An Introduction to Vocal Acoustics and Spectrographic Analysis

1
An Introduction toVocal AcousticsandSpectrograp
hic Analysis

• Dr. Philip Sargent
• Vocal/Choral Division
• Shenandoah University

2
What is Sound

• What we can hear
• Vibrations (pressure variations) that can produce
  the sensations of hearing
• 20-20,000 Hz (Hertz or cycles per second)
• Travels through a medium
• Displacement and velocity of medium molecules
  correspond to the pressure variations

3
Nature of Sound

• Vibration
• Travels as waves of compression/rarefaction
• Actual medium doesnt move far it just
  oscillates in place

4
Nature of SoundOscillating systems demos

• Pendulum
• What changes when string is shortened?
• What changes when weight is increased?
• Spring and mass
• What changes when spring is stiffened?
• What changes when weight is increased?

5
Nature of Sound

• Vibration
• Travels as waves of compression/rarefaction
• Actual medium doesnt move far it just
  oscillates in place
• Travels at a constant speed (1130 fps)
• Speed dependant on the conducting medium
• Faster through helium, steel, water, than air
• 335 m/s in normal air 350 m/s in the throat

6
Nature of Sound

• Vibration
• Travels as waves of compression/rarefaction
• Actual medium doesnt move far it just
  oscillates in place
• Travels at a constant speed (1130 fps)
• Speed dependant on the conducting medium
• May form a repetitive pattern (or not)
• Travels in all possible directions

7
Basic properties of sound

• Duration
• Amplitude
• Quality
• Direction or apparent location/source
• (psycho-acoustic property)

8
Basic properties of sound

• Duration measured by
• Tempo and rhythm
• Actual length (minutes, seconds. milliseconds)
• We are most interested in events measured in
• Seconds
• Milliseconds (ms)

9
Basic properties of sound

• Duration
• Amplitude - measured by
• Dynamics
• Fixed power units (e.g. Watts/meter2 )
• Decibels (logarithmic similar to the way we
  perceive loudness and pitch)
• Phons measures perception of intensity
  (loudness)
• Not used much in scientific studies too
  subjective

10
Basic properties of sound

• Duration
• Amplitude
• Quality
• Due to multiple frequencies being perceived as a
  single quality (Casio demo)
• Displayed as frequency spectrum
• Instrument identification, timbre
• Registration
• Vowel recognition

11
Basic properties of sound

• Duration
• Amplitude
• Quality (and possibly pitch)
• Direction or apparent location/source
• Result of simultaneous perception of slightly
  different sounds (Stereo, 5.1)

12
Basic properties of sound

• Duration
• Amplitude
• Quality
• Due to multiple frequencies being perceived as a
  single quality (non-harmonic synthesis demo)
• Can be displayed as frequency spectrum
• Voice type and gender identification timbre
• Registration
• Vowel recognition

13
Pitched and Non-pitched Sounds

• Non-pitched
• Impulse (t,k, clap, pop)
• Random (s,f, hiss, white and pink noise)
• Pitched
• Repeating pressure variation pattern (waveform)
• Perceivable pitch (frequency)
• Period and wavelength (Boxcars at the Xing)

14
Pitched Sounds

• Simplest Sine wave

15
Pitched Sounds

• Simple Sine wave
• Pendulum
• single spring and mass
• Complex
• Vowels, sustained instrument tones
• Non-harmonic struck instruments
• Harmonic Voice, strings, winds, piano
• Remember harmonics?

16
Pitched SoundsThe Harmonic Series

• Fundamental (F0 or F0)
• called F sub zero or F zero
• Lowest frequency partial
• Same as first harmonic (H1)
• Perceived as the sounds pitch (even if its not
  there!)
• Overtones
• Integral multiples of fundamentals frequency
• Produce regular pattern of musical intervals
• Not all present in every sound
• Progressively weaker than fundamental

17
Pitched SoundsThe Harmonic Series
18
Building a complex wavefrom harmonics(sine
wave components)
Pitched Sounds

• Backus 1969

19
Adding harmonics to build

• a
• Sawtooth
• or
• a
• Square
• wave
• Backus 1969

20
Displaying Complex SoundsVoceVista
21
Pause to studyVoceVistaProWindows and Interface

• Waveform
• Spectrogram
• Power Spectrum
• EGG - later

22
VoceVista and Fourier

• Jean Baptiste Joseph Fourier (1768 1830)
• All repeating wave patterns can be analyzed /
  synthesized as a sum of sine waves
• FFT displays the energy distribution in a complex
  wave by frequency
• In mathematics, the discrete Fourier transform
  (DFT) converts a finite list of equally-spaced
  samples of a function into the list of
  coefficients of a finite combination of complex
  sinusoids, ordered by their frequencies, that has
  those same sample values. It can be said to
  convert the sampled function from its original
  domain (often time or position along a line) to
  the frequency domain.

