Resonant frequency
that amplies some groups of
harmonics above others.
Appears as a dark band on a
spectrogram.
Transitions into and out of the
vowels.
15 / 99
Speech soundSpeech productionMeasuring speech
Formants
Since 60s, linguists found a strong
correlation between the tongue
position and formants.
Dierent vowels are characterized by
dierent formant proles.
(Figure adapted from Goldstein (2010), Fig. 13.2)
By changing its shape, the vocal
tract create distinct patterns of
resonating formant frequencies for
each vowel.
)Source-lter theory
In a gure of spectral slice(frequency
on the x-axis and amplitude on the y-axis),
the rst peak is dened asF1, the
second peak is termedF2, and so
on.
16 / 99
Speech soundSpeech productionMeasuring speech
Formants
Since 60s, linguists found a strong
correlation between the tongue
position and formants.
Dierent vowels are characterized by
dierent formant proles.
(Figure adapted from Goldstein (2010), Fig. 13.2)
By changing its shape, the vocal
tract create distinct patterns of
resonating formant frequencies for
each vowel.
)Source-lter theory
In a gure of spectral slice(frequency
on the x-axis and amplitude on the y-axis),
the rst peak is dened asF1, the
second peak is termedF2, and so
on.
17 / 99
Speech soundSpeech productionMeasuring speech
Tongue and formants
18 / 99
Speech soundSpeech productionMeasuring speech
Tongue and formants
19 / 99
Speech soundSpeech productionMeasuring speech
Tongue and formants
20 / 99
Speech soundSpeech productionMeasuring speech
Tongue and formants
Classical formant-cavity aliation:
F1 { back cavity
F2 { front cavity
21 / 99
Speech soundSpeech productionMeasuring speech
Tongue and formants
English vowels (high front [i], high
back [u], and low back [A]) from
MRI data: right midsagittal view
Tongue height and tongue advancement
Tongue height: High vowels (/i/, /u/))
tongue moves out of the pharyngeal area)
increase air volume in the pharyngeal area
)lowers F1
Tongue advancement: Back vowels (/A/,
/u/))far back in the oral cavity)
lengthens the anterior oral cavity)
lowers F2
NOTE: Lip protrusion can also increase
the anterior oral cavity!
22 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Now you understand why the IPA
vowel chart is arranged this way.
But, some of the details remained
unexplained.
I know [i], but what is [y]?
I know [u], [U], but what is [W]?
This vowel chart also encodes
another vowel quality:
roundedness
23 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Now you understand why the IPA
vowel chart is arranged this way.
But, some of the details remained
unexplained.
I know [i], but what is [y]?
I know [u], [U], but what is [W]?
This vowel chart also encodes
another vowel quality:
roundedness
24 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Now you understand why the IPA
vowel chart is arranged this way.
But, some of the details remained
unexplained.
I know [i], but what is [y]?
I know [u], [U], but what is [W]?
This vowel chart also encodes
another vowel quality:
roundedness
25 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Roundedness
[i] and [y] contrast in roundedness,
so do [u] and [W].
)[i] is [-round]
)[y] is [+round]
Try this:¨(“;ýø)
26 / 99
Speech soundSpeech productionMeasuring speech
About vowels
Roundedness
[i] and [y] contrast in roundedness,
so do [u] and [W].
)[i] is [-round]
)[y] is [+round]
Try this:¨(“;ýø)
27 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Larynx is a complicated system.
By vibrating the vocal folds (VF), we
produce (voiced) sounds.
Unsurprisingly, we can change the way
we vibrate our VF to make dierent
sound qualities (i.e., phonations).
Modal
Breathy
Creaky
Whisper
Falsetto
29 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Larynx is a complicated system.
By vibrating the vocal folds (VF), we
produce (voiced) sounds.
Unsurprisingly, we can change the way
we vibrate our VF to make dierent
sound qualities (i.e., phonations).
Modal
Breathy
Creaky
Whisper
Falsetto
30 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Modal
Characterized as most common form of
voicing, with (more or less) a balance of
closed and open vocal fold frequency.
Breathy
Combines voicing with glottal friction.
Forms a continuum with modal voice.
31 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Modal
Characterized as most common form of
voicing, with (more or less) a balance of
closed and open vocal fold frequency.
Breathy
Combines voicing with glottal friction.
Forms a continuum with modal voice.
32 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Creaky
LowerF0
Quite ecient, used when tired
Whisper
voiceless speech, sometimes with
high4P
Very insucient - run out of air
quickly
33 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Creaky
LowerF0
Quite ecient, used when tired
Whisper
voiceless speech, sometimes with
high4P
Very insucient - run out of air
quickly
34 / 99
Speech soundSpeech productionMeasuring speech
Larynx and phonations
Falsetto
Makes the vocal fold very thin and
high-frequency
High medial compression and
longitudinal tension
35 / 99
Speech soundSpeech productionMeasuring speech
II.
