Voice, Speech, Voiced, Unvoiced, Speech Spectrum

1. The Voice Mechanism

Muscular contractions within the chest force air
from the lungs through components of the vocal
mechanism

Stream of air serving as the energy carrier is
modulated in its velocity and pressure to produce
sounds

The modulation is achieved in two distinct ways
giving voiced and unvoiced sounds

2. Fig. 1: Elements of Voice Mechanism

3. Voiced Sounds

Include vowels of ordinary speech and tones
characteristic of the singing voice

Primary modulating agent is the larynx, across
which are stretched the vocal cords -- two
membrane-like bands with a slit-like opening which
modulates the air-stream as it vibrates

Fundamental frequency of modulation is determined
by the tension to which the vocal cords are
stretched

Vocal cords produces a sawtooth type of variation in
pressure

4. Fig. 2a: Pressure output at vocal chords
Fig. 2b: Pressure output for vowel sound ‘a’ as in
father

5. Modulation

Resonating cavities of the nose, mouth and throat
form an acoustic network which modulates the
pressure wave

A variety of voiced sounds are produced by
changing the position of the tongue or altering the
configuration of the lips

Fig. shows how the output at the vocal cords may
be modified so as to produce the vowel sound ‘a’ as
in father

6. Unvoiced sounds

Sounds produced without using the vocal cords!
Also called breath sounds

Produced by steady and forcible exhaling of air,
causing turbulences in the stream through
irregularities along the vocal tract

Include fricative consonants as ‘f’ and ‘s’ ; stop
consonants ‘p’, ‘t’, and ‘k’ - modulation using lips,
teeth and tongue respectively

7. Note

A Fourier of the voiced sounds shows a large
number of harmonically related frequency
components

Analysis of unvoiced sounds reveals a continuous
band of frequencies largely confined to the upper
portion of the audible frequency range

8. Acoustic Power Output
of Speech

Average speech power in conversation is about 10
microwatts, averaged over time intervals of 2-4
seconds

When loud it rises to about 200 microwatts and
upon shouting to about 1000 microwatts

The speech power associated with whispering is
about 0.001 microwatt.

whisper-normal-loud-shout

9. Fluctuations


When power is averaged over a short time interval
compared to the duration of a syllable (about 0.2
sec), large fluctuations in level are observed as
different speech sounds are uttered
e.g. the power of the vowel o as in low is about 50
microwatts, whereas the weak consonant v has an
average power of only 0.03 microwatt.

10. Fig. 3: Spectrum level curve for average
conversational speech re 0 . 0002 microbar at a 1m for
males

11. Speech Spectrum

Average speech spectrum: distribution of speech
power with frequency averaged over wider time
intervals

Ordinate represents pressure spectrum level i.e.
SPL in bands of a single cycle/sec width, measured
at a distance of 1m from the lips of the talker

SPL integrated over entire band from 100 to 10 kHz
is ~ 65 dB

Total acoustic output ~ 20 microwatts: computed
assuming intensity to be uniformly distributed over a
1m radius hemisphere centered on the mouth of the
talker