23
VoceVista Analysis Settings

• Narrowband vs. Wideband
• LTAS
• Control-Drag for Hz and vibrato rate
• Time and frequency range settings
• Reference lines
• Ctrl-F1, F2, F3 for over/under overlay
• F1-8 window choices
• Grayed out controls and Reset

24
Source/Filter Model of the Voice

• Source
• Lungs, and breathing muscles
• Larynx (Vocal folds)
• Filter
• Vocal tract
• Laryngeal spaces
• Nasal passages
• Trachea and bronchial tubes?

25
Source Breath Management

• Subglottic pressure (Psg)
• Cm H2O, KPa, other units
• Volume of air (Transglottal airflow)
• litres/second
• Must balance effort against the natural
  elasticity of the breathing apparatus

26
Source Breath Management

• elasticity of the breathing apparatus
• Sundberg 1987

27
Source Breath Management

• Subglottic pressures
• Sundberg 1987

28
Psg vs. dB for various phonations

• Pressure dB
• 14cm 70dB
• 9cm 76dB
• 8cm 78dB
• 5cm 68dB

Sundberg 1987
29
Source Vocal Folds

• Series of glottal puffs
• Psg overcomes glottal resistance
• Transglottal airflow begins
• Bernoulli effect and restorative tension in
  folds close glottis
• Pattern repeats
• Vocal folds may thin or thicken vertically
  depending on CT/TA balance

30
Sundberg SSV 1987
Sundberg 1987
31
Source Vocal Folds
Sundberg 1987
32
Source Vocal Folds
Sundberg 1987
33
Source Vocal FoldsExamining the Trans-glottal
flow

• EGG
• Measures conductivity between folds
• Correlates with open/closed phase
• Inverse filtering
• Removes effect of Vocal Tract resonances
• Result is the glottal waveform as flow

34
Pause to studyVoceVistaProEGG Window and
Interface

• EGG
• Polarity and order of input
• Scrolling to set EGG/microphone delay
• Setting Criterion Level
• CQ OQ 1

35
Waveview Inverse Filter
36
Source Vocal Folds
Source Vocal Folds
Sundberg 1987
37
MFDR and Loudness
Sundberg 1987
38
Source Vocal Folds (continued)

• Glottal waveforms differ with changes in
  registration
• Falsetto longer open phase nearly sinusoidal
• Soft head voice long open phase
• Chest/operatic long closed phase
• Belt longer closed phase most high harmonic
  content

39
Source to Filter

• Series of glottal puffs
• Vocal fold closure generates oscillations
  (standing waves) in the tube (vocal tract)
• Faster closure less spectral slope (MFDR)
• Total/peak trans-glottal flow affects relative
  F0 strength

40
Source to Filter

• Open and closed phase resonate differently
• Closed phase
• Standing wave is moderately dampened
• Higher CQ values important to UE (D. Miller)
• Allow stronger source H3 ( H4 if CQ gt.80)
• Open phase
• Rapid loss of wave energy
• heavily dampens the standing wave
• Subglottal cavities become part of the filter

41
Source/Filter InteractionString instruments

• Pitch determined by
• Length
• Tension
• Thickness (mass/length)
• Pitch independent of resonators characteristics
  
• Loosely coupled
• All overtones present in diminishing strength

42
Source/Filter Interaction Wind instruments

• Sound produced by
• Edge tone (flute, recorder)
• Vibrating reed (clarinet, oboe)
• Lips buzzing in a mouthpiece (brass)
• Pitch dependant on resonator (largely tube
  length) Tightly coupled, esp. in Woodwinds
• Output may contain all or just odd partials
  (based on resonator shape)

43
Source/Filter Interaction Vocal Inertance

• Increasing supraglottal resistance to flow may
  aid in glottal closure
• Non-linear system Titze
• Tightly coupled - Vennard

44
Filter Vocal Tract

• Voice
• Very irregular tube / coupled cavities Bottle
  Demo
• Cavity (Helmholtz) resonance
• Pitch factors
• Cavity size
• Neck length
• Neck diameter
• Shape can affect narrowness of response
• Resonator wall condition affects
  efficiency/attenuation
• Coupling of resonators

45
Formants

• Fixed pitch areas of resonance
• Affect amplitude of overtones
• Dependant on frequency, NOT the number of the
  harmonic
• Like the tone controls on a stereo or equalizer

46
The /a/ Formant
47
F1/F2 Vowel Formant PlotVennard 1967
48
Formants, Continued

• Singers Formant - Clustering of F3, F4, F5
  around 2800 Hz
• Vowels - F1, F2, and perhaps F3
• Formant cavities in front and behind tongue hump
  S.P.P.1 Demo
• IPA Formant Frequencies
• (PSS, spoken) (PSS, sung G3)
• /i/ 350 1800 400 1500
• /e/ 350 2000 400 1600
• /e/ 525 1700 550 1300
• /æ/ 600 1600 650 1400
• /a/ 650 1100 650 1100
• /o/ 350 650 425 750
• /U/ 400 950 500 700-900
• /u/ 300 700 450 700-900

