Non-Stationary Modelling approaches to Speech Recognition
Roberto Togneri
UWA
Non-Stationary Modelling approaches to Speech Recognition
Contents
Speech Recognition: Too many varieties!
Speech Recognition: Too many problems!
What is speech? Physical View
What is speech? Phonetic view
Speech Analysis I
Speech Analysis II
Time-Domain Analysis
Frequency Domain Analysis
Speech Analysis: Summary
Feature Vector Sequences
Modelling Feature Vectors
Hidden Markov Model: Concept
Hidden Markov Model: Algorithms
Stochastic Segment Model: Concept
Stochastic Segment Model: Algorithms
Non-Stationary HMM: Concept
Non-Stationary HMM: Algorithms
Non-Stationary HMM: Results
Trajectory Model: Method
Trajectory Model: Experiment
Trajectory Model: Results I
Trajectory Model: Results II
Trajectory Model: Observations
Issues
Future Work
Conclusions