Have a basic question regarding speech recognition. There is something called a decision tree
which is used to classify senones by asking linguistic questions for creating acoustic model.
My question is how is this process incorporated in sphinxtrain and does one have to provide these
questions for training an acoustic model for english and other languages (Using sphinxtrain) ?
Thanks and regards,
Li
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
My question is how is this process incorporated in sphinxtrain and does one have to provide these
There are tools to build the tree which invoked during training on stage 40 (buildtree) and 45 (prunetree). You can look on their source in sphinxtrain.
does one have to provide these questions for training an acoustic model for english and other languages
It's possible to provide the questions. There is a corresponding item in sphinx_train.cfg config file. Or they could be created automatically. Manually provided questions are usually better.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
Have a basic question regarding speech recognition. There is something called a decision tree
which is used to classify senones by asking linguistic questions for creating acoustic model.
My question is how is this process incorporated in sphinxtrain and does one have to provide these
questions for training an acoustic model for english and other languages (Using sphinxtrain) ?
Thanks and regards,
Li
There are tools to build the tree which invoked during training on stage 40 (buildtree) and 45 (prunetree). You can look on their source in sphinxtrain.
It's possible to provide the questions. There is a corresponding item in sphinx_train.cfg config file. Or they could be created automatically. Manually provided questions are usually better.