3. Acoustic Phonetic Basics

2y ago
21 Views
2 Downloads
2.08 MB
54 Pages
Last View : 9d ago
Last Download : 3m ago
Upload by : Braxton Mach
Transcription

Prosody: speech rhythms and melodies3. Acoustic Phonetic BasicsDafydd GibbonSummer SchoolContemporary Phonology and PhoneticsTongji University 9-15 July 2016

Contents The Domains of Phonetics: the Phonetic CycleArticulatory Phonetics (Speech Production)–– Acoustic Phonetics (Speech Transmission)–––––– The IPA (A Alphabet / Association)The Source-Filter Model of Speech ProductionThe Speech Wave-FormBasic Speech Signal ParametersThe Time Domain: the Speech Wave-FormThe Frequency Domain: simple & complex signalsPitch extractionAnalog-to-Digital (A/D) ConversionAuditory Phonetics (Speech Perception)–The Auditory Domain: Anatomy of the Ear

The Domains of Phonetics Phonetics is the scientific discipline which deals with––– speech production (articulatory phonetics)speech transmission (acoustic phonetics)speech perception (auditory phonetics)The scientific methods used in phonetics are––direct observation (“impressionistic”), usually based onarticulatory phonetic criteriameasurement ––of position and movement of articulatory organsof the structure of speech signalsof the mechanisms of the ear and perception in hearingstatistical evaluation of direct observation and measurementscreation of formal models of production, transmission andperceptionAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 20143

The Domains of Phonetics: the Phonetic CycleA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 20144

The Domains of Phonetics: the Phonetic CycleSender:ArticulatoryPhoneticsA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 20145

The Domains of Phonetics: the Phonetic CycleSender:ArticulatoryPhoneticsA tiger and a mouse were walking in a field.0.2628speech signal0-0.253902.86633Time (s)Abidjan 9 May 2014Channel:AcousticPhoneticsDafydd Gibbon: Phonetics Course 3 Abidjan 20146

The Domains of Phonetics: the Phonetic PhoneticsA tiger and a mouse were walking in a field.0.26280speech signal-0.253902.86633Time (s)Abidjan 9 May 2014Channel:AcousticPhoneticsDafydd Gibbon: Phonetics Course 3 Abidjan 20147

Quiz on the Phonetic Cycle Define each of the following:––– articulatory phonetics?acoustic phonetics?auditory phonetics?Which parts of the head are they associated with?What is the “phonetic cycle”?Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 20148

Articulatory Phonetics(Speech Production)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 20149

The articulatory domain Domain of speechproductionArticulatory organsare relatively easilyobservableDomain of referencefor phoneticcategories of the IPAInvestigated via––corpus creationexperimentparadigmAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201410

The IPA (A Alphabet / Association) Abidjan 9 May 2014IPA: 120 years oldregularly re-examined andrevised by Associationbased on articulatorycategoriesdesigned to capture thephonemes of all languages ofthe world: i.e. phoneticdistinctiveness of thecorresponding sound in alanguage of the world is onekey criterion for adopting asymbolDafydd Gibbon: Phonetics Course 3 Abidjan 201411

The Source-Filter Model of Speech Production A “model” is a simplified representation of relevantfeatures of reality (but it also adds its own artefacts)In the Source-Filter Model of speech production, thesound is generated by the SOURCE and modified by theFILTERThe Source-Filter Model represents the speechproduction process in two phases:–The SOURCE of the sound: –LARYNX (for resonant, voiced sounds)CONSTRICTION OF THE ORAL CAVITY (for noisy sounds suchas obstruents)The FILTER through which the sound has passed: Abidjan 9 May 2014the PHARYNGEAL CAVITYthe ORAL CAVITYthe NASAL CAVITYDafydd Gibbon: Phonetics Course 3 Abidjan 201412

The Source-Filter Model of Speech 25391.5561702.866331.65788Time (s)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201413

