introduction to velopharyngeal function and...
TRANSCRIPT
Introduction to Velopharyngeal Function and Dysfunction
Xiaochuan [email protected]
Center for Spoken Language UnderstandingOGI School of Science and EngineeringOregon Health and Science University
CSLU Xiaochuan NiuAugust 24, 2006
Slide 3
IntroductionTopic:
Velopharyngeal function and dysfunction
Velopharyngeal (VP) portPhysiological structure >
StatusOpening/ClosureIntermediate positions
nasal tract
oral tract
pharyngeal tract glottal folds
velum
rear pharyngeal wall
CSLU Xiaochuan NiuAugust 24, 2006
Slide 4
IntroductionVelopharyngeal function in normal speech
Appropriate opening of the velopharyngeal (VP) port during nasal consonants (/m n ng/), nasal vowels (in French etc.), nasalized phonemes (due to coarticulation)
Appropriate closure of the VP port during “oral” phonemes, such as plosives and fricatives
What is an appropriate VP function?Right timing and right degree
CSLU Xiaochuan NiuAugust 24, 2006
Slide 5
IntroductionVelopharyngeal dysfunction (VPD) in disordered speech [Witt & Rapley 06]
VP insufficiency: the tissue to accomplish closure of the VP port is insufficient (e.g. cleft palate)
VP incompetence: due to neurological etiologies such as motor disorders (e.g dysarthria)
VP incorrect learning: the result of sensory deficits (e.g. hearing impairment), or congenital disorders (existing at birth)
CSLU Xiaochuan NiuAugust 24, 2006
Slide 6
IntroductionMultiple clinical goals of speech pathology
Differential diagnosis to identify the disease or reason causing the dysfunction
Measurement of speech mechanisms to assess the severity of the dysfunction or the disease
Development of approaches to intervention (behavioral, prosthetic or surgical)
Design of the strategy to reduce or stabilize the disability
CSLU Xiaochuan NiuAugust 24, 2006
Slide 7
IntroductionRelated goals of the research on speech science and technology
To understand the mechanism of both the normal and abnormal speech process
To provide technical batteries for diagnosis, assessment, and rehabilitation of disordered speech
These general goals are applicable for the research on VP function and dysfunction
CSLU Xiaochuan NiuAugust 24, 2006
Slide 8
IntroductionGoals of this talk
To introduce the background knowledge of speech physiology and speech production in order to understand the complexity of VP function during speech
To review some measurement techniques used in the study of VP dysfunction and some research aspects on the production model of nasalization
To present our research work related to VP function
CSLU Xiaochuan NiuAugust 24, 2006
Slide 10
BackgroundCoordinating components of speech production
Respiration: to provide subglottal air pressure
Phonation: to keep tension of vocal folds
Resonance: to open or close nasal and oral cavities
Articulation: to move and shape articulators (tongue, lips, teeth)
Prosody: to convey stress and intonation
VP function affects resonance & interacts with others
CSLU Xiaochuan NiuAugust 24, 2006
Slide 11
BackgroundVP movements during speech
Xray filmsSample #1 (male) & Sample #2 (female)
Endoscopic video in the nasal cavityVelum view & Port view
Combined movements of the velum, rear pharyngeal wall and lateral pharyngeal wall (involving many muscles)
Speaker dependent patterns of movements
CSLU Xiaochuan NiuAugust 24, 2006
Slide 12
BackgroundThe action and location of VP muscles [Laver 80]
1. Soft palate
2. Tongue
3. Thyroid cartilage
4. Palatoglossus muscle
5. Palatopharyngeus muscle
6. Palatal tensor muscle
7. Palatal levator muscle
8. Azygous uvulae muscle
9. Hamular process of the pterygoid bone
CSLU Xiaochuan NiuAugust 24, 2006
Slide 13
BackgroundTerms used in the study of VP function/dysfunction
Nasalization: significant communication of the nasal cavity with the rest of the vocal tract during speech
Nasality: perceptual quality of nasal resonance
Hypernasality: excessive nasally escaping air reverberating in the nasal cavity
Hyponasality: blocked nasal resonance caused by nasal obstruction
Nasal emission: increased nasal instead of oral airflow during the production of pressure consonants (not necessarily acoustic)
Nasal turbulence: fricative sounds caused by nasal airflow
CSLU Xiaochuan NiuAugust 24, 2006
Slide 14
BackgroundResearch and clinical challenges