Speech production
36 / 99
Speech soundSpeech productionMeasuring speech
Speech production
1
Denition ofspeech production
(in terms of motor control)?Wiki: \Speech production is the
process by which thoughts are
translated into speech. This includes
the selection of words, the
organization of relevant grammatical
forms, and then the articulation of
the resulting sounds by the motor
system using the vocal apparatus."
37 / 99
Speech soundSpeech productionMeasuring speech
Speech production
1
Denition ofspeech production
(in terms of motor control)?Wiki: \Speech production is the
process by which thoughts are
translated into speech. This includes
the selection of words, the
organization of relevant grammatical
forms, and then the articulation of
the resulting sounds by the motor
system using the vocal apparatus."
38 / 99
Speech soundSpeech productionMeasuring speech
Speech production
1
Denition ofspeech production
(in terms of motor control)?Wiki: \Speech production is the
process by which thoughts are
translated into speech. This includes
the selection of words, the
organization of relevant grammatical
forms, and then the articulation of
the resulting sounds by the motor
system using the vocal apparatus."
39 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
46 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
47 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
48 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
49 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
50 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
51 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Speech production requires a sequence of well-coordinated
articulatory movements. For example, to produce the wordpalm
/p
h
am/, a speaker needs to
1
shut the mouth for a bilabial closure
2
pump up intraoral pressure,
3
open the glottis; burst open the lips
4
make sure the tongue is at its designated position during the
bilabial burst, in this case, low-back position
5
leave the velopharyngeal port open in order to make it nasal
6
close the mouth again for the /m/ closure while the vocal folds
keep vibrating
52 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
We can visualize this overlapped yet coordinated \gestures,"
)This is the essence of \Articulatory Gestures."(Browman and
Goldstein, 1989)
)Dierent tiers of gesture scores
53 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
We can visualize this overlapped yet coordinated \gestures,"
)This is the essence of \Articulatory Gestures."(Browman and
Goldstein, 1989)
)Dierent tiers of gesture scores
54 / 99
Speech soundSpeech productionMeasuring speech
Speech behaviours
Connected speech (usually) involves
contraction (i.e., some kind of
omission).
Note: You can't have complete
contraction. (Think about boxing
combinations.)
55 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"!
Saturdayhas a very unique
sequence: [RõR]
In English, there are two ways to
produce rhotics: tongue tip-up [õ]
and tongue tip-down [ô].(Derrick et
al., 2015)
There are four identiable /R/ in
English.(Derrick et al., 2015)
1
alveolar tap
2
down-ap
3
up-ap
4
postalveolar tap
56 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"!
Saturdayhas a very unique
sequence: [RõR]
In English, there are two ways to
produce rhotics: tongue tip-up [õ]
and tongue tip-down [ô].(Derrick et
al., 2015)
There are four identiable /R/ in
English.(Derrick et al., 2015)
1
alveolar tap
2
down-ap
3
up-ap
4
postalveolar tap
57 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"!
Saturdayhas a very unique
sequence: [RõR]
In English, there are two ways to
produce rhotics: tongue tip-up [õ]
and tongue tip-down [ô].(Derrick et
al., 2015)
There are four identiable /R/ in
English.(Derrick et al., 2015)
1
alveolar tap
2
down-ap
3
up-ap
4
postalveolar tap
58 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"!
Saturdayhas a very unique
sequence: [RõR]
In English, there are two ways to
produce rhotics: tongue tip-up [õ]
and tongue tip-down [ô].(Derrick et
al., 2015)
There are four identiable /R/ in
English.(Derrick et al., 2015)
1
alveolar tap
2
down-ap
3
up-ap
4
postalveolar tap
59 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"! Phonetically, is the retroex produced as [õ] or as [ô]?
Phonetically, is this sequence produced as [R
-
õR
&
] or as [R
l
õR
l
]? Derrick et al.(2015)predict an up-down sequence: R
-
õR
&
]
60 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"! Phonetically, is the retroex produced as [õ] or as [ô]?
Phonetically, is this sequence produced as [R
-
õR
&
] or as [R
l
õR
l
]? Derrick et al.(2015)predict an up-down sequence: R
-
õR
&
]
61 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about \Saturday"! Phonetically, is the retroex produced as [õ] or as [ô]?