The Source-Filter Model of Speech onstrictions)Abidjan 9 May 201400-0.05087-0.25391.5561702.866331.65788Time (s)ORALFILTERDafydd Gibbon: Phonetics Course 3 Abidjan 201414

Quiz on Articulatory Phonetics What are the main articulators involved in––– Produce the following consonants, followed by thevowel [a]:–––––– vowel production?consonant production?tone production?voiceless bilabial fricativevoiced alveolar affricatevoiced palatal stopvoiceless labial-velar stopimplosive velar stopvelar nasalWhat is the source-filter model?–Illustrate this, referring to the difference in sound betweenspeaking in a tiled bathroom and in the open air.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201415

Acoustic Phonetics(Speech Transmission)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201416

The acoustic domain Acoustic phonetics is concerned with investigating thetransmission of speech signals through–– The basic parameters of the speech signals are–– amplitudetime (duration)The main derived parameters of speech signals are––– gases such as air, other substances (e.g. bone, tissue)electronic amplification and storageintensitynoise vs. resonance (voicing)frequency and formantsThe methods used to analyse speech signals are:––analog-to-digital (A/D) conversionmathematical definitions of filters and transformationsAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201417

The Speech Wave-Form Speech is transmitted through air (and other substances)as a regular wave of pressure changes:0 .1 1 8 90- 0 .1 5 7 61 .0 1 8 3 81 .0 5 5 3 4T im e ( s ) The changes in air pressure––––but can be heardand cannot be seen (unlike the waves on the ocean)but can be measured (like the waves on the ocean)and the measurements can be visualised and used forcalculating statistical models of the structure of speechAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201418

Visualisation of Speech Signal ParameterstPraat screenshotimeoscillogramamplitudeA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201419

Visualisation of Speech Signal ParameterstPraat screenshotimeformantsfundamental frequencyspectrogrampitch trackA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201420

Visualisation of Speech Signal ParameterstPraat screenshotimeannotation (labelling) on different tiersA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201421

Visualisation of Speech Signal ParameterstPraat al frequencyspectrogrampitch trackannotation (labelling) on different tiersA tiger and a mouse were walking in a field.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201422

The Time Domain: the Speech Wave-FormThe positive or negative amplitude A of the speechsignal at any given point in time is the distance of thewave from zero at this point in time.positive 0 .1 1 8 9negative0- 0 .1 5 7 61 .0 1 8 3 81 .0 5 5 3 4T im e ( s )Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201423

Derived parameter INTENSITY The intensity of the speech signal at any given point intime is the square of the amplitude of the wave fromzero at this point in time:7 5 .3A2I A2-A 0.248705 4 .2 50 .8 2 3 11 .1 1 6 0 3T im e ( s )tiger-0.25390.82311.11603Time (s)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201424

Derived parameter ENERGY The energy E (root-mean-square energy) is–the square root of the mean of a sequence of intensity valuesI1, ., In (remember: intensity is amplitude squared) i 1.n A x i 2E n Energy is therefore intensity averaged over time– In fact, intensity measurements are, in practice, energymeasurements over very short periods of timeCompare other measurement units per time unit: Abidjan 9 May 2014miles per hourkilowatts per hourDafydd Gibbon: Phonetics Course 3 Abidjan 201425

Derived parameters PERIOD & WAVELENGTH The period or interval of a single wave in a speech signalis the duration of this single wave.–– A signal is resonant if its periods are regular in duration.A signal is noisy if its periods are irregular in durationThe wavelength λ (lambda) in metres of a speech signalis the speed of sound in m/sec divided by the number ofperiods per second.A task: Abidjan 9 May 2014What is the speed of sound?What is the wavelength of a sound with 100 periods per second?Dafydd Gibbon: Phonetics Course 3 Abidjan 201426

The Frequency Domain: simple & complex signals The frequency of a speech signal is the number ofwaves (periods) per second in the waveform0.24870-0.25390.82311.11603Time (s)0.15 seconds Question:–tigerIgnoring irregularities: what is the approximate averagefrequency of the segment between the red lines?Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201427

