LINGUISTICS 401 LECTURE #1 Topics in Phonetics THE SOURCE-FILTER THEORY OF ...

14 Pages·2011·0.12 MB·English

LINGUISTICS 401 LECTURE #1
Topics in Phonetics

THE SOURCE-FILTER THEORY OF SPEECH PRODUCTION
A review

A. THE GLOTTAL SOUND SOURCE

There are different ways in which the larynx can produce sounds: the
process is called phonation.

Phonation is the use of the laryngeal system, with the help of an airstream
provided by the respiratory system, to generate an audible source of acoustic
energy which can be modified by the articulatory actions of the vocal
apparatus.

PHONATION TYPES: (only two types will be reviewed here)

1. VOICELESSNESS

a. NIL PHONATION

• the vocal folds can be widely
abducted or they can be fully adducted

• when abducted, the glottis is widely open → the acoustic input
to the vocal tract will be zero
↓
if the rate of transglottal airflow is below the level that would
generate local turbulence in the glottis.

A smooth (=laminar) airflow is silent.

A turbulent airflow creates an audible hiss.

1
Whether the flow is laminar or turbulent depends on two factors:
i. rate of airflow
ii. area of the glottis

(A man with a large glottis will require a higher rate of airflow to
reach the threshold of turbulence than a woman or a child).

NIL PHONATION:

silent, smooth (laminar) airflow through the wide open glottis

or

zero airflow with the glottis closed

b. BREATH PHONATION

• The threshold of airflow through a typical adult male glottis for
creating turbulence: 200-300 cc/s (cubic centimetres/second)

• Below this threshold the airflow is laminar: NIL PHONATION

• Above this threshold the airflow is turbulent: AUDIBLE
↓
very gentle, rustling sound
(e.g., hat)

c. WHISPER PHONATION

• The airflow through the glottis is turbulent → hissing quality.

• More intense than in breath phonation -- the vocal folds are
positioned much closer (or completely together) except
between the arytenoids:
triangular opening at the back of the larynx (one
third of the length of the larynx);
2
the pulmonic air flows at high velocity into the
pharynx with considerable turbulence.

• Whisper can also be produced with a slightly more open glottis
→ as long as the wide open position for breath phonation is not
reached.

• Breath and whisper phonations: they represent different
degrees of constriction on a scale of decreasing width of glottis
opening.

Breath phonation: voiceless adjustment of the glottis is set between 60-95%
of the maximum glottal area.

Whisper phonation: the adjustment is less than 25% of the maximum glottal
area; the rate of airflow through the very constricted glottis for whisper is
25-30 cc/s.

2. VOICED PHONATION (= VOICING or MODAL VOICE)

• Acoustically different from the breath and whisper phonation states:

breath phonation

CONTINUOUS acoustic input into the vocal
tract

whisper phonation

voiced phonation: PULSED input, with the frequency of the pulsing
being the product of muscular and aerodynamic
factors

The muscle activity needed to adduct and tense the vocal folds simply makes them
ready for vibration, BUT DOES NOT CAUSE THE VIBRATION ITSELF.

3
The two aerodynamic forces which produce vibration of the vocal folds are:

i. SUBGLOTTAL AIR PRESSURE applied to the lower part of the folds,
forcing them to open (= POSITIVE PRESSURE!)

ii. BERNOULLI EFFECT → The vocal folds cannot stay apart due to the
reduced pressure (= NEGATIVE PRESSURE!)

These positive and negative pressures set the vocal folds vibrating due to the
ELASTICITY OF THE FOLDS.

MYOELASTIC-AERODYNAMIC THEORY OF VOICE
PRODUCTION

Course of events during voiced phonation (study these events from the Handout
and Figure 4-3, p. 50):

• The glottis is closed or nearly closed

• Respiratory pressure from the pulmonic egressive airstream builds up →
subglottal pressure rises.

• When the subglottal pressure becomes high enough to overcome the
muscular forces which are holding the vocal folds together, the vocal folds
are blown slightly apart.

• The compressed air bursts through the narrow gap into the pharynx; the
airflow will reach high velocity because of the narrow constriction.

Consequence: Drop in pressure in the zone of glottal constriction; the high
speed airflow will suck the vocal folds together → BERNOULLI EFFECT!

• The elastic tension of the laryngeal muscles is at work to close the larynx
against the diminishing subglottal pressure → when the combined force of
the myoelastic and aerodynamic factors is sufficient to overcome the force
of the respiratory pressure, the vocal folds snap shut.

4
• in the instant of glottal closure an acoustic shock-wave will travel up
through the vocal tract:

THE ACOUSTIC OUTPUT OF THE LARYNX IS MOST INTENSE
AT THIS POINT
↓
THIS ACOUSTIC OUTPUT PROVIDES MAXIMUM ENERGY TO THE
RESONANCES OF THE VOCAL TRACT.

• With the vocal folds closed, the subglottal pressure, driven by the
respiratory system, begins to rise again → the cycle of events will be
repeated.

• The repetition of the cycle of events is very rapid:
about 120 cycles per second (male)
220 cycles per second (female)

The production of all speech sounds is the result of manipulation of air from the
lungs.

Review the role of the intrinsic laryngeal muscles in the control of
vocal fold behaviour (abduction, adduction and tensioning).

