Unit 3 Flashcards
Resonant frequencies determined by
length and shape of tube
Acoustic resonators
principles of acoustic resonators (vocal tract) that allows us to produce and perceive difference vowels
Vocal tract length
tube shaped, extends from vocal folds to outer border of lips
running sound through tubes amplifies harmonics
based on shape, certain harmonics will be amplified more than others which will be suppressed
Tube makes voice louder
16/17 centimeters
Transfer function:
mathematical formula called a function describes which harmonics will be amplified and which will not
As tube changes shape/ size, transfer function will change as well
Different harmonics will be amplified (different vowel sound heard) with different shapes and sized
Resonance
fundamental frequency determines pitch of voice which is driven by VF
Resonant frequency= frequency of peaks based on size and shape of vocal tract
Natural (resonant) frequency
Based on elasticity and characteristics of the mass, the mass spring model is going to want to vibrate in free vibration at a particular frequency
When set into free vibration, mass will vibrate with maximum amplitude given certain force at certain frequency
Turning fork
Different resonant frequencies
Vibrates at different frequencies
What influences resonant frequency
Smaller things have higher resonant frequencies
Resonance and standing waves: Mechanical explanation
Standing wave patterns based on length of vibrating mass
Incident wave frequency= reflective wave frequency
Points where they meet in phase, results in standing wave
Standing wave goes up and doesn’t move
Constructive inference (amplification)
- VF sends pulse of pressure; wave travels up vocal tract and hits resistance of air at lips and reflects wave back down vocal tract
- Meet at different points along tube; point where waves meet and become more powerful are resonances of vocal tract
- Length of tube and frequency of driving force determine number of standing waves
- can have several at once
Example acoustic and mechanical resonators/ amplifiers
Mechanical resonators:
turning fork: sticking on a box will vibrate a particular frequency
Acoustic resonators: string: take high or low string and it’ll vibrate at a particular frequency based on mass, length, tension; vocal tract is inside guitar and tissue around vocal tract is body of guitar- depending on acoustic resonator we can create certain sounds
Tubes as acoustic resonators
Resonating mass = air column in a tube closed at one end and open at the other
Resonant (natural) frequencies of the tube are the standing waves determined by the length and shape of the tube
Resonant Frequencies: Numbered from lowest to highest (e.g. R1, R2, R3, etc.)
Multiple resonant frequencies in tubes
R1: lowest resonant frequency
Two tubes will have resonant frequencies
Calculating Resonant Frequencies of Straight Tubes Open at One End
Lowest Resonant Frequency (R1):
Original Formula: Frequency = velocity of sound / wavelength
Resonance Formula: Frequency = velocity of sound / 4X length of the tube
Example: Tube Length = 6 inches (.5 feet)
R1 = 1130 ft/sec / (4 X .5’) = 1130/2 = 565 Hz
Additional Resonances:
R1 = 1 x 565 = 565 Hz
R2 = 3 x 565 = 1695 Hz
R3 = 5 x 565 = 2825 Hz
The wavelength will always be 4x length of tube
Odd number multiples!
Loudness will go up and down, we have a source with all the harmonics , the harmonics that fall below the peaks will be the loudest harmonics that fall between them is the softest
The main part of vocal tract that resonates for vowels is the mouth
The vocal tract as an acoustic filter
When a mass is set through a filter, it rejects some mass but allows others to flow through it, which is what the vocal tract does with harmonics- some harmonics go through, some not
3 types of filter:
Low pass filter (lower than cutoff frequency)
High pass filter
Bandpass filter
Low-pass filter designed that such that when material is set through it, only the material below the threshold of that filter will go through it, anything above it will not go through it
Want to pass the only frequencies that are lower than the peak of the voice
High pass filter: passing only what the cut off frequency is (high pass filter gives you pitch)
Bandpass filter: two cutoff frequencies; low frequency cut off and high frequency cut off. Only accepts the middle portion.
