Acoustic Characteristics of Consonants Robert A. Prosek, Ph.D.
CSD 301 Robert A. Prosek, Ph.D. CSD 301 Slide 2 Consonants
Consonant articulations are more complex than vowel articulations
consonants are usually described in groups according to their
significant acoustic and articulatory properties stops fricatives
affricates nasals glides liquids Consonant articulations are more
complex than vowel articulations consonants are usually described
in groups according to their significant acoustic and articulatory
properties stops fricatives affricates nasals glides liquids Slide
3 Stop Consonants (1) Stop consonants are characterized by a
complete closure somewhere in the vocal tract Three phases closure
release transition reverse the steps for postvocalic stops Stop
consonants are characterized by a complete closure somewhere in the
vocal tract Three phases closure release transition reverse the
steps for postvocalic stops Slide 4 Stop Consonants (2) Stop gap
this event corresponds to the complete closure of the vocal tract
minimum radiated acoustic energy silence for voiceless stops voice
bar for voiced stops 50 - 150 ms Stop gap this event corresponds to
the complete closure of the vocal tract minimum radiated acoustic
energy silence for voiceless stops voice bar for voiced stops 50 -
150 ms Slide 5 Slide 6 Stop Consonants (3) Stop release (burst)
pressure has been rising behind the obstruction rapid release
produces a transient 20 - 30 ms thus, suitable temporal resolution
is needed voiceless stops follow the burst with frication noise
generated at the place of articulation low frequency for /p/ (500 -
1500 Hz) (falling spectrum) high frequency for /t/ (above 4 kHz)
(rising spectrum) mid-frequency for /k/ (1.5 - 4 kHz) (peaked
spectrum) Stop release (burst) pressure has been rising behind the
obstruction rapid release produces a transient 20 - 30 ms thus,
suitable temporal resolution is needed voiceless stops follow the
burst with frication noise generated at the place of articulation
low frequency for /p/ (500 - 1500 Hz) (falling spectrum) high
frequency for /t/ (above 4 kHz) (rising spectrum) mid-frequency for
/k/ (1.5 - 4 kHz) (peaked spectrum) Slide 7 Slide 8 Stop Consonants
(4) Cues for voicing /p t k/ are phonetically distinguished from /b
d g/ by voicing VOT is the interval between the release of the stop
and the onset of vocal fold vibration for /b d g/ VOT from -20 to
+20 ms with a mean of 10 ms for /p t k/ VOT from 25 to 80 ms with a
mean of 45 ms voice bar for intervocalic stops length of preceding
vowel for final stops Cues for voicing /p t k/ are phonetically
distinguished from /b d g/ by voicing VOT is the interval between
the release of the stop and the onset of vocal fold vibration for
/b d g/ VOT from -20 to +20 ms with a mean of 10 ms for /p t k/ VOT
from 25 to 80 ms with a mean of 45 ms voice bar for intervocalic
stops length of preceding vowel for final stops Slide 9 Slide 10
Slide 11 Stop Consonants (5) Formant transitions articulatory
movement from stop to vowel entails a formant movement as the
resonating chamber of the vocal tract changes, the formant
frequencies change formant transitions are important for perception
formant transitions are approximately 50 ms in duration Formant
transitions articulatory movement from stop to vowel entails a
formant movement as the resonating chamber of the vocal tract
changes, the formant frequencies change formant transitions are
important for perception formant transitions are approximately 50
ms in duration Slide 12 Stop Consonants (6) Formant transitions
(continued) F1 usually rises for the stop consonants F2 and F3 are
not so simple for /p b/ F2 and F3 rise slightly for /t d/ F2 falls
and F3 rises slightly for /k g/ F2 and F3 separate steeply and
rapidly However, a given stop is associated with a variety of
transitions (see Fig. 5-14) there is no fixed transition pattern
for perception Cue trading in stop consonants Formant transitions
(continued) F1 usually rises for the stop consonants F2 and F3 are
not so simple for /p b/ F2 and F3 rise slightly for /t d/ F2 falls
and F3 rises slightly for /k g/ F2 and F3 separate steeply and
rapidly However, a given stop is associated with a variety of
transitions (see Fig. 5-14) there is no fixed transition pattern
for perception Cue trading in stop consonants Slide 13 Fricatives
Articulation Narrow constriction in the vocal tract When air flow
rate is high, turbulence results Turbulence is complex,
unpredictable air flow Turbulent airflow is perceived as turbulent
noise Fricatives have a relatively long duration Fricatives are
divided into sibilants (stridents) nonsibilants (nonstridents)
