Physics of Music Lecture 4: Analysis and Synthesis of Tones

1
Physics of MusicLecture 4 Analysis and
Synthesis of Tones

• Berg Stork, Chapter 4
• Fourier Analysis
• Pure Tones
• Noise

2
Frequency and Intensity scales

• Frequency
• Each Octave is a factor of two change in
  Frequency (Pythagorous)
• Musical Octave sounds in unison
• Intensity
• 3 dB (deci-Bel) is a factor of 2 change in
  Intensity 3.0110.log(2)

3
Joseph Fourier (1768-1830)

• (1807) Any periodic wave of frequency f can be
  constructed as a superposition of standard
  (particularly sine cosine) waves of frequency
  f, 2f, 3f, 4f,
• Any non-periodic wave can be constructed as a
  superposition of standard waves of all
  frequencies.

4
Digital Sampling

• CD sampling is 44100 Samples/sec
• Highest possible frequency is 22050 Hz
• CoolEdit2000 Fast Fourier Transform (FFT)
• e.g. 4096 samples, Frequency resolution is 44100
  Hz / 4096 10.8 Hz

5
Fourier Synthesis Mathematical functions

• Square wave (frequency f) BS Fig 4-5
• Frequencies f, 3f, 5f,
• Amplitudes 1, 1/3, 1/5,
• Sawtooth (ramp) BS Fig 4-6
• Frequencies f, 2f, 3f, 4f,
• Amplitudes 1, ½, 1/3, ¼, 1/5,
• Repetitive Pulse BS Fig. 4-7
• Frequencies f, 2f, 3f, 4f, 5f,
• Amplitudes 1, 1, 1, 1, 1,

6
Triangle Wave
7
Square Wave (Synthesized)
8
Soprano Recorder, C above Middle C
9
PanPipes Noisy Triangle Wave
10
Chalumeau

• Fundamental at 155.7 Hz D
• (I was trying to build D)
• Expect odd harmonics at
• 3x155.7 Hz 467 Hz (A)
• 5x155.7 Hz 778.5 Hz (G)
• 7x155.7 Hz 1090 Hz
• 9x155.7 Hz 1401 Hz
• Observed frequencies
• 467 Hz, 768 Hz, 1041 Hz, 1284 Hz

11
Non Periodic Sound Sources

• Real music (vs. synthesized sound)
• Vibrato, tremolo, attack, decay, etc.
• Inharmonicities
• Noise
• Air Flow (flute)
• Instrument noises (keys, valves)
• Articulation transients
• Sung Consonants
• Percussion

12
White Noise Pink Noise(see Noise.wav)

• White Noise Equal (random) intensity at all
  frequencies
• Pink Noise Equal acoustic power in each octave.
  Intensity diminishes 3dB with each octave.
• Integrated acoustic energy intensity times
  bandwidth. Each octave is twice as large in
  frequency range as octave below.

13
Sample of White Pink Noise
14
Frequency Spectra White and Pink (1/f) Noise
White Noise
Pink Noise
15
White Noise, Normal Distribution

• White noise has a well defined mean and standard
  deviation.
• S.d. square root of variance
• Variance mean of the square of the deviations
  from the mean
• Precision of measurement of mean improves with 1
  over square root of number of samples.
• Public opinion polls have a precision of 3 with
  just 1000 people questioned
• 0.03 1/square root (1000)

16
Pink Noise, 1/f-Noise

• Pink Noise does not have a well defined mean or
  variance.
• If weather follows a Pink Noise distribution,
  then there is no such thing as climate.
• 100-yr flood is a flood expected once in 100
  years, and 10 x worse than the average flood in
  any 10 year period.

17
Frequency Spectrum of Symphonic Music (Pink
Noise?)

• A. Dvorak, Concerto for Cello Orchestra
• 1st 210

18
Timbre / Tone Quality

• Attack / Decay Transients
• Spectrum of driving force
• Reed, Lips, Bowing, Hammer, Plucking
• Harmonic response of each instrument
• Clarinet has some even harmonics (slightly
  conical bore)
• Sounding box/board of String instruments
• Formants (especially human voice)
• Inharmonicities
• Finite string size stretches Piano octaves.
• Wind harmonics not necessarily pure integers

19
Resonancesin open and closed tubes
20
Musical Instrument