Overview

A while back, I taught a PhD level course on speech signal processing. It was complemented by content by Fabio Valente (in 2009 and 2011) and Milos Cernak (in 2013).

Slides

The course covered a fourteen week semester. Below are slides from the lectures that Milos and I gave. The missing weeks are laboratory excercises.

  • Week 1
    • Speech production
    • Vowels
    • Consonants
  • Week 2
    • Perception
    • Sampling
    • Framing
    • Overlap add
  • Week 3
    • z-transform
    • Homomorphic processing
    • Mel frequency ceptral coefficients
  • Week 5
    • Linear prediction
    • Perceptual linear prediction
  • Week 6
    • Bilinear transform
    • Root cepstrum
    • Vocal tract length normalisation
  • Week 8 (by Milos)
    • Transforms
    • Excitation
    • Inverse filtering
  • Week 9 (by Milos)
    • Speech coders
    • Excitation coding
    • ASR / TTS paradigm
  • Week 10 (by Milos)
    • Speech synthesis signal processing
    • Synthesis vocoders
    • Speech quality evaluation
  • Week 12
    • Noise
    • The Gaussian model
  • Week 13
    • The Wiener filter
    • The Ephraim Malah suppression filter
    • Voice activity detection

Caveat: Some of the slides above contain images “from the web” that omit an attribution or copyright. It’s my bad; I’ll fix them as I find them.