sphinxtrain

INDEX
source: http://www.speech.cs.cmu.edu/sphinxman/scriptman1.html#24

1. Before you train
   • The general-procedure chart
   • Modeling context-dependent phones with untied states: some memory requirements
   • Data preparation
     • When you have a very small closed vocabulary
   • The set of base and higher order feature vectors
     • Feature streams
   • Force-alignment
2. Training continuous models
   • Creating the CI model definition file
   • Creating the HMM topology file
   • Flat initialization of CI model parameters
   • Training CI models
   • Creating the CD untied model definition file
   • Flat initialization of CD untied model parameters
   • Training CD untied models
   • Building decision trees for parameter sharing
     • Generating the linguistic questions
   • Pruning the decision trees
   • Creating the CD tied model definition file
   • Initializing and training cd tied gaussian mixture models
3. Training semi-continuous models
   • Vector quantization
   • Creating the CI model definition file
   • Creating the HMM topology file
   • Flat initialization of CI model parameters
   • Training CI models
   • Creating the CD untied model definition file
   • Flat initialization of CD untied model parameters
   • Training CD untied models
   • Building decision trees for parameter sharing
     • Generating the linguistic questions
   • Pruning the decision trees
   • Creating the CD tied model definition file
   • Initializing and training cd tied models
   • Deleted interpolation
4. SPHINX2 data and model formats
5. SPHINX3 data and model formats
6. Training multilingual models
7. The training lexicon
8. Converting SPHINX3 format models to SPHINX2 format
9. Updating or adapting existing model sets
10. Using the SPHINX-III decoder with semi-continuous and continuous models