The Simplest Sources produce Sine Waves A sine wave with frequency F is produced by anevenly swinging pendulum – a rather slow sine wave!0.99990-0.999900.0133172Time (s) The speech signal is not a simple sine wave but acomplex signal composed of many sine waves ofdifferent frequencies.Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201428

The Frequency Structure of Speech The SOURCE–– for harmonic, voiced soundsis the larynx (‘voicebox’, ‘Adam’s apple’)The larynx produces:–an approximately triangular complex waveform, consisting of a fundamental frequency–– Abidjan 9 May 2014about 80 Hz - 150 Hz for men (greater range possible)about 160 Hz - 300 Hz for women (greater range possible)many overtones, which are audible up to about 20 kHzdifferent intensities of overtones, relative to each other, whichdetermines the overall waveform, and therefore the timbre orquality of the sound which the source producesduring voicing, the larynx generates a waveform which is ratherlike a “sawtooth” sequenceDafydd Gibbon: Phonetics Course 3 Abidjan 201429

Complex Sources: noisy & harmonic signals If many sine waves of arbitrary frequencies occurtogether, the result is NOISE.If many sine waves occur together, each being aninteger multiple of some lowest frequency,–––the resulting overall wave is a HARMONIC wave:the lowest frequency of a harmonic waveform is thefundamental frequency, F0 (f-zero, f-nought)the higher frequencies in a harmonic waveform are called theharmonics or overtones of the fundamental frequency0.2487NOISYtigerHARMONIC, RESONANT0-0.25390.8231Abidjan 9 May 20141.11603Time (s)Dafydd Gibbon: Phonetics Course 3 Abidjan 201430

Sources with Integer Multiples of Sine Waves Harmonic, resonant frequencies are created by addingseveral sine waves together, point by pointThe larynx sound source is a special case of thisAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201431

Harmonics / overtones in complex signals If a complex signal consists of–a series of sine waves with frequencies of f, 2f, 3f, ., nf ––– e.g. frequencies of 150 Hz, 300 Hz, 450 Hz, 600 Hz, .then the signal is a resonant signaland f is the fundamental frequency F0while 2f, 3f, ., nf are harmonics of the fundamental frequencyStylised example of source signal with harmonicsenergyfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201432

The Spectrum of Complex Signals If a complex signal consists of–a series of sine waves with frequencies of f, 2f, 3f, ., nf ––– e.g. frequencies of 150 Hz, 300 Hz, 450 Hz, 600 Hz, .then the signal is a resonant signaland f is the fundamental frequency F0while 2f, 3f, ., nf are harmonics of the fundamental frequencyStylised example of source signal with harmonicsfundamental frequency, F0energyfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201433

The Spectrum of Complex Signals If a complex signal consists of–a series of sine waves with frequencies of f, 2f, 3f, ., nf ––– e.g. frequencies of 150 Hz, 300 Hz, 450 Hz, 600 Hz, .then the signal is a resonant signaland f is the fundamental frequency F0while 2f, 3f, ., nf are harmonics of the fundamental frequencyStylised example of source signal with harmonicsfundamental frequency, F0energyfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201434

The Spectrum of Complex Signals If a complex signal consists of–a series of sine waves with frequencies of f, 2f, 3f, ., nf ––– e.g. frequencies of 150 Hz, 300 Hz, 450 Hz, 600 Hz, .then the signal is a resonant signaland f is the fundamental frequency F0while 2f, 3f, ., nf are harmonics of the fundamental frequencyStylised example of source signal with harmonicsfundamental frequency, F0harmonics, overtones(integer multiples of F0)energyfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201435

The Spectrum of Complex Signals The filter system consists of pharyngeal, nasal, oralcavities, with resonant frequencies which amplify ordamp the overtones with these frequenciesThese filter frequency bands are called formantsFormant frequencies of the oral cavity can be modifiedby the variable filters (articulators tongue and lips)energyformantsfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201436

