Speech Analysis

http://svr-www.eng.cam.ac.uk/~ajr/SA95/node1.html


Next: Introduction Up: Speech Analysis Previous: Speech Analysis

Contents

 
• Introduction
  • What is Speech Analysis?
    • So what is an acoustic vector?
  • Why Speech Analysis?
  • The problems of speech analysis
  • Standard references for this course
• Background
  • Sampling theory
    • Sampling frequency
    • Sampling resolution
  • Linear filters
    • Finite Impulse Response filters
    • Infinite Impulse Response filters
  • The source filter model of speech
• Filter bank Analysis 
  • Spectrograms
  • Non-linear frequency scales
• Short-Term Fourier Analysis 
  • Properties of the DFT
    • Linearity
    • Time Shift
    • Frequency Shift
    • Convolution
    • Real valued input
  • Windowing
  • The short-term Fourier transform
  • Zero padding
  • Fast Fourier transforms
  • Practical application of the short-term Fourier transform
  • Overlap and add for linear filtering
    • Example: Spectral subtraction
  • Cepstral analysis
  • Homomorphic filtering
  • Mel scaled analysis
  • The Autocorrelation from the FFT
• Z transforms
• Linear Prediction analysis
  • Motivation from lossless tubes
  • Parameter estimation
  • The autocorrelation method
  • The covariance method
  • Pre-emphasis
  • The LP spectrum
  • Gain computation
  • The lattice filter implementation
  • The Itakura distance measure
  • The LP cepstrum
  • Log area ratios
  • The roots of the predictor polynomial
  • Line spectral pairs
• Perceptual Linear Prediction
  • Spectral Analysis
  • Critical-band spectral resolution
  • Equal loudness preemphasis
  • Intensity-loudness power law
  • Autoregressive modelling
  • Discussion
• Formant analysis
  • Motivation
  • Obtaining candidate values
    • Peak picking on the smoothed spectrum
    • Peak picking on the LP spectrum
    • Factoring for the LP roots
  • Fitting ``bumps''
  • Combining candidates
• Voicing analysis
  • Pitch synchronous analysis
  • Zero-crossing points
  • Peak in the autocorrelation function
  • Peak in the autocorrelation of the LP residual
  • The average magnitude difference function
  • Peak in the cepstrum
  • Combining candidates
• Degree of voicing
  • Voicing determination
  • Usage
• Speech coding
  • Waveform coders
    • Pulse Code Modulation (PCM)
    • Differential Pulse Code Modulation (DPCM)
    • Adaptive Differential Code Modulation (ADPCM)
    • Example: lossless waveform coding
  • Sub-band coders
  • Linear prediction vocoders
    • Parameter updating
    • Example: LPC10
  • Formant coders
  • Multi-pulse coders
  • Backwards adaptation
  • CELP coders
• Audio demonstration
• References
• About this document ...

---

Speech Vision Robotics group/Tony Robinson