4.2 Source Filter Model of Speech Production

vocal tract made up of

vocal folds (soud nosurce) and resonating cavities: glottal, pharyngeal and nasal

production of sound has 3 steps

initiation from energy source- vocal fold vibration or via turbulence of air through a constriction

resonance of sound producer

radiation of sound into air

Comes out of our oral cavity or nasal cavity via pharyngeal cavity after going through filters.

Depending on the length or the shape of these different cavities that sound is passing through, different frequencies are being resonated or attenuated.

block diagram of speech production explain

air goes up vocal folds (vibrating or not) → makes noise (periodic or aperiodic) → goes through vocal tract (or constricted) and filtered to make speech

when breathing vocal folds are

open

when eating vocal folds are

closed

Cartilage is responsible for tightening and loosening of

vocal folds, changing tension.

what causes vocal folds to open and close really fast.

Subglottal pressure

Fundamental frequency affected by

-          length of vocal folds (men and women)

 

-          tension in vocal folds

 

-          mass of vocal folds

 

-          pressure in trachea (affected for people with respiratory issues)

Men have longer vocal folds so

lower fundamental frequency.

Mucus increases mass of vocal folds

lower fundamental frequency of source sound.

subglottal pressure relationship with F0

If we increase the subglottal pressure we increase the F0 of a sound.

Linear relationship.

linguistic significance of F0

we can change it as we speak. rises and falls during speech = intonation to speech

provides prosody or suprasegmentals (musicality) of speech providing meaning, emotion and emphasis

3 major contributors to prosody

fundamental frequency (voice pitch)

amplitude envelope (vocal effort)

duration and rhythm (timing)

F0 of speech is in what range

60 to 500Hz

quarter wavelength resonators

The vocal tract acts as a resonant tube or quarter wave resonator (tube open at one end and closed at the other) that amplifies certain frequencies produced by the vocal folds.

prosody (musicality) suprasegmental information is affected by what

•       Requires good perception and control of one’s own voice (laryngeal and breathing musculature)

•       Hearing impairment (particularly congenital) may affect this control leading to unnatural speech, having trouble with altering the F0 of voice, and so have trouble providing intonation and prosody in their speech

what frequencies are maximally resonated by quarter wavelength resonators

-          When the wavelength of the tone is 4 times the length of the tube, that frequency will be maximally resonated.

 

-          Quarter (of the) wavelength resonators provide acoustic conditions where compressions of the longitudinal wave line up and reinforce themselves.

F1 also named

first formant

the centre frequency of the first of a series of resonances.

F1 =

s/4L where L is length in metres and s is the speed of sound

EXAMPLE: If a tube is 30cm long, what is the first resonant frequency (i.e. first formant, F1)?

F1 = s/4L

F1 = s/4 x 0.30m

F1 = 334/1.2

F1 = 278.3 Hz

the centre frequency of a formant is the

formant frequency or (F_n).

first peak = formant 1

spectrograms

Show amplitude and frequency variations of speech against time:

y-axis shows frequency in Hz (linear scale).

x-axis shows time (ms).

darkness (or lightness) shows amplitude (dB).

spectrograms made using what type of filters

sets of band pass filters.

wide band or narrow band

Better accuracy for timing leads to poorer accuracy for frequency

Prosody requires good

biofeedback. The speaker must accurately perceive their own voice to monitor and control their laryngeal and respiratory musculature.

Individuals with hearing impairment, particularly those with congenital or early-onset hearing loss, lose this auditory feedback loop (produce reduced prosodic variability flatter more monotone f0 contour)

 

spectrograms frequency part that is important for speech

100Hz to 3kHz is the region where the most important speech information lies

brain hears for the relative difference between

f1 and f2

formants change position formant transition due to

movements of tongue and mouth during pronunciation of different words