Articulation Narrow constriction in the vocal tract When air flow
rate is high, turbulence results Turbulence is complex,
unpredictable air flow Turbulent airflow is perceived as turbulent
noise Fricatives have a relatively long duration Fricatives are
divided into sibilants (stridents) nonsibilants (nonstridents)
Slide 14 Fricatives (2) Sibilants Intense noise Differentiated
among themselves by voicing noise spectrum Voicing pulses (glottal
closures) for /z / no pulses for /s / Noise spectrum Alveolar
sibilants have higher frequency energy Palatal sibilants have
energy down to 3 kHz Spectral irregularities arent important in
perception Formant transitions Formant transition locations depend
on the articulation, but the transitions are not important
perceptually for sibilants Sibilants Intense noise Differentiated
among themselves by voicing noise spectrum Voicing pulses (glottal
closures) for /z / no pulses for /s / Noise spectrum Alveolar
sibilants have higher frequency energy Palatal sibilants have
energy down to 3 kHz Spectral irregularities arent important in
perception Formant transitions Formant transition locations depend
on the articulation, but the transitions are not important
perceptually for sibilants Slide 15 Slide 16 Slide 17 Slide 18
Fricatives (3) Nonsibilants /f v h/ Less noise energy than
sibilants Voiced nonsibilants will have quasi-periodic pulses Noise
spectra are fairly flat diffuse The relationship between noise
spectrum nonsibilant identification is not known Formant
transitions play the primary role in perception Noise spectrum may
play a secondary role Nonsibilants /f v h/ Less noise energy than
sibilants Voiced nonsibilants will have quasi-periodic pulses Noise
spectra are fairly flat diffuse The relationship between noise
spectrum nonsibilant identification is not known Formant
transitions play the primary role in perception Noise spectrum may
play a secondary role Slide 19 Slide 20 Slide 21 Fricatives (4)
Acoustical needs for fricatives Measures that are economical
Economical Valid Reliable Problems Ambient noise Filtering values
Acoustical needs for fricatives Measures that are economical
Economical Valid Reliable Problems Ambient noise Filtering values
Slide 22 Affricates Described as a combination of stop and
fricative / / Articulation complete obstruction in the vocal tract
intraoral pressure builds up release to generate fricative noise
Acoustic features rise time duration of frication relative
amplitude in third formant region stop gap Described as a
combination of stop and fricative / / Articulation complete
obstruction in the vocal tract intraoral pressure builds up release
to generate fricative noise Acoustic features rise time duration of
frication relative amplitude in third formant region stop gap Slide
23 Nasals Articulation complete closure in vocal tract sound
radiated through nasal cavities sometimes called nasal stops /m n /
Acoustics Nasal murmur - sound of a nasal associated strictly with
nasal radiation of sound there are many spectral peaks, but most
have low amplitude antiformants nasal formant low frequency (~300
Hz) highest energy Articulation complete closure in vocal tract
sound radiated through nasal cavities sometimes called nasal stops
/m n / Acoustics Nasal murmur - sound of a nasal associated
strictly with nasal radiation of sound there are many spectral
peaks, but most have low amplitude antiformants nasal formant low
frequency (~300 Hz) highest energy Slide 24 Nasals (2) Nasal
formant (continued) consonant energy, overall, is reduced because
higher formants have reduced energy Other acoustic features highly
damped formants (broad bandwidths) formant transitions in connected
speech Nasal formant (continued) consonant energy, overall, is
reduced because higher formants have reduced energy Other acoustic
features highly damped formants (broad bandwidths) formant
transitions in connected speech Slide 25 Slide 26 Glide Consonants
Also called approximants and semivowels /w / Articulation gradual
articulatory motion narrow, but not closed, vocal tract Acoustics
Formants for /w/ F1 and F2 are both low for / / low F1 and high F2
Also called approximants and semivowels /w / Articulation gradual
articulatory motion narrow, but not closed, vocal tract Acoustics
Formants for /w/ F1 and F2 are both low for / / low F1 and high F2
Slide 27 Liquid Consonants Also included as semivowels / l/
Characterized by rapid movements formant structure F3 is the main
difference Antiformants for /l/ Also included as semivowels / l/
Characterized by rapid movements formant structure F3 is the main
difference Antiformants for /l/ Slide 28
