LINGUISTICS 401 LECTURE #3

Topics in Phonetics

ACOUSTIC ANALYSIS: THE SOUND SPECTROGRAPH

Registering a sound by the amplitude of movement of an air particle in the

path of the sound wave – see p. 89

Variations continuously with time – sound wave displays.

Study the Types of acoustical displays Handout.

The most common practice of accounting of any type of sound wave is that
of acoustic analysis or frequency analysis.

1. Sound intensity

Loudness depends on amplitude.

Vibrations of greater amplitude require more energy – consequently,

loudness is related to the energy in the sound: the sound provided with
greater amplitude is more loud.

What about frequency and amplitude in relation to loudness?

If we have two sound waves of equal amplitude with one of them a higher
frequency than the other, the more energy will be needed for the higher
frequency. Why?

Study p. 90

More energy is needed for the sound with higher frequency – it will be
louder, even though the amplitudes are the same.

Loudness depends on the energy in the sound and this is proportional both
the amplitude and frequency.

There is a need for a quantity which will take account of both amplitude and
frequency. This quantity is termed sound intensity.

1
Relationship between (i) amplitude and intensity
Study p. 90
(ii) frequency and intensity

Speech sounds are complex sounds; their intensity bears no simple relation
to the amplitude or frequency of the components.

Most times it is practical to compare the intensities of two different sounds –

thus the unit of intensity has been defined.

The unit of intensity: decibel

Two facts about decibel:

(i) it expresses a relation between two quantities: it is a ratio

(ii) it is a logarithmic unit (the logarithm of the ratio is given)

Intensities are measured in watts.

The energy of the sound waves is small: it has been estimated that it would
require more than 3 million voices talking at once to produce the equivalent
which light a 100 watt lamp.

The faintest sound that the ear can detect is 10-16 watts (0.000000000000000 1 watts);
the upper limit (we can hear without pain is 10-4 watts (0.0001 watts).

Unit: intensity ratio 10:1 bel

It was found that this unit was too large, thus the unit decibel (one-tenth of a
bel) is being used.

Study the formula on p. 92

Study Table 4.

2
2. The relationship between intensity and loudness (a review)

INTENSITY: physical; it can be measured by instruments.

LOUDNESS: psychological; subjective sensation of judged intensity.

Like frequency and pitch, intensity and loudness are not in perfect correlation.

Phon: A unit of equal loudness

STUDY the Handout!

equal loudness levels at different frequencies!

Absolute threshold of audibility (the heavy line in the diagram): the intensities of
each frequency which are just audible to young, healthy ears.

The human auditory system is designed to receive the middle frequencies

(1000-6000 Hz) with much less intensity than is needed for the extremely
low of high frequencies.
Zero setting (on instruments; e.g. sound- level meters)
↓
the heavy line in the diagram

PHON CURVES: The lighter lines in the diagram

The 20-phon line is equally loud at all frequencies to a 1000 Hz tone at 20 dB.

The 70-phon line is equally loud at all frequencies to a 1000 Hz tone at 70 dB.
etc.

At low loudness levels there is a large difference between the middle and the
extreme frequencies in the amount of intensity needed to perceive equal
loudness;

3
At higher loudness levels the large intensity differences disappear.

Sone: unit of loudness

1 sone = loudness to a 1000 Hz tone at 40 dB.

The perception of loudness increases more slowly than the actual increase in
intensity.

3. The relationship between frequency and pitch (a review)

FREQUENCY: physical; it can be measured by instruments.

PITCH: psychological; frequency as perceived by listeners.

The human auditory system is more responsive to some frequency changes than to
others.

LOWER FREQUENCIES (below 1000 Hz): The perceived pitch is linear in

relation to frequency.

HIGHER FREQUENCIES: larger changes in frequency are needed in order to

perceive changes in pitch.

Unit of frequency: The number of cycles per second, Hz

Unit of pitch: mel

Mel: The pitch of a 1000 Hz tone is used as reference → 1000 mel

Whatever frequency is judged to be half of that pitch is called 500 mels; twice that
pitch is 2000 mels.

