Acoustics of Spoken Language

Acoustics of Spoken Language Dafydd Gibbon
First draft
(Printed: May 22, 1996 )

• Contents
• 1 Overview
• 2 Speech systems
  • Applications systems
  • The speech system cascade
  • The articulatory subsystem
  • The channel subsystem
  • The auditory subsystem
  • Extending the speech system cascade
  • Systems and Filters
  • Source systems
  • Modulator systems
  • Scaling: amplification and attenuation
  • Filter systems
    • Simple filters
    • Complex filters
  • Transmission media
  • Correlation: cross-correlation and autocorrelation
  • Convolution
  • Transformation systems
  • Fourier Transformation
  • Inverse Fourier Transformation
  • Fast Fourier Transform
  • Frequency resolution
  • Other transformations
  • Sampling systems
  • Windowing systems
• 3 Simple and complex signals
  • Signals: Terms and basic concepts
  • Fundamental signal types
  • Properties of sinusoid signals
  • Properties of complex periodic one-dimensional signals
  • Signal idealisation in phonetic models
• 4 Units of measurement
  • The unit signals
  • Basic mathematics
  • Amplitude
    • Magnitude
    • DC offset
    • Peak-to-peak amplitude
    • Average amplitude
    • Root-mean-square amplitude
    • Intensity
    • Energy, Power
    • Phon, Decibel
  • Period and Frequency
    • Period
    • Hertz, Cycles per second
    • Octaves, semitones
    • Mel
  • Phase
    • Phase and phase shift
    • The unit circle, its radius and its circumference
    • Degree
    • Radian, and
    • Angles and right-angled triangles
    • Signals and complex numbers
  • Bits and Bytes
• 5 Signals and systems in the time domain
  • The addition operation: spectrum synthesis
  • Multiplication: amplitude modulation
  • Clipping operations
  • The correlation transformation: spectral analysis
  • The autocorrelation relation: fundamental frequency extraction
  • Convolution: filtering
  • PSOLA: frequency and phase modulation
• 6 Signals and symbols in the frequency domain
  • Spectral analysis
  • The Fourier transformation
• 7 Vowels: theory and practice of formant analysis
• 8 Consonants: temporal and spectral structure
• 9 Intonation: fundamental frequency and fundamental frequency extraction
• 10 Speech recognition: architectures and methods
  • Types of ASR system
  • Segmentation and classification
  • Basic components
  • Knowledge constraints and stochastic constraints
  • The HMM approach
  • The ANN approach
  • Speaker adaptation
  • Hybrid approaches: adding linguistic constraints
    • Phonemic constraints
    • Autosegmental constraints
    • Levels of time structure
    • Event logics
• 11 Speech synthesis: architectures and methods
  • Parametric synthesis
  • Diphone synthesis
  • PSOLA
• References
• About this document ...