In feature, there are some feature extracting functions.
- (power) spectrogram feature.
- fbank feature.
- mfcc feature.
- energy feature.
- add delta to feature.
In gmm, there are some functions to train and test a GMM model.
- k-means algorithm.
- compute MFCC feature from files.
- train a GMM model with EM algorithm (Support parallel threads).
Hidden Markov Model
In hmm, there are some functions to train and query a HMM model.
- forward and backward algorithm of a HMM model.
- token passing viterbi decoding algorithm.