Mel curve: the solid line in the diagram .

Mel: An auditory (=perceptual) unit of frequency. It reflects the non-linear
properties of the human perception of frequency.

STUDY the Handout!

4
4. The sound spectrograph

Complex tone: fundamental frequency and its harmonics.

Spectrum: represents the relative strength of the components

The sound spectrograph is an instrument that produces a hard copy of

a three-dimensional display of time, frequency and intensity.
It has an adjustable band-pass filter (= resonators!):

Objective: to analyze a complex wave into its component frequencies

by way of employing filters.

Bandpass filter: Study p. 95

Bandwidth: The range of frequencies over which a resonator will respond

effectively.

Study the Continuous spectrum bandwidths Handout.

Important: the resonances of the vocal tract are changing all the time!
Why? Study p. 97

sections: a pattern of formants in one

moment of time
Study Figure 35

WIDE-BAND FILTER has an analyzing bandwidth of 300 Hz;

NARROW-BAND FILTER has an analyzing bandwidth of 45 Hz.

Wide-band filter: detects and distinguishes signals which are close together
in TIME;

Narrow-band filter: detects and distinguishes signals which are close

together in FREQUENCY.

TIME RESOLUTION IS INVERSELY PROPORTIONAL TO

FREQUENCY RESOLUTION.

Study pp. 100-101

5
 Narrow-band filtering is useful for tracking Fo:

if you select a given harmonic, the frequency measured on the y-axis

of that harmonic divided by the harmonic number will be the Fo.

e.g. if the 10th harmonic is at 2000 Hz, Fo is 200 Hz.

Wide-band filtering: the individual harmonics are lost (unless they are
extremely widely spaced as in a small child’s voice), but the resonances of
the vocal tract, THE FORMANTS, are well defined.

TIME appears on the horizontal axis, running from left to right.

FREQUENCY is plotted on the vertical axis, increasing from bottom to top.

INTENSITY is represented by the blackness of the pattern.

THE SPECTROGRAM SHOWS HOW SPECTRAL ENERGY CHANGES

OVER RELATIVELY BRIEF INTERVALS OF TIME.

5. The three forms of the acoustic speech signal

i. THE ACOUSTIC WAVE: The signal that can be heard by the ear or sensed by a
microphone.

The ear or a microphone respond to sound pressure variation in the atmosphere.

The ear converts the air pressure variations into neural impulses that are sent to the brain

for interpretation.

The microphone converts air pressure variations into electrical signals.

Microphones are TRANSDUCERS

converting one form of energy into another

6
Technically, the airborne acoustic signal of speech is called the PROPAGATED or
RADIATED acoustic signal: it propagates or radiates into space after it emerges from a
speaker’s vocal tract.

ii. THE STORED ANALOG SIGNAL

e.g. audio tape recording

An analog signal varies continuously in its pressure (= amplitude) and time properties.

Magnetic tapes store the speech signal as a magnetic field; like the original airborne
acoustic signal, it varies continuously in its properties.

Advantage: it can be played back for listening or analysis:

magnetic energy → converted back into electrical energy → converted into acoustic energy
(e.g. by a loudspeaker).

iii. THE STORED DIGITAL SIGNAL

(digital: numerical)

This form can be stored in a digital computer or on digital magnetic tapes or disks.

To store a speech signal in a digital computer, it is necessary to convert the analog

(continuous) signal into a series of numbers = DIGITIZATION.

AN ANALOG-TO-DIGITAL (A/D) CONVERTER IS A TECHNICAL DEVICE THAT

CHANGES AN ANALOG SIGNAL TO A DIGITAL ONE.

IN THE REVERSE PROCESS, A DIGITAL-TO-ANALOG (D/A) CONVERTER

CHANGES THE DIGITAL SIGNAL TO AN ANALOG FORM (FOR PLAYBACK).

The three forms of speech -- airborne acoustic signal, stored analog signal, and stored digital
signal -- are interchangeable: one form can be converted into another form and back again.