Sphinx Documentation

Disclaimer : Information provided here may not represent the standpoints of Carnigie Mellon University or CMU Sphinx Group. This page contains all the documents and links I gathered about Sphinx. Always under heavy construction so check it out form time to time.

Sphinx 2 Information

Standard Sphinx 2 distribution that you can find in Sourceforge contains a documentation directory. That contains most information you need to use Sphinx 2. I will start with here . It is written by Ravi Mosur, the creator of Sphinx2/3/3.3.

Also check out the resource I gathered for sphinx 2 at the resource page .

Recently, at Jul, 2004, Ravi Mosur also built a FSM's version of Sphinx 2. You can found his presentation at here

Sphinx 3 Information

Sphinx 3 is the currently (at 20050115) the most active branch of CMU speech recognizer development.

For documentation of Sphinx3, check out the documentation in the package of sphinx3. I keep a local copy at here

Several things you will find interested in the package:

Also, please check out Sphinx 3 Twiki

If you want to dig out more, you could also check out the code documentation of Sphinx 3 . Remember, they are not completed.

Sphinx 4 Developer Information

Sphinx 4 is the latest recognizer jointly developed by CMU, Sun and Mitsubishi and HP with contribution from UCSC and MIT. Sphinx 4 has a very strong development team and aims at providing the best platform for speech researchers.

For documentation of Sphinx 4, please browse the official Sphinx 4 page

You can also find the Twiki forum of Sphinx 4 developer.

SphinxTrain Information

For user: If you want to train Sphinx II or Sphinx III models. The best document can be found is Rita Singh's Sphinx III's manual In instruction set for training , you will see a diagram that illustrates how training can be done using SphinxTrain.

There is also a pretty comprehensive tutorial from the robust group. In the tutorial, a recipe is included in the RM corpus. If you run it outside CMU, you need to specify the paths for the resource. You can find it here . This is based on the perl script written by Rich Houghton. There is also a csh version that is widely used internally in CMU. You could find it at here .

If you are very new to SphinxTrain, you might find that "10 Common Pitfalls of Using SphinxTrain" a good read.

new at 20040729 Another piece of information you will find it useful is the README.tracing I wrote for SphinxTrain. This is a note file I wrote while I am tracing SphinxTrain. It is more a hacker's point of view of the software.

Dhananjay Bansal's page (Obsolete): at here.

CMU LM Toolkit Information

Check out the document attached in the tar ball. You can access the html version at here and the text file at here

I also create the doxygen documentation for the code of LM Toolkit . As usual, it is not completed. V2.

OpenSALT Information

Check out Yitao Sun's maintained page on OpenSALT

He and Jahanzeb Sherwani also recently started a discussion at here