Phonetically, is this sequence produced as [R
-
õR
&
] or as [R
l
õR
l
]? Derrick et al.(2015)predict an up-down sequence: R
-
õR
&
]
62 / 99
Speech soundSpeech productionMeasuring speech
Let's talk about "Saturday" The rhotics in \Saturday" is more likely to be produced as a
tongue tip-up retroex [õ].
Most of the rst ap (T variant) is [R
-
] (191/213).
Of the 193 tokens ofSaturdayproduced with [õ], fully 187
ended with [R
&
].
64 / 99
Speech soundSpeech productionMeasuring speech
Speech pre-plans
Two possible approaches:
1
One holistic event (i.e., [R
-
õR
&
])
2
Two (or multiple) independent events (i.e., [R
l
õR
l
])
)Derrick et al. (2015) use ultrasound and 3D model simulation.
They concluded: R
-
õR
&
]
)There ISpre-planning!
65 / 99
Speech soundSpeech productionMeasuring speech
Speech pre-plans
Two possible approaches:
1
One holistic event (i.e., [R
-
õR
&
])
2
Two (or multiple) independent events (i.e., [R
l
õR
l
])
)Derrick et al. (2015) use ultrasound and 3D model simulation.
They concluded: R
-
õR
&
]
)There ISpre-planning!
66 / 99
Speech soundSpeech productionMeasuring speech
III.
Measuring speech
67 / 99
Speech soundSpeech productionMeasuring speech
Palatography; linguography
Palatography
A way to record contact between the tongue and the roof the
mouth to get articulatory records for the production of speech
sounds.
Apply marker paint on the tongue, produce the word, and see
where ink rubbed o onto the roof of the mouth.
)palatograms.
Since palatography requires contact, i.e. some kind of
obstruction, for the paint to rub o, palatography is most
suitable for looking atconsonants.
69 / 99
Speech soundSpeech productionMeasuring speech
Palatography; linguography
An example of palatogram:
70 / 99
Speech soundSpeech productionMeasuring speech
Palatography; linguography
Palatography
Merits:
Easy and safe; suitable for eld work.
(Relative) More comprehensive view of the lingual contact.
Downsides:
No timing information.
Careful speech; does not allow errors.
Not suitable for sounds with no lingual contact.
Only captures the production of a single sound, not a sequence of
sounds.
71 / 99
Speech soundSpeech productionMeasuring speech
Palatography; linguography
Palatography
Merits:
Easy and safe; suitable for eld work.
(Relative) More comprehensive view of the lingual contact.
Downsides:
No timing information.
Careful speech; does not allow errors.
Not suitable for sounds with no lingual contact.
Only captures the production of a single sound, not a sequence of
sounds.
72 / 99
Speech soundSpeech productionMeasuring speech
Palatography; linguography
Linguography
Palatography applies marker paint on the tongue
and see where the ink stays on the palateafter
the contact.
Linguographyapplies marker paint on the
palate and see where the ink stays on the surface
of the tongue.
Everything else is identical to palatography.
Palatographyfocuses on where in the palate the
contact was made whereaslinguographyfocuses
on which part of the tongue is making the
contact.
73 / 99
Speech soundSpeech productionMeasuring speech
EGG
Electroglottography (EGG)
Non-invasive measurement of laryngeal
movements.
Easy set-up.
Measuring the resistance between the two
electrodes placed outside the thyroid
notch.
Very good temporal resolution.
Bad location information)Don't know
where the opening of contact is or the
width of the opening
Raising or lower the larynx may cause
loss of data acquisition.
74 / 99
Speech soundSpeech productionMeasuring speech
General Impression
When talking aboutultrasound, we
think about this:
And ultrasound images should look like
this:
75 / 99
Speech soundSpeech productionMeasuring speech
General Impression
When talking aboutultrasound, we
think about this:
And ultrasound images should look like
this:
76 / 99
Speech soundSpeech productionMeasuring speech
General Impression
When talking aboutultrasound, we
think about this:
And ultrasound images should look like
this:
77 / 99
Speech soundSpeech productionMeasuring speech
General Impression
When talking aboutultrasound, we
think about this:
And ultrasound images should look like
this:
78 / 99
Speech soundSpeech productionMeasuring speech
General Impression
When talking aboutultrasound, we
think about this:
And ultrasound images should look like
this:
79 / 99
Speech soundSpeech productionMeasuring speech
Basics about ultrasound
The transducer produces high-frequency
sound (ultra-high, in fact!) ranging from
316 MHz.(We human can only hear 2020
KHz sounds!)
Some of these sound waves penetrate
skin, fat, and muscle(parts of the body that
hold a lot of water), while some reect back
to the transducer.
The sound waves get absorbed by bone
and reect sharply o of air boundaries,
so that ultrasound doesn't image bone or
air very well.
