Speech Recognition By Java - Simplified

Speech recognition is one of the challenging areas in computer science, a lot of pattern recognition methodology tried to resolve a good way and higher percentage of recognition.

One of the best ways to be use is Hidden Markov Model :



"The process of speech recognition is to find the best possible sequence of words (or units) that will fit the given input speech. It is a search problem, and in the case of HMM-based recognizers, a graph search problem. The graph represents all possible sequences of