49
Loudness Curves

• Sundberg
• 1973

50
Simplified Loudness Curves

• Backus
• 1969

51
Resonant modes of a tubewith one closed end
52
Formants based on summing the first four resonant
modes of a tube closed at one end(inaccurately
drawn)

• Kent and Read 1992

53
Resonant Response of a Uniform Tube(F1-F5
425, 1275, 2125, 2975, 3825 Hz)to a Sine Wave
Sweep

• Sundberg 1973

54
Pressure Nodes and Antinodes

• of the
• first two
• resonant modes
• of the
• vocal tract
• Kent and Read 1992

55
Constriction

• At A lowers F1 F2
• At B raises F2
• At C lowers F2
• At D would raise F1 F2
• But would involve constricting the larynx
• Kent and Read 1992

56
Display and analysis toolsSpectrographic
analysis (Gram, SpeechStation 2, Dr. Speech,
KayPentax)VoceVista

• Waveform display
• Shows frequency components and their amplitudes
• Real time or frozen slice of time (averaged
  over duration of slice)
• Spectrograph
• Shows persistence or change of waveform (timbre)
  over time
• Horizontal position (X axis) time
• Vertical position (Y axis) pitch of
  partial/overtone
• Display intensity (grayscale, color) intensity
  of sound

57
Pedagogical Uses of Spectrograms

• Vibrato
• Rate
• Width
• Evenness
• Presence/Absence (patterns?)
• Late onset
• Absent on alternate notes
• Effects of technical changes immediately observed

58
Pedagogical Uses of Spectrograms

• Effects of technique changes immediately observed
• Perceptions are current, not recalled
• Feedback comes while singer is singing and
• Experiencing the sensations of
• vibration
• physiological activity
• Sound generated
• .
• No verbal intervention necessary
• Expert students may use for feedback in teachers
  absence
• Models for individuals must be established
• Must not become a crutch

59
Pedagogical Uses of Spectrograms

• Vibrato
• Rate
• Width
• Evenness
• Presence/Absence (patterns?)
• Late onset
• Absent on alternate notes

60
Pedagogical Uses of Spectrograms

• Vowel articulation
• Clear differentiation between vowels
• Dynamic and tonal balance
• Consistency from attack to release
• Early diphthong closure
• Accuracy of phoneme (markers helpful)

61
Pedagogical Uses of Spectrograms

• Consonant articulation
• Length/duration
• Shaping of sibilant noise (s, ? , etc)
• Affect on adjacent vowels (Co-articulation)
• Strength of pitched sounds (balance with vowels)

62
Pedagogical Uses of Spectrograms

• Onsets (attacks) and releases
• Glottal
• Aspirate
• Bloomed (Nair)
• Gloriously simultaneous and perfect
• Nairs Articulators Set, Abdominally Initiated
  (ASAI)

63
Pedagogical Uses of Spectrograms

• Singers Formant
• Presence or absence
• Frequency
• Center
• Width

64
Sources

• Backus, John The Acoustical Foundations of Music
  (Norton 1969)
• Everest, F. Alton Master Handbook of Acoustics
  (McGraw Hill 2001)
• Kent and Read The Acoustic Analysis of Speech
  (Singular 1992)
• Sundberg, Johan The Science of Musical Sounds
  (Academic Press 1991)
• Vennard Singing the Mechanism and the
  Technique (Carl Fischer 1967)
• Also recommended
• Appleman, Ralph The Science of Vocal Pedagogy
• Benade, Arthur Fundamentals of Musical Acoustics
• Sundberg, Johan The Science of the Singing Voice
• Sundberg, Johan Everything else that hes
  written
• Titze, Ingo Everything hes published

65
Types of sounds

• Impulse
• Random
• Repetitive waveform
• Mixed sounds

66
Types of sounds

• Impulse
• Plosive consonants ( /t/ /k/)
• Instrumental attacks (organ chiff, harpsichord
  pluck)
• Bangs, slaps, snaps

67
Types of sounds

• Impulse
• Random
• Contains all pitches/frequencies
• (like an orchestra before tuning, but more so)
• Continuant consonants ( s, f , ? )
• White and pink noise
• Scrapes, scratches, tape hiss, wind noise

68
Types of sounds

• Impulse
• Random
• Repetitive waveform pitched
• Simple sine wave
• Complex harmonic most non-percussive musical
  tones (These are of most interest to us)
• Complex non-harmonic Vibes, marimba, chimes

69
Types of sounds

• Impulse
• Random
• Repetitive waveform pitched
• Mixed sounds
• Speech
• Gong