Ling2005 week3

1
Ling2005 week3

• Topics
• Review nature of sound (week 2)
• Complex waves
• Waveforms and their spectra
• Fundamental frequency
• Frequency analysis

2
Review Week 2 Preliminaries Acoustic Phonetics

• Sound waves and barometric pressure waves
  Differences?
• Compare with regard to wavelength, velocity,
  period, frequency.
• Define pressure.
• Force per unit area
• Why do we need a medium for sound?
• Sound transmission relies on kinetic energy of
  particles in a medium
• Descriptive terms for sound quality?
• Pleasant noisy, mellow harsh, .
• Three ways of classifying sounds/signals?
• Periodic/aperiodic, transient/continuous,
  simple/complex

3
Review continued

• Properties of waves
• Amplitude
• Displacement from baseline arbitrary unit but
  see decibels
• Period
• Time between wave fronts Measured in
  milliseconds
• Frequency
• Number of repeats per unit time. Measured in Hz
• Auditory and acoustic dimensions of sound
• Perceptual versus physical properties.
• Pitch frequency
• Loudness Pressure or Intensity
• Quality/Timbre distribution of spectral energy.

4
Perceptual and acoustic properties of English
consonants

• How may we map perceptual relations between
  speech sounds?
• Obtaining similarity judgments
• Mapping perceptual distances multi-dimensional
  scaling.
• What features might be responsible for clustering
  the sounds in perceptual space?
• The X-axis?
• Stimulus duration (time)?
• The Y-axis?
• A resonance/sonority vs noise/hiss dimension?
• Correlates with manner of articulation
• Are these general auditory properties of sounds
  or are they specific to sounds of speech
  (phonetic properties)?

5
Waveform and spectrum

• There are two complementary ways of looking at
  signals
• As a waveform in the time domain
• As a spectrum in the frequency domain
• Consider this sound
• What kind of sound is it?
• Can you guess its frequency?

6
A simple (sinusoidal) wave

• Its waveform

• Its spectrum

7
The spectrum

• A waveform shows amplitude fluctuations in a
  signal over time.
• A spectrum shows the distribution of frequency
  (energy) components present in a signal
• their relative strength (magnitude) and where
  they are located in the frequency domain.
• A spectrum is a snapshot of the energy of the
  signal at a particular point in time
  (instantaneous spectrum),
• Or the average energy present over a period of
  time (the long-term spectrum).
• In practical terms, a spectrum must always be
  computed over some minimal time period (Hence the
  instantaneous spectrum is an idealization.

8
Spectra of complex signals

• What is a complex signal?
• One with more than a single frequency component.
• This is a (minimally) complex signal

9
Notes on previous slide

• How many periods are highlighted in the waveform?
• How many frequency components are present in the
  spectrum?
• Two
• What are their respective frequencies?
• 170 Hz and (approximately) 240 Hz.
• On the lower part of the spectrum we see spectral
  time window over which the spectrum was
  calculated.
• How many periods of the signal are captured in
  the spectral window?
• Approximately four.
• The spectral window has a certain shape
  (envelope). The envelope indicates the weighting
  applied at each time point about the centre of
  the spectral window. How would you describe the
  shape of the window?
• Like a normal curve (Gaussian).

10
More waves and their spectra

• Recall the sounds we examined last week?
• The buzz

11
Notes on previous slide

• The buzz is made up of a series of rapid clicks
  (pops).
• The time between clicks defines its Period and
  (fundamental) Frequency.
• From the spectrum, you see that the buzz has many
  frequency components.
• Each of these components occurs at a regular
  frequency interval.
• This frequency interval corresponds to the
  frequency of repetition of the waveform. It is
  known as the fundamental frequency of the
  signal.

12
White noise

• A continuous, aperiodic signal.
• Has a continuous spectrum an infinite number of
  frequency components.
• Has no fundamental frequency.

13
A quasi-periodic signal

• This sounds like a sinusoid in noise.
• A composite signal with periodic and aperiodic
  components

14
Questions on previous slide

• Can you distinguish the periodic and the
  aperiodic components of this signal on the
  spectrum?
• Does this signal have a fundamental frequency?
• If so, what is that frequency?
• Ans 366 Hz.