80 / 99
Speech soundSpeech productionMeasuring speech
Basics about ultrasound
The transducer produces high-frequency
sound (ultra-high, in fact!) ranging from
316 MHz.(We human can only hear 2020
KHz sounds!)
Some of these sound waves penetrate
skin, fat, and muscle(parts of the body that
hold a lot of water), while some reect back
to the transducer.
The sound waves get absorbed by bone
and reect sharply o of air boundaries,
so that ultrasound doesn't image bone or
air very well.
81 / 99
Speech soundSpeech productionMeasuring speech
Basics about ultrasound
The transducer produces high-frequency
sound (ultra-high, in fact!) ranging from
316 MHz.(We human can only hear 2020
KHz sounds!)
Some of these sound waves penetrate
skin, fat, and muscle(parts of the body that
hold a lot of water), while some reect back
to the transducer.
The sound waves get absorbed by bone
and reect sharply o of air boundaries,
so that ultrasound doesn't image bone or
air very well.
82 / 99
Speech soundSpeech productionMeasuring speech
Basics about ultrasound
Compared to X-ray and other imaging
techniques, ultrasound machines don't
produce ionizing radiation.
Ultrasound provides immediate images
(wonderful temporal resolution)across large
areas(fairly good spatial resolution).
Very easy to use; can be portable
)Great fortongueimaging!
83 / 99
Speech soundSpeech productionMeasuring speech
B-mode; B/M-mode
B-mode
Bi-dimenstional:
For tongue, phoneticians care
about:
- height
- backness (frontness)
B/M-mode
B-mode plusmovementon ONE
dimension, along the
line
84 / 99
Speech soundSpeech productionMeasuring speech
B-mode; B/M-mode
B-mode
Bi-dimenstional:
For tongue, phoneticians care
about:
- height
- backness (frontness)
B/M-mode
B-mode plusmovementon ONE
dimension, along the
line
85 / 99
Speech soundSpeech productionMeasuring speech
B-mode; B/M-mode
B-mode
Bi-dimenstional:
For tongue, phoneticians care
about:
- height
- backness (frontness)
B/M-mode
B-mode plusmovementon ONE
dimension, along the
line
86 / 99
Speech soundSpeech productionMeasuring speech
B-mode; B/M-mode
B-mode
Bi-dimenstional:
For tongue, phoneticians care
about:
- height
- backness (frontness)
B/M-mode
B-mode plusmovementon ONE
dimension, along the
line
87 / 99
Speech soundSpeech productionMeasuring speech
Data
Aligning ultrasound images with sounds.
88 / 99
Speech soundSpeech productionMeasuring speech
Data processing
1
Grab still images
2
Tongue surface tracing
89 / 99
Speech soundSpeech productionMeasuring speech
Data processing
1
Grab still images
2
Tongue surface tracing
90 / 99
Speech soundSpeech productionMeasuring speech
Data processing
4
Fitting(Smoothing spline analysis of variance; SS ANOVA)
92 / 99
Speech soundSpeech productionMeasuring speech
Data processing
4
Fitting(Generalized Additive Mixed Model; GAMM)
Figure adapted from Chiu & Sun (2020)
93 / 99
Speech soundSpeech productionMeasuring speech
Data processing
4
Fitting(Generalized Additive Mixed Model; GAMM)
Figure adapted from Chiu & Sun (2020)
94 / 99
Speech soundSpeech productionMeasuring speech
Challenges
Transducer stabilization problem:
Frame rate and resolution tradeo. Some of the images may be grainy.
95 / 99
Speech soundSpeech productionMeasuring speech
Challenges
Transducer stabilization problem:
Frame rate and resolution tradeo. Some of the images may be grainy.
96 / 99
Speech soundSpeech productionMeasuring speech
Challenges
Transducer stabilization problem:
Frame rate and resolution tradeo. Some of the images may be grainy.
97 / 99
Speech soundSpeech productionMeasuring speech
Challenges
Transducer stabilization problem:
Frame rate and resolution tradeo. Some of the images may be grainy.
98 / 99
Speech soundSpeech productionMeasuring speech
References
Chiu, C., & Sun, J. T.-S. (2020). On pharyngealized vowels in
Northern Horpa: An acoustic and ultrasound study.Journal of the
Acoustical Society of America,147(4), 2928{2946. doi:
10.1121/10.0001005
Derrick, D., Stavness, I., & Gick, B. (2015). Three speech rounds, one
motor action: Evidence for speech-motor disparity from English
ap production.Journal of the Acoustical Society of America,
137(3), 1493 - 1502. doi: 10.1121/1.4906831
99 / 99