Quiz #5 - LING 313

How many types of Basic waves

2 types

• transverse

• longitudial

Transverse Wave

particles move upward and downward in a perpendicular motion

Longitudinal Wave

particles move forward and backward

what is similar between transverse and longitudinal waves

the particles themselves do not travel but the wave does

can be represented the same way as the transverse wave

both follow the principles of simple sine waves

• something pushes its neighbour once they recover from it push it retracts (ossilations)

Explain transverse waves

Particles that we see move up and down

the particles are moving perpendicularly to the way the wave is moving

ex: a stadium people don’t move they just go up and down

Propogate

the movement of a wave through a medium that transfers energy, momentum, and information from one location to another without permanently displacing the medium's particles

Explain longitudinal waves

particle movement travels in the same direction as the wave

high-pressure zones correspond to the peaks, low pressure zones to the vally

Characteristics of simple waves

Frequency

Intensity

Phase

Frequency

how many cycles per second (how many full circles of movement)

measured in Hz

number of full cycles per second

Intensity

how strong are the oscillations (amplitude value gives us this)

measures in dB

Phase

when did the wave start

often measured in degrees (relative to another wave)

What is the formula for frequency when the period is in milliseconds?

Frequency = 1000 ÷ period (ms)

1 second = 1000 ms, so the number of cycles in one second = 1000 ÷ period in ms.

Frequency vs Pitch

frequency is an acoustic property

pitch is something that we perceive after interpreting those frequencies

the perception of frequencies are colored by the limitations of our hearing system

Amplitude

the distance betwwen the zero line and the peak or vally

• displacement from the zero line

What would amplitude look like in a perfect sine wave

always be equal displacement positively and negatively from the zero line

ni real life there is a bit of buffer (but that is negligable)

amplitude vs loudness

amplitude is an acoutic property

loudness is something we perceive after interpreting amplitude in our hearing system

Phase

the timepoint when the wave started

Noise cancellation

180 degrees after the wave which cancels the sounds out

A wave graph has DISTANCE on the x-axis. What does it show, and what is one cycle called?

A picture of the wave spread out in space; one cycle = wavelength

Distance on the x-axis means it's a snapshot of the wave spread out in space (many particles at one moment), so one cycle is the wavelength (λ).

A wave graph has TIME on the x-axis. What does it show, and what is one cycle called?

One particle moving over time; one cycle = period

Time on the x-axis means it's one particle moving over time, so one cycle is the period (T).

what are speech sounds made up of

complex waves

Sine waves

simple waves with regular oscilating patterns

explain this sine wave

100hz (oscillates 100 times per second)

x and y axis represent

x - time

y- amplitude (pressure varaition in the air)

100 vs 200 hz wave

Sine wave → complex wave (how is it done)

adding multiple sine waves results in a complex wave

What do period waves mean in speech sounds

vowels, sonorants, prevoicing

periodicity

repeating patterns

aperiodic waves in speech sounds

voiceless fircatives

combination of periodic and aperiodic waves

voiced fricatives

cues in waveform (5)

complex waves to power spectrum

make a list of the sine waves in a complex wave

• frquency

• amplitude

complex waves to spectra

make a graph with frequency on the x axis and amplitude on the y axis

how are specturm and waveform different in terms of the info they provide

specturm doesnt have time

not changes across time its just a snapshot

spectra of vowels interesting info from spectrum

Fundamental Frequency (F0)

Formants (F1, F2, etc)

Individual Harmonics - component frequencies

Spectral Tilt

Fundamental Frequency

1st harmonic

lowest frequency that your vocal folds produce

Formants

vowels show these

spectra to spectrogram

1. find the peaks in a spectrum and draw shades as verical lines

2. rotoate the spectrum 90 degrees anticlockwise

3. extract a clip from the right side of the figure (marked in red)

4. repeat with mulitple spectrum

5. 
   

what does high darkness mean on a spectrogram

higher amplitude

amplitude

how far the wave departs from zero

period

T

duration of one cycle in seconds

frequency

f

cycles per second in Hz

amplitude vs frequency vs phase

relationship between freqeuncy and periods

reciprocals

where does the sine wave come from

as you move on the wheel you can draw right angled triangles for your position

then extract the legnth of the hypotenuse and the “opposite” side

the sin of an angle is the ratio b/w opposite and hypotensue

as the angle changes the ratio (sine value) also changes

how does the sign wave become a curve

when the sine of an angle is plotted against that angle, the result is the classic “sine curve”

analog signals challenges

analong signals are continuous

• a value exists for every possile time poin t

• each value can be precise to infinitely many decimal points

this make them difficult to store, access, and analyze

• computers can only store a finite list of numbers

  • to take advantage of modern computing we must digitze them

analog - to- digital conversion (ADC)

two steps one for each axis of the graph

1. sampling

• measure the signal at regular moments in time (the time axis) keep those values and dump the rest

2. quantization

• round each measurement to one of a limited set of values (the amplitude axis)

3. after sampling we have a table of values, these “discrete” values are what is stored on the computer

sampling rate/frequency

we decide which values are kept and which are dumped (forever)

• the sampling rate says how many samples are kept per second

ex: if we keep 100 samples from a one-second interval, our sampling rate is 100hz

Nyquist theorem

to capture a compnent of freqeuncy f the sampling rate fs must be more than 2f

nyquist frequency

half the samplimng rate ie: fs/2

highest frequency a recording can represent and is a property of tthe recording settings

why more than 2 times

exactly 2 samples per cycle can land on the zero crossings, the wave looks like silence

with strickly more than 2 per cycle the samples cannot all miss

Aliasing

a distortion phenomenon that happens when a continuous signal is sampled too slowly, causing high-frequency components to fold into lower, false frequencies

frequencies above the nyuist freq are not lost they “fold back” and show up as false lower frequencies

anti-aliasing filter

before sampling, the recorder passes the signal through a low-pass filter that removes everything above about the nyquist frequency

this is the anti aliasing filer once aliasing has happened it cannot be undone

these day this is built into digital records and audio interfaces so we rarely have to handle it ourselves

why aliasing occurs during downsampling

quantization/sample size/ bit depth

the second issue in digitization is quantization

how precise should our y axis be

analog signals have infinitely many possible values and we cannot allow infinite numbers

quantization example

why do we have more levels

levels to bits

quantization noise

file size

source filter theory

speech = sound produced at the larynx x shaping by the vocal tract

the product of the signal/sound that is produced @ the larynx then shaped @ the vocal tract

who formalized the source-filter theory

gunnar fant (1919-2009) in acoutic theory of speech production (1960)

two parts of the source filter theory

1. the source

2. the filter

define the source

what is generating the signals

glottal vibration (periodic) ex vowel

noise generated at a constriction (aperiodic) ex voiceless fricative

define the filter

the supra-laryngeal cavities (the vocal tract)

payoff

source x filter = output

what are harmonics mathamatically

integer mulitples of F0/H1

filters

a filter passes some frequncies and attenusates others

types of filters

low pass, high pass, band pass

low pas

only frequencies lower than this threshold

highpass

only frequencies above x threshold

band-pass

only frequencies between these thresholds

cut off theory

where output drops (conventionally -3dB)

roll-off

how steeply it drops (dB/octave)

bandwith

width of the band pass

narrow bandwidth

the vocal tract as a filter

the vocal tract acts as tube with a series of band-pass filters, it boosts harmonics near the center frquencies and attenuates others

what are the spectral peaks

formats

super imposed