The Spectrum of Complex Signals The filter system consists of pharyngeal, nasal, oralcavities, with resonant frequencies which amplify ordamp the overtones with these frequenciesThese filter frequency bands are called formantsFormant frequencies of the oral cavity can be modifiedby the variable filters (articulators tongue and lips)This means that the energy of the harmonics is modifiedfundamental frequency, F0harmonics, overtones(integer multiples of F0)energyformantsfrequencyAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201437

Fourier Analysis: the Spectrum Complex waveforms can be analysed as sums of sinewaves (Joseph Fourier, Fourier Analysis): –7the mathematical operation is the Fourier Transform (FT)the Discrete Fourier Transform (FFG) applies to digitised signalsthe Fast Fourier Transform (FFT) is an optimised versionThe spikes (harmonics) are generated by the SOURCE, andthe peaks (formants) are generated by the FILTER:tiger40FFT200098231Abidjan 9 May 20141.11603Time (s)0Dafydd Gibbon: Phonetics Course 3 Abidjan 20142000Frequency (Hz)38

The Speech Sound Source: sawtooth waveforms The sum of harmonics which are integer multiples, with Ainverse to F, creates a sawtooth waveform:sin i h For x x 1 . x n : x i h 1.mh This example illustrates the sum of four sine waves:100 Hz 200 Hz 300 Hz 400 Hz0.482SUM 0-0.4819Abidjan 9 May 20140Dafydd Gibbon: Phonetics Course 3 Abidjan 2014Time (s)390.01

Fourier Analysis: the Spectrogram A single spectral analysis of an interval in a speechsignal, yields a spectrum and requires a at leat oneperiod:In order to track the changing structure of a speechsignal, a sequence of spectra is needed.–7tigerA representation of a sequence of spectra is called aspectrogramFFT series098231Abidjan 9 May 20141.11603Time (s)Dafydd Gibbon: Phonetics Course 3 Abidjan 201440

Broad band spectrogramAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201441

Narrow band spectrogramAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201442

Spectrogram Filtering: Formants The FILTER which modifies the SOURCE signalconsists of the pharyngeal, nasal and oral cavities.Formants are frequency bands in a spectrogram whichdiffer in intensity from other frequency bands––harmonics in these areas are differ in strengthformants sonorant sounds (vowels, liquids, nasals,approximants)tigerVOWELFORMANTSAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201443

Spectrogram Filtering: Consonantal Noise Obstruent consonants involve–obstruction in the oral tract which causes noise stops: closure of (oral and nasal) tracts, followed by noiseburstfricatives: near-closure of oral tract (and closure of nasal tract)causing noisetigerCONSONANTSAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201444

Pitch extraction Separation of F0 from harmonics is pitch extractionMethods of pitch extraction are:––––counting zero-crossings in the same directioncounting peaks in the signalauto-correlationcepstrum analysistiger25070F0 extraction982311.11603Abidjan 9 May 2014Time (s)750Dafydd Gibbon: Phonetics Course 3 Abidjan 20140.292Time (s)45

Analog-to-Digital (A/D) Conversion In order to enter a speech signal into a computer it isdigitised:–––the signal is sampled regularly and the amplitude of the sampleis measured automaticallythe speed with which the measurements are made is called thesampling ratestandard sampling rates are: 44.1 kHz (CDs) 2 x 2 x 3 x 3 x 5 x 5 x 7 x 7 (prime numbers)48 kHz (DAT tapes)22.05 kHz (laboratory recordings). (other sampling rates, e.g. 16 Hz, are also found)The minimum sampling rate is twice the frequency of thehighest harmonic in the signal (Nyquist theorem),otherwise false measurements are made and “aliasing”occurs (ghost frequencies)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201446

Analog-to-Digital (A/D) Conversion The corners in the visualisation represent measuringpointsThe measuring points are joined by straight lines to givean impression of continuityAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201447