The position of the VP port is “hidden”It's hard to measure the movement
The perception of nasality is affected by factors besides VP opening
tongue posture / pith and intensity / voice quality
Timing pattern is complex (due to coarticulation)It's almost impossible to discriminate mistiming by listening to the running speech
CSLU Xiaochuan NiuAugust 24, 2006
Slide 15
OutlineIntroduction
Background
ReviewsInstrumental techniques
Acoustics of nasalization
Our work
Summary
CSLU Xiaochuan NiuAugust 24, 2006
Slide 16
Reviews Instrumental techniquesInstrumental techniques [Krakow & Huffman 93, Baken & Orlikoff 00]
Sources of movements
Muscle activities
Movements
ImagingTracking
Effects of movements
Aerodynamic & acoustic consequences
CSLU Xiaochuan NiuAugust 24, 2006
Slide 17
Reviews Instrumental techniquesMuscle activities:
Electromyography (EMG)Surface electrodesIntramuscular electrodes
VP study usually needs the intramuscular ones:
Invasive( Surface EMG )
CSLU Xiaochuan NiuAugust 24, 2006
Slide 18
Reviews Instrumental techniquesImaging
Fiberoptic endoscope
Xray / MRI / Ultrasound
Endoscope: stability
Xray: health risk
MRI: 21ms/frame( MRI: /p a s i/ )
CSLU Xiaochuan NiuAugust 24, 2006
Slide 19
Reviews Instrumental techniquesTracking
Point tracking Radiography (pellets) Magnetometry (coils)
Aperture trackingMechanical
VelotraceStrain gauge
Photodetection (eg. Nasograph)( Magnetmometry )
CSLU Xiaochuan NiuAugust 24, 2006
Slide 20
Reviews Instrumental techniquesAerodynamic & acoustic consequences
AirflowPneumotachograph: split mask with airflow sensors
Sound pressureMicrophones (eg. Probemic, Nasometer)Accelerometers or contact microphones for tissue vibrations
CSLU Xiaochuan NiuAugust 24, 2006
Slide 21
Reviews Instrumental techniquesSummary of instrumental techniques
Research vs. clinical measurement (different goals)
Aspects of practical considerationsInvasiveness / discomfort / health risk / ease of use / time consumed / potability / expense / inherent advantage and limitation / available analysis methods
No “gold standard”: no single measure answer all clinical questions
CSLU Xiaochuan NiuAugust 24, 2006
Slide 22
OutlineIntroduction
Background
ReviewsInstrumental techniques
Acoustics of nasalization
Our work
Summary
CSLU Xiaochuan NiuAugust 24, 2006
Slide 23
Reviews Acoustics of nasalizationResearch on acoustics of nasalization
Speech recognitionAcoustic correlates to nasalization (spectral features)
Speech synthesis / simulationProduction models of nasalization (articulatory or parametric)
Speech analysisModels and algorithms considering the nasal tract for accurate inverse filtering analysis
CSLU Xiaochuan NiuAugust 24, 2006
Slide 24
Reviews Acoustics of nasalizationSome spectral features of nasalization
Reduction in amplitude of the first formant [Fant 60, House & Stevens 56]
Introduction of a polezero pair in the region of the first formant [House & Stevens 56, Maeda 82, Huffman 90]
Extra nasal resonance between 250 and 450Hz [Maeda 82]
Distance between the two spectral peaks in low frequency range [Maeda 93, Chen 97]
CSLU Xiaochuan NiuAugust 24, 2006
Slide 25
Reviews Acoustics of nasalizationProduction models of nasalization
Parametric synthesis [Hawkins & Stevens 85]Sourcefilter model with polezero pair
Articulatory simulation [Maeda 82]Transmissionline model with the extension of nasaltract area function
The production models are useful tools to test hypotheses of the relation between the production and perception of nasalized sounds
CSLU Xiaochuan NiuAugust 24, 2006
Slide 26
Reviews Acoustics of nasalizationAnalysis models and algorithms with the nasal tract assumption
ARMR model of the vocal filter and algorithms to estimate its parameters [Fujisaki & Ljungqvist 87, Liu & Lacroix 96]
Branched vocal tract, lossless tube model [Lim & Lee 93, Schell & Lacroix 02]
CSLU Xiaochuan NiuAugust 24, 2006
Slide 27
Reviews Acoustics of nasalizationSummary of the research on acoustics of nasalization
Focus on the production mechanism or characteristics of normal nasals and nasalized vowels
The analysis target is simply the sound power signal without other additional information
CSLU Xiaochuan NiuAugust 24, 2006
Slide 28
OutlineIntroduction
Background
Reviews
Our workAn airflow device
Dual channel acoustic analysis
Summary
CSLU Xiaochuan NiuAugust 24, 2006
Slide 29
Our work – An airflow deviceMotivations
