Presentation on theme: "Structure of Spoken Language"— Presentation transcript:
1 Structure of Spoken Language CS 551/651:Structure of Spoken LanguageLecture 3: Phonetic Symbols and Physiology of Speech ProductionJohn-Paul HosomFall 2010
2 Phonetic Symbols: the IPA The International Phonetic Alphabet (IPA) (reproduced compliments of the International Phonetic Association, Department of Linguistics,University of Victoria, Victoria, British Columbia, Canada)
7 Phonetic Symbols: Worldbet An ASCII representation of IPA, developed by Hieronymous (AT&T)
8 Phonetic Symbols: ARPAbet, TIMITbet, OGIbet ASCII representation of English used in TIMIT corpus.
9 Phonetic Symbols: SAMPA An ASCII representation for multiple (European) languages
10 Acoustic Phonetics: Anatomy nasal tract(hard) palatevelic portoral tractalveolar ridgevelum (soft palate)lipstongueteethpharynxglottis (vocal folds and space between vocal cords)tongue tipvocal folds (larynx)= vocal cordsThe Speech Production Apparatus (from Olive, p. 23)
11 Acoustic Phonetics: Anatomy Breathing and Speech (from Daniloff, chapter 5):Duration of expiration in soft speech is 2.4 to 3.5 seconds; maximum value (singers, orators) is 15 to 20 seconds without distress.Louder voice requires inhaling more deeply after expiration; also deeper inhalation if followed by longer speech.More intense voicing requires greater lung pressure.Lung pressure relatively constant throughout an utterance.Emphasis in speech: greater tenseness in vocal folds yielding higher F0; greater lung pressure increases airflow (energy).
12 Acoustic Phonetics: Anatomy the false vocal folds narrow theglottis during swallowing, preventingpieces of food from getting into thetrachea.
13 Acoustic Phonetics: Anatomy Phonation (from Daniloff, chapter 6):Phonation is “conversion of potential energy of compressed airinto kinetic energy of acoustic vibration.”For voiced speech:1. Buildup of Pressure:air pressure from the lungs pushes against closed vocal foldsso that Psubglottal > Poral; buildup continues untiluntil Psubglottal – Poral > elastic recoil force of vocal folds2. Release:vocal folds forced open by pressure difference;burst of compressed air hits air in vocal tract, causingacoustic shock wave moving along vocal tract
14 Acoustic Phonetics: Anatomy Phonation3. Closure of Vocal Folds, two factors:(a) force of elastic recoil in vocal foldsVocal folds have elastic or recoil force proportional tothe degree of change from the resting position.(b) Bernoulli Effect(i) energy at location of vocal folds is conserved: E = KE + PE(ii) increase in KE causes decrease in PE(iii) PE corresponds to pressure of air(iv) drop in pressure causes walls of glottis to bedrawn closer togetherSummary: air burst causes high rate of airflow, causesdrop in pressure, causes folds to be pulled together
15 Acoustic Phonetics: Anatomy Implications:vocal folds do not open and close because of separate musclemovements2. opening and closing is automatic as long as the resting positionof the vocal folds is (near) closure, and there is sufficientpressure buildup below vocal folds3. Factors governing vocal fold vibration:(a) position of vocal folds (degree of closeness between folds)(b) elasticity of vocal folds, depending on position anddegree of tension(c) degree of pressure drop across vocal folds
16 Acoustic Phonetics: Anatomy Types of phonation (from Daniloff, p. 194)quietbreathingforcedinhalationnormalphonationwhisper
17 Acoustic Phonetics: Anatomy The cycle of glottal vibration (from Daniloff, p. 171)1. folds at rest2. musclecontraction5. “explosion”open6. acousticshockwave3. increase inpressure4. forcing foldsapart7. rebound towardclosure8. folds close,goto step (3)
18 Acoustic Phonetics: Anatomy The cycle of glottal vibration (from Pickett, p. 50)opening to closure, 2.4 to 4.5 msecclosure to opening, 0 to 2.1 msec(F0 = 222 Hz)
19 Acoustic Phonetics: Anatomy Types of phonation (from Daniloff, p. 174)voiceless, whisper, breathyvoiced, creak, glottal stop
20 Acoustic Phonetics: Anatomy Some cool (gross?) videos:Video of fiberoptic stroboscopy exam:(ignore the background music!)And here’s another video fromshowing the vibration of the vocal folds as a person’spitch increases:
21 Acoustic Phonetics: Anatomy The effects of nasalization on vowels (from Pickett, p. 71)Figure An example of theeffects of vowel nasalization onthe vowel spectrum. The spectrumenvelopes of a normal [a] and a heavilynasalized [a] were plotted… The firstthree formants are labeled in thenormal vowel. In the nasalized vowel,there are three local reductions inspectrum level, indicated by “z’s”;these are the result of the additionof anti-resonant zeros to the vocaltract response, due to a wide-openvelar port.
22 Acoustic Phonetics: Anatomy The effects of nasalization on vowels (from Pickett, p. 71)Coupling of the oral and nasal tract introduces pole-zero pairs(resonances & anti-resonances, occurring in pairs) in the spectrum.The amount of coupling affects the spacing between each poleand its corresponding zero, as well as their frequency locations.The presence of a pole-zero pair increases the apparent bandwidth of the neighboring formants.The presence of spectral zero below F1 tends to make the location of F1 appear slightly higher ( Hz) than it normally wouldIf the zero is higher in frequency than its corresponding pole, the net effect is to reduce the amplitude of higher frequencies