The midterm/quiz will be March 7 in class.

Material covered: 
        Johnson, Chapt 1-3 (but there will be no questions about  autocorrelation or RMS measurement).
                         Chapt 5, 6 (6.1-6.5 only)

        Ladefoged,  Pp. 181-190.
        Port's Handouts:  Basic Acoustics,  Acoustics of Speech (Sections 1-5),
             Digital Signal Processing Overview   [Note the improved new (borrowed) graphics.] 
             Acoustics Problems, Audition for Linguists

BASIC ACOUSTICS

• sound, acoustic medium,
• acoustic waveform: compression, rarifaction
• periodic sound. Eg, sine wave with frequency, amplitude and phase (re another sine wave)
• frequency (Hertz = Hz = cycles/second); ms = millisecond; period (T)
• amplitude (usually pressure). Greater amplitude usually implies greater perceived loudness.
  
• complex periodic wave, fundamental frequency, component waves
• harmonics of a periodic wave (F0 = first harmonic, plus 2d, 3d etc). Equally spaced at multiples of F0.
  
• Fourier's Theorem, Fourier analysis for periodic waves only.  Says any periodic wave can be modelled as close as you please by summing sinusoidal frequencies that are multiples of the fundamental frequency - that is, using sinusoidal harmonics.
  
• power spectrum - display of energy at various frequencies. Represents a waveform in a way that is easier to interpret than the waveform itself.
  
• aperiodic sounds.  Eg, white noise, acoustic transients, impulse or pulse
• low-pass filter, highpass filter, pass-band vs. reject-band
• bandpass filter with center frequency and bandwidth

• Deriving the resonant freqs of a uniform tube:
• `standing wave' - identical waves passing in opposite directions (eg, from reflection of a single wave);
•   at the (velocity) node the particles always stand still, while at the (velocity) antinode they exhibit maximal motion
• Quarter wave resonator: One high impedence (hard) boundary plus one low impedence (soft) boundary. Assume uniform tube or bar.
• The lowest frequency full wave is 4 times length of tube or bar.
• distance = rate * time
• wavelength = speed.snd * wave.period, ... thus
• wavelength = speed.snd / frequency, ... thus
• freq = speed.snd / wavelength
• resonant.freq = M * (speed.snd / (4 * tube length)) where M  is odd (ie, M = 1,3,5,...).
• Why no even-numbered harmonics?  Because the even harmonics would have a node at the open end - impossible.
  
• Half-wave resonator: Tube or bar with both ends having high impedence (eg, closed or clamped) or both low impedence (open or free). Ie, tube closed at both ends or open at both ends.
• wavelength = speed.snd / frequency
• freq = speed.snd/wavelength
• resonant freq = N * (speed.snd / (2 * tube length)) where N  is any integer

• What is the acoustic theory of speech production?
  
• Understand the basic idea of:  Source X Transfer Function = Output Wave
• Explain how the X (multiply) operation works in the previous line.
  

• Schwa vowel is modelled by a quarter-wave resonator.
• F1 and F2 of the vowels [a] and [u] can be crudely modelled using the equations above.
• Modelling F1 for [i] also requires the `Helmholtz bottle' equation.
• Quantal vowels - where cavity affiliations switch and articulatory error has minimal effects
• pressure + velocity = constant.  Ave pressure and ave. velocity are complementary
• Perturbation theory:   Constriction at a pressure maximum (minimal motion) -> resonance rises (eg, F1 of [a], pharyngeal Cs)
  Constriction at velocity max (minimal pressure) -> lower frequency of resonance (eg, F1, F2 of [u], F3 of [r])
• Be able to produce a figure of the corner vowels on Freq X Time and  F1 X F2.

•  Sampling Theorem (Nyquist's Theorem): you need at least 2 samples for cycle or you get aliassing.
• Quantization noise: So LP filter input (before sampling) and again on output (after conversion to analog).
  
• filter types:  LoPass, HiPass, Bandpass, Notch,... any special shape
• windowing,   FFTs
• linear prective coding (LPC)
  

• Basic hearing anatomy. Properties of basical membrane, organ of Corti, etc
• inner and outer haircells.
  
• Auditory representations (vs. physical acoustic ones)
• Empirical ransformation of frequency scale into mels, barks or critical bands.
• lateral inhibition
• place code (200 Hz-20 kHz) and time code (20 Hz - 4 kHz
• learning and creating lead to occurrence of signs.