Simultaneously recording both nasal airflow and acoustic signals without interference with articulation and degradation of the acoustic signal
Device designBernoulli's equation
Pt1=Ps12V2
Pt2=Ps−12V2
Ps Pt1 Pt2
flow
CSLU Xiaochuan NiuAugust 24, 2006
Slide 30
Our work – An airflow device
Data acquisition of DC signals with a sound card:Differential pressure sensor
Processing modules: power supply, offset and gain control, and frequency modulation (FM)
CSLU Xiaochuan NiuAugust 24, 2006
Slide 31
Our work – An airflow deviceDemodulation algorithm
Sampling & lowpass filtering (5kHz)
Discrete Hilbert transform to obtain the analytical signal
Get phase signal by canceling the carrier frequency
Compute the derivative of the phase signal
CSLU Xiaochuan NiuAugust 24, 2006
Slide 32
Our work – An airflow deviceAcoustic, demodulated nasal airflow, and static airflow
CSLU Xiaochuan NiuAugust 24, 2006
Slide 33
Our work – An airflow deviceObservations:
The demodulated airflow signal contain strong harmonic components during the sections of voiced speech sounds;
The filtered DC and lowfrequency components represents the static airflow as it moves in and out of the nostril;
The static nasal airflow peaks occur ...
CSLU Xiaochuan NiuAugust 24, 2006
Slide 34
Our work – An airflow deviceSummary
A quantitative measurement of the static nasal airflow
Nonspeech information, such as inhalation and exhalation
VP information about the detailed timecourse of nasal, nasalized vowel, and nasal emission events during normal speech
Potable, lowcost, less interference with the articulatory process during speech
CSLU Xiaochuan NiuAugust 24, 2006
Slide 35
OutlineIntroduction
Background
Reviews
Our workAn airflow device
Dual channel acoustic analysis
Summary
CSLU Xiaochuan NiuAugust 24, 2006
Slide 36
Our work – Dual channel acoustic analysis
MotivationsTo make full usage of the information from two acoustic channels (with a device like the Nasometer)
To study the relationship between the nasal and oral acoustic signals during nasalization
To derive acoustic features of nasalization from the dual channel analysis
CSLU Xiaochuan NiuAugust 24, 2006
Slide 37
Our work – Dual channel acoustic analysis
Theoretical formulation – transmission line model
po
v
pp
pp
pi
g
U
P
DC
BA=
U
P
mo
ml
mm
mm
mi
v
U
P
DC
BA=
U
P
=
no
nl
nn
nn
ni
v
U
P
DC
BA
U
P
[Flanagan 72]
CSLU Xiaochuan NiuAugust 24, 2006
Slide 38
Our work – Dual channel acoustic analysis
Theoretical formulationTransfer ratio function of the volume velocity from lips to nostrils
T n /m ≝U no
U mo
=Am Z mlBm
An Z nlBn
PropertiesIndependent on the acoustic system below the velumAll poles of the transfer ratio function stem from transfer admittance function of the nasal cavity
CSLU Xiaochuan NiuAugust 24, 2006
Slide 39
Our work – Dual channel acoustic analysis
Analysis methodAssuming the transfer ratio function in the Zdomain has the following form
The parameters are estimation by minimizing the mean square error of the following system
Tn /m
Z =Uno Z
Umo Z =
b0b1 Z−1⋯bN Z N
1a1 Z−1⋯aM Z M
CSLU Xiaochuan NiuAugust 24, 2006
Slide 40
Our work – Dual channel acoustic analysis
Articulatory simulationTo test the analysis method against the synthesized nasalized signals
CSLU Xiaochuan NiuAugust 24, 2006
Slide 41
Our work – Dual channel acoustic analysis
ResultsThe estimated transfer ratio function matches the preset parameters
The polezero pattern of the transfer ration function mainly reflects the property of the nasal tract and the velum opening
CSLU Xiaochuan NiuAugust 24, 2006
Slide 42
Our work – Dual channel acoustic analysis
SummaryThe analysis method provides a new way to nasalization measurement
Our further work will includeApplying the method to dual channel recorded dataDeveloping and test the acoustic feature of nasalization from this approachDeveloping the method into an assessment tool for VP dysfunction
CSLU Xiaochuan NiuAugust 24, 2006
Slide 43
Outline
Introduction
Background
Reviews
Our work
Summary
CSLU Xiaochuan NiuAugust 24, 2006
Slide 44
SummaryUnderstanding the complexity of VP function during speech
Standing on the bridge between speech pathology and speech technology
CSLU Xiaochuan NiuAugust 24, 2006
Slide 45
ReferenceBaken, R.J. & Orlikoff, R.F (2000), Clinical Measurement of Speech and Voice, 2nd edition, Thomson, Delmar Learning.
