ICSI Speech FAQ:
5.9 How can I run the SRI front-end standalone?

Answer by: wooters - 2002-09-06

Many people have asked how to run the front-end of the SRI system in a standalone mode. Here is a note from Andreas about how to do it: 

Subject: running the SRI front-end standalone
Date: Wed, 28 Aug 2002 15:18:29 PDT
From: Andreas Stolcke

I installed at SRI and ICSI a new program that Ramana wrote,
"create-normalized-feature-files" that computes feature files using
the Decipher frontend.  It is installed in
$EVAL2000/bin/$MACHINE_TYPE.  Running it without arguments gives a
usage message.

I also wrote a wrapper script, go.dump-features, that uses standard
eval system configuration values to dump feature files for a set of
wavefiles.  It has a man page.  By combining it with go.fe-normalize
(or not) you can produce either speaker-normalized or raw features.

--Andreas
The environment variable "$EVAL2000" should be set to:
/u/drspeech/opt/eval2000/system
Finally, to be able to access the man page for "go.fe-normalize", you need to add $EVAL2000/man to your $MANPATH environment variable.

Previous: 5.8 How do different features compare performancewise? - Next: 6.1 What is the function of the neural net?
Back to ICSI Speech FAQ index

Generated by build-faq-index on Tue Mar 24 16:18:16 PDT 2009