Study Chapter 4 (text and figures!) and the Handouts.

B. SOUNDS, RESONANCE, AND SPECTRUM ANALYSIS

a. SOUND PRODUCTION AND PROPAGATION

Study Chapter 2, pp. 15-17

b. SIMPLE HARMONIC MOTION

Study Chapter 2, pp. 17-18

5
c. DEFINITIONS OF SINE WAVE CHARACTERISTICS

CYCLE: Unit of wave: (=one complete crest and trough)

The number of cycles per second: FREQUENCY

cycles per second (cps) = Hertz (Hz)

e.g., 100 Hz refers to 100 cycles per second.

The time taken for each cycle is termed a PERIOD.
If the frequency is 20 cps, then the period is 1/20th of a second
(= 50 msec)

Algebraically, period and frequency have the relationship:

1
f =
T

where f = frequency
T = the duration of the period

d. PHYSICAL DIMENSIONS OF WAVES

d1. AMPLITUDE: the increase or decrease of air pressure at a
given point during a sound (= the amount of acoustic
energy present!)

The greater the amplitude the higher the crest (and deeper
the troughs) of the waves.

The amplitude indicates the INTENSITY of the sound.

d2. WAVELENGTH: The distance occupied by one cycle.

Study Chapter 2, pp. 18-20

6
e. TEMPORAL DIMENSIONS OF WAVES

e1. FREQUENCY (see above)

e2. SPEED OF PROPAGATION (=velocity of sound)

At normal atmospheric conditions, sound travels through air at
about

344 meters per second

or

1130 feet per second

or

758 miles per hour

Velocity of particle movement must not be confused with
velocity of sound wave propagation!

Particles CHANGE velocity; sound wave
movement is CONSTANT!

Study the Handout.

Zero velocity and maximum acceleration at B, D, F, H, J, L;
(velocity gradually diminishes as the particle approaches these points).

Maximum velocity at the zero crossing, C, E, G, I and K.

e3. PERIOD (see above)

7
f. SPECTRUM ANALYSIS

f1. WAVEFORMS

WAVEFORMS are common representations of sound signals.

A waveform is an amplitude by time display.

• it represents particle motion
• it also represents pressure variation as a function of distance
from the source

The movement of any particle, were it visible, would not look like
a waveform by which it is traditionally represented.

A waveform is an abstract representation of the displacement from rest which
the particle undergoes during a certain time span.

By convention: AMPLITUDE is represented along the ordinate
(y or vertical axis)

TIME is represented along the abscissa
(x or horizontal axis)

Study Figure 2-5, p. 19.

Wave motion can be

(i) PERIODIC: the pattern of vibration repeats itself

(ii) APERIODIC: the vibration has no repeatable pattern.

Strictly speaking, only waves with an infinite number of repetitions are
periodic (DAMPING!)
↓
causing the vibrations to die away

But, in practice, many speech sound waves have enough repetitions to be
regarded as periodic.
Periodic waves are perceived as ‘musical’ (e.g., vowels); aperiodic waves
are perceived as ‘noise’ (e.g., a fricative sound).
8
METHODS USED FOR MAKING THE AIR AUDIBLE

(i). VOICING: The creation of near-periodic sound waves by the rapid opening
and closing of the vocal folds.
The air from the lungs is chopped into tiny puffs of air which are audible.

(ii). CONSONANT NOISE: Part of the speech mechanism is positioned in such
a way that aperiodic sound waves are created in the vocal tract.

(iii). COMBINED METHOD:
Combinations of periodic and aperiodic sound waves are created (voiced
consonants)

Waves can be divided into:

f2. SIMPLE or SINUSOIDAL WAVES

• a representation of PURE TONE
• always periodic

ONE OF THE SIMPLEST KINDS OF VARIATION IN AIR PRESSURE:
PURE TONE

Sounds of the same frequency and in phase (see below) may be added: the
result will be a pure tone (= contains only one frequency)

f3. COMPLEX WAVES

• periodic : if all of its components are periodic

• aperiodic

g. ADDITION OF WAVES:

g1 Adding pure tones of the same frequency: the result will be a pure
tone having the same frequency (phase and amplitude may vary)

9
PHASE: the difference in the timing of the components is known
as a difference in phase.

Study p. 24 and the Handout.

g2. Adding pure tones of different frequencies:
Fourier’s Theorem (Chapter 2, p. 24)
↓
ANY COMPLEX PERIODICAL WAVE CAN BE ANALYZED AS THE
SUM OF ITS COMPONENT FREQUENCIES AND AMPLITUDES.

Study the Handout.
Periodic complex variations produce signals in which the component
frequencies are integral multiples of the lowest frequency of pattern repetition:

FUNDAMENTAL FREQUENCY (Fo)

The frequency of the repetition of a complex wave (p. 28)

HARMONICS: whole-number multiples of the fundamental frequency of the
waveform. (Chapter 2, pp. 28-29)
Why whole-number multiples?

Harmonics (=overtones)

h. LINE SPECTRUM

Another type of display for vibrating patterns is called a LINE
SPECTRUM (or AMPLITUDE SPECTRUM).

horizontal axis : FREQUENCY
vertical axis: AMPLITUDE

Study the Handout.

i. APERIODIC COMPLEX SIGNALS

They consist of more than one frequency, but the frequencies are not
harmonically related.
10