Chen, M.Y (1997), “Acoustic Correlates of English and French Nasalized Vowels,” J. Acoust. Soc. Am. 102 (4), pp 23602370.
Fant, G. (1960), The Acoustic Theory of Speech Production, Mouton, The Hague.
Flanagan, J.L. (1972), Speech Analysis Synthesis and Perception, 2nd edition, SpringerVerlag
Fujisaki, H. & Ljungqvist, M. (1987), “Estimation of Voice Source and Vocal Tract Parameters Based on ARMA Analysis and a Model for the Glottal Source Waveform,” Proc. IEEE Int. Conf. ASSP 12, pp 637640.
Hawkins, S. & Stevens, K.N. (1985), “Acoustic and Perceptual Correlates of the NonnasalNasal Distinction for Vowels,” J. Acoust. Soc. Am. 77, pp 15601575.
House, A.S. & Stevens, K.N. (1956), “Analog Study of the Nasalization of Vowels,” J. Speech Hear. Dis. 21, pp 218232.
CSLU Xiaochuan NiuAugust 24, 2006
Slide 46
ReferenceHuffman, M. (1990), “Implementation of Nasal: Timing and Articulatory Landmarks,” UCLA Working Papers in Phonetics 75, pp 112143.
Krakow, R.A., & Huffman, M.K. (1993), “Instruments and Techniques for Investigating Nasalization and Velopharyngeal Function in the Laboratory: An Introduction,” in Phonetics and Phonology Volume 5: Nasals, Nasalization and the Velum, edited by M. K. Huffman and R. A. Krakow, Academic, New York, pp359.
Laver, J. (1980), The Phonetic Description of Voice Quality, Cambridge University Press, pp 6892.
Lim, IT. & Lee, B.G. (1993), “Lossless polezero modeling of speech signals,” Trans. IEEE SAP, pp 269276
Liu, M. & Lacroix, A. (1996), “Improved Vocal Tract Model for the Analysis of Nasal Speech Sounds,” Proc. IEEE Int. Conf. ASSP, pp 801804.
Maeda, S. (1982), “The Role of the Sinus Cavities in the Production of Nasal Vowels,” Proc. IEEE Int. Conf. ASSP 2, pp 911914.
CSLU Xiaochuan NiuAugust 24, 2006
Slide 47
ReferenceMaeda, S. (1993), “Acoustic of Vowel Nasalization and Articulatory Shifts in French Nasal Vowels,” in Phonetics and Phonology Volume 5: Nasals, Nasalization and the Velum, edited by M. K. Huffman and R. A. Krakow, Academic, New York, pp 147167.
Schell, K. & Lacroix, A. (2002), “Parameter Estimation of Branched Tube Models by Iterative Inverse Filtering”, Proc. IEEE Int. Conf. DSP, pp 333336.
Witt, P. & Rapley, J. (2006), “Craniofacial, Postpalatoplasty Speech Dysfunction,” eMedicine Specialties.
Website (for some videos and figures in the slides):
http://www.shlrc.mq.edu.au/speech/resources.html
http://www.phonetik.unimuenchen.de/indexeng.html