Skip to content

Training a hidden Markov model through expectation-maximization, using Baum-Welch formulae, for applications in speech recognition

Notifications You must be signed in to change notification settings

rachelwiles/HMM-Speech-Recognition

Repository files navigation

HMM Speech Recognition 🎤

Training a hidden Markov model through expectation-maximization, using Baum-Welch formulae, for applications in speech recognition.

MATLAB assignment

A matrix, B matrix and time steps must be added to read_initialmatiricies before use.

Run in the following order:

  1. read_initialmatricies
  2. calc_B
  3. calc_forward
  4. calc_backward
  5. get_occupationlikelihoods
  6. reestimate_meanvariance
  7. gauss_graph
  8. get_transitionlikelihoods
  9. reestimate_A

File retrain.m may be used to perform many iterations of above procedure.

About

Training a hidden Markov model through expectation-maximization, using Baum-Welch formulae, for applications in speech recognition

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages