Sphinx Update

2004 May 27 at 15:26 » Tagged as :

Since I last wrote about sphinx I ran a cvs update on the source tree. I made progress with this update, or at least I was now beginning to get a different set of exceptions(which is encouraging).

I noticed that most of the exceptions generated by the demos involved the WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz In my previous attempts I found that only the HelloDigits demo would start up and even that couldn't recognize what I was saying. Many a web page later I found someone had commented the the sphinx 4 project didn't check out correctly from the CVS repository. This could indeed be correct, because what's available on the webcvs interface and what's available on my local source tree appeared completely different.

One error in particular deals with a missing dictionary.

Problem when loading HelloNGram: java.io.FileNotFoundException: JAR entry edu/cmu/sphinx/model/acoustic/WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz/dict/cmudict.0.6d not found in /opt/voice/sphinx4/lib/sphinx4.jar

java.io.FileNotFoundException: JAR entry edu/cmu/sphinx/model/acoustic/WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz/dict/cmudict.0.6d not found in /opt/voice/sphinx4/lib/sphinx4.jar

at sun.net.www.protocol.jar.JarURLConnection.connect(JarURLConnection.java:97)

at sun.net.www.protocol.jar.JarURLConnection.getInputStream(JarURLConnection.java07)

at java.net.URL.openStream(URL.java:913)

at edu.cmu.sphinx.linguist.dictionary.FastDictionary.allocate(FastDictionary.java)

at edu.cmu.sphinx.linguist.lextree.LexTreeLinguist.allocate(LexTreeLinguist.java:403)

at edu.cmu.sphinx.decoder.search.WordPruningBreadthFirstSearchManager.allocate(WordPruningBreadthFirstSearchManager.java:324)

at edu.cmu.sphinx.decoder.Decoder.allocate(Decoder.java09)

at edu.cmu.sphinx.recognizer.Recognizer.allocate(Recognizer.java82)

at demo.sphinx.hellongram.HelloNGram.main(HelloNGram.java:53)

Indeed the dictionary reffered is not avalable in my local tree, I promptly downloaded the cmudict.0.6d file, copied to the source tree, compiled and re ran the demo. Geranimo!! another exception.

Exception in thread "main" java.lang.Error: Error loading word: ##

at edu.cmu.sphinx.linguist.dictionary.FastDictionary.loadDictionary(FastDictionary.java:228)

at edu.cmu.sphinx.linguist.dictionary.FastDictionary.allocate(FastDictionary.java)

at edu.cmu.sphinx.linguist.lextree.LexTreeLinguist.allocate(LexTreeLinguist.java:403)

at edu.cmu.sphinx.decoder.search.WordPruningBreadthFirstSearchManager.allocate(WordPruningBreadthFirstSearchManager.java:324)

at edu.cmu.sphinx.decoder.Decoder.allocate(Decoder.java09)

at edu.cmu.sphinx.recognizer.Recognizer.allocate(Recognizer.java82)

at demo.sphinx.hellongram.HelloNGram.main(HelloNGram.java:53)

Progress I am happy.

I then decided to clean up the whole tree and start all over again. Yesterday I tried an update but cvs reported an error (about some lock) I am all but ready to give up on CVS as a whole by the way.

The fresh check out is currently in progress internet connections in this country are abisimally slow.