Filter characteristics
Center frequency: center of the filter, width of the band between the low cut off and high cut off is called the band width
To measure band width: go to the peak find out what pressure or decibel level and go down 3 decibels
Bass of car: lowering everything down and turning up the low frequencies
Treble: raising the high frequencies and lowering the low frequencies
Center frequency (peak) Cutoff frequencies (low and high) Bandwidth (what frequencies are included
Bandwidth factors and application of speech
Shape of the tube influences the band width: straight tube will have a fairly narrow band width
If u bend the tube, it will bandwidth will get wider and amplitude will go down a little bit
More bending= wider band width
Damping characteristics of tube: more absorbent you have, the wider the band width, the lower the amplitude
Waves that overlap each other will start to sound muffled
Higher, amplification gets more narrow
Formants
Formant: a name that is given to the resonance of the vocal tract
Each of these peaks is like a harmonic
In the vocal tract, there are 7 formants
1,2,3 are in the oral cavity
Only the formants associated with the mouth cavity are involved in speech production
The bandwidth gets wider with larger frequencies
Formant 4,5,6,7 down in larynx
Ignore formant 4,5,6,7 irrelevant to speech production
Formant 1 and 2 are most important
Formant 3 occurs in nasal cavity but does not playa significant role in vowel productions
The sound spectrogram
Transmit speech signal and transmit it and only people know how to transmit it, can read it
Visualize the speech signal at 3 dimensions
Time is on horizontal axis
Vertical axis is frequency
Amplitude: shading, darkness
Pattern play back
Developed pattern playback, first attempt at developing speech synthesis
Sound they drew came out of it
Development of sound sonogram
2.5 seconds of speech, and speeds up really fast
Goes through band path filter and takes the lower frequencies and tells how powerful they are and keeps doing it until it goes to the top of the frequencies and creates an electrical signal and more power, more electricity generated
More power= more electricity
Goes to a metal wire and take a speech of specialized paper and wrap it around the drum and burns the paper and you get what looks like the sound spectrogram
Not they have digital sonographs on computer
Wide band (300 Hz) vs. Narrow band (45 Hz)
Use both a Narrow band and wide band filter for analysis
It can show you each individual harmonic
Each bars indicate harmonics
And you can see fundamental frequency
Narrow band good for: glottal wave form
Wide band: 300 hz, chunks of frequenices and anlyzing them together so you don’t see individual harmonics, come out as broad bands / bars rather than individual harmonic lines so less information
The spectrograph takes entire voice spectrograph and will pick out first harmonic and do it again but move up
Advantages of narrow band and wide band
With narrow band: we can see intonation (up and down of pitch)
More interested with formants then may switch over the wide bands where harmonics are broad, and you can see formant structure
Narrow Band (45Hz) A Sound source Harmonic structure Fo Intonation
Wide Band (300 Hz) B
Formants
Relative timing
Glottal pulses
Formants and vowels
ah, (F2: coems closer to F:1
Front vowels, f 1 and f 2 far apart
Back vowels: both at low end and close together
Aperiodic sounds: voiced and voiceless
voiced
voiced and voiceless have different bars
fricatives will be long
stops will be very short
affricates will be about in the middle
more intensity/ power in the higher frequencies
high concentrated energy (s, z)
tell they are all voiced because they all have something on the bottom
you see formants with voiced (dark areas on spectrum)
voiceless
no formants
getting the client to produce a different acoustic pattern
Bands of Energy diffuse energy rather than formants
Formants sometimes visible in voiced fricatives
Voice bar = voiced
Duration influences sound category (plosive, affricate, fricative)
Oscilloscope
scope: monitor, shows on screen
monitor that shows amplitude of a signal by time which is what a waveform would look like
so as amplitude changes, it goes up and down on the screen (2D)
The vocal tract as an acoustic resonator
Source= voice filter= vocal tract output= vowel
Influence of tube shape on formant pattern
The more open and more unrestricted it is, greater power in a signal Open vowels (neutral vowels) have more power and emphasis (ahhhh) The degree of openness is changing, as we change the openness/ degree of restrictedness, we will find the formants will change
Factors that determine formant patterns for vowels
Oral cavity shape
Location of tongue constriction (front- back)
Degree of constriction (tongue height)
Size of mouth opening and ip protrusion/ rounding
Certain characteristics important in determining what vowel we will produce
Vocal tube length: how long it is from beginning of the back part of the oral cavity to the very front edge of lips
The lips have significant impact on vowel sounds
If the tube gets shorter: formants are higher
If the tube is longer, the formants will be shorter
We can intrinsically change the length of the tube: rounding the lips (eeee) is shorter
The formant frequencies will shift when we change roundness of lips
The oral cavity can be divided into front cavity and back cavity, divided by the high point of the tongue. The high point of tongue is the point of constriction
The front- back location of where it occurs- in front of the high point of tongue and behind high point of tongue
Formants: what to look for
A: relatively open tube
i: constriction more towards front
U: constriction more towards back
Formant 1 and 2 most important to produce the vowel you hear
Height: amplitude of the formants
We have where they are on the frequency scale, and how high they are (powerful)
Higher formant frequencies will have wider band widths in general
You can change amplitude and bandwidths but still hear the same vowel
It is the location that makes the difference
The amplitude is not necessarily related to bandwidth
Lower frequencies have more amplitude than wider frequencies
Influence of vocal tract length on formant frequencies
We are going to be different in terms of where the resonants occur
As you change length of tube, where the resonants are change because the tubes are a different length