Quiz on Acoustic Phonetics What are the basic parameters of the speech signal?Define the following terms:––– amplitudeintensityenergyHow are time-domain representations of speech signalconverted to frequency domain representations?Define the following ntal frequency, F0, pitchharmonicformantanalog-to-digital conversionAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201448

Auditory Phonetics(Speech Perception)Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201449

The Auditory Domain: Anatomy of the EarAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201450

The Auditory Domain: Anatomy of the EarAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201451

The Auditory Domain: Anatomy of the EarmicrophoneAbidjan 9 May 2014amplifierFourier transformDafydd Gibbon: Phonetics Course 3 Abidjan 201452

Quiz on Auditory Phonetics What are the functions of––– the outer ear?the middle ear?the inner ear?What are––––the ossicles?the oval window?the cochlea?the basilar membrane?Abidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201453

Final RemarksAfter the first unit––you should have learned the basic theoretical foundations onwhich phonetic activities with Praat are basedyou should be able to use a Praat TextGrid file with the TGAonline timing analysis toolAfter the second unit–you should thoroughly understand what you are doing withPraatAbidjan 9 May 2014Dafydd Gibbon: Phonetics Course 3 Abidjan 201454

Abidjan 9 May 2014 Dafydd Gibbon: Phonetics Course 3 Abidjan 2014 3 The Domains of Phonetics Phonetics is the scientific discipline which deals with – speech production (articulatory phonetics) – speech transmission (acoustic phonetics) – speech perception (auditory phonetics) The scientific methods used

Related Documents:

Phonetic Transcription Articulation of Sounds Phonetic Alphabet Transcription Transcription Notes Phonetic transcriptions are written in square brackets [ ]. Transcribe words based on sound, not spelling. Don’t use a schwa ([@]) in stressed syllables. Upper- and lowercase letters are not interchangeable. Some morphemes, like past tense -ed .File Size: 2MB

Play Acoustic – Reference manual (2014-05-09) 7 Welcome to the Play Acoustic manual! First, thank you so much for purchasing Play Acoustic. We at TC-Helicon are confident that your vocal and acoustic guitar performances will be positively impacted with this great effects processor. As you discovered in the Quick Start Guide (the

TECHNOLOGY RISK REDUCTION Developing next-generation acoustic core Improve acoustic performance through unique non-conventional geometries Large acoustic cell configuration in development Producible/Cost-effective large acoustic cavity configuration Producible design concepts for acoustic testing “On Hold” until 2021 . Large .

Introduction 1 An Introduction to Acoustic Emission—/?. B. Liptai, D. O. Harris, and C. A. Tatro 3 Research on the Sources and Characteristics of Acoustic Emission—fi. H. Schofield 11 Dislocation Motions and Acoustic Emissions—P. P. Gillis 20 Acoustic Emission Testing and Microcracking Processes—y4. S. Tetelman and R. Chow 30

Welcome to Variax Acoustic Thanks for buying a Variax Acoustic and joining us in our quest to apply the miracle of modern technology to the pursuit of great acoustic guitar tone. You now own detailed models of some of the most distinctive acoustic instruments of all time–wrapped up in a single comfortable and highly playable guitar. How does .

AT, an Acoustic Transmissometer Albert J. Williams 3rd Woods Hole Oceanographic Institution MS#12, 98 Water St. Woods Hole, MA 02543 USA Abstract-The combination of attenuation measurement with acoustic travel-time current measurement along a common path has produced a new acoustic sensor of suspended particles, the Acoustic Transmissometer (AT).

As a result of this physical nature of acoustic waves, the composition of the material through which an acoustic wave travels will impact its speed and the energy that is lost due to absorption as the wave propagates through the mate-rial. When a propagating acoustic wave encounters a sudden change in the acoustic impedance (product of sound speed

A.4. Performance analysis - Consideration of variable amplitude acoustic emission sources Only the detectability of an acoustic emission source equivalent to a Hsu-Nielsen source (0.5 mm - 2H) was considered in the previous calculations. It can be assumed that detectable acoustic emission sources in a real structure do not necessarily give