The ear hears the relationship between formant 1, formant 2, and formant 3
The ear rarely relies absolute values of things
Under the people, most likely a neutral vowel because evenly spaced
Formant 1: influence by the back cavity and tongue height
Front-forward tongue
As tongue goes from neutral to more forward, formant 1 goes down (lower frequency)
As tongue moves backward, you can see formant 1 begins to rise
As tongue goes higher, formant 1 also goes down
As tongue goes lower, formant 1 goes up
Maximum forward and maximum height= lowest formant 1
As tongue goes maximum forward and height you get the best sound
When we round our lips, the effect of the mouth opening comes in and drops formant 1 back down again
Formant 2: influenced by the front cavity and front- back tongue position
Formant 1 and formant 2 are inverse
Formant 2 associated with front cavity and formant 1 associated with back cavity
As tongue goes up and forward, formant 2 goes up
As tongue goes down and back as f1 comes up, f 2 will go down and back
Formant 3: influenced by front- back location of tongue and lip protrusion
formant 3 influenced by front back location
Formant bandwidth and its impact on vowel perception
Ratio of formant 1 and formant 2 and ratio between formant 2 and formant 3 had impact on vowel identification
Changing the bandwidth or amplitude did not have significant impact on vowel identification
The bandwidth did change the tonality
The vowels became less distinct when the bandwidths get wider
The distinctiveness of those peaks decreases
Higher frequency formants have wider bandwidths
If they get too wide, speech becomes non clear but if they get too narrow than it sounds unnatural
Bandpass filter with center frequency of 500 Hz and bandwidth of 300, you half it and subtract it from the bandpass
Vowel duration: intrinsic features and consonant environment
The duration of vowels varies significantly
You can create subtle difference sin pronunciations of vowels
Some vowels are longer than other vowels
Vowel height: /i/ /I/ /ou/ tend to be shorter
the /i/ is tenser
The distance the tongue has to travel between letters can cause vowels to become faster and not go as far and the tongue may not go all the way to the e /i/ position and then move rapidly if it has further to go
When u have a voiceless sound before the vowel will tend to be shorter
Degree of constriction affects vowels
Voicing lengths the vowels more than non voicing
The greater the degree of constriction (plosive) the vowel will be shorter
As you decrease degree of constriction (fricatives), the vowel will be longer
The vowels will be the longest for the voiced sibilant z
Vowel duration and syllable stress
Duration of vowel in a stressed syllable or stressed word will be longer than the same vowel in a non stressed syllable
The stressed syllable has a longer, higher pitch, frequency, and amplitude
Vowel duration and speaking rate
You are much more restricted in duration of consonants as you go from one category of consonant sounds to another
As you increase speech duration of vowels and consonants can shorten
Vowel fundamental frequency
Fundamental frequency: pitch in the vowel
Intrinsic: tongue height impacts the vowel
High front and back vowels: tensing tongue and stretching it
The vocal folds will tense if tongue tenses and frequency will go up (vibrate faster)
Stress syllable: longer duration, higher fundamental frequency
Right hemisphere encodes emotional states from language
The more open, the larger the megaphone, the more powerful the vowel
Formant transitions as sound connectors and indicators of coarticulation
Formants don’t change in frequency across time: this tells you that the vocal tract is keeping its figuration across time
The physiological vocal tract is in a steady state- it is not changing shape and the articulators are not moving
Bending formant depicts the articulators are moving from one sound to another- transitioning from one acoustic position to another
The fluidity and blending of motion from one articulatory to another that affects the dynamic of speech production
When we talk its always blended with sounds either before or after (coarticulation) the structures will move with concert with one another anticipating what is going to come back and remembering what just happened
Bending: acoustic transitions of that change
Acoustic information in the bending formants that tells your brain what came before it and what is going to come after it
Acoustic formant transitions
Each of the formants will act independently and might show different characteristics of transitions
The point at which the formant frequency begins to change regardless of direction is called the Onglide point
Formant 2 begins to change shape as its moving towards the formant configuration you would want for the next vowel
The point where formant 2 stops moving is called the Offglide
The steady state is the point between B and C (where it doesn’t change shape)
If formant 2 started higher for some reason, it would go down and then steady state
Positive change is going up and negative is going down
We can measure the duration of the transition: how long it took to get from point A to point B, how fast we are moving our articulators
Frequency change: amount of frequency change that occurs, ear is providing us information about the configuration of the vocal tract
Rate of change: number of frequency changes over a given time, integrates everything, the rate of change provides the most information about the sounds that come before and after
Each formant has its own transition characteristics
dipthongs
If we take two vowels and connect them with a transition usually of about 50 msec, it changes. It no longer becomes two vowels or two syllables it becomes one vowel with two nuclei
The neutral vowel then goes upward to get to the high neutral vowel for the I
If that transition becomes too long (should be 50msec) then it will sound like two different vowels with voicing in between
If the transition is too short, sounds like one vowel
dipthongs and transitions
The brain doesn’t necessarily care for the absolute vowels of acoustic features when it comes for speech
Absolute values don’t matter because our vocal tracts are different
The brain looks for relative values: how much change not the absolute vowels and looks for relative changes
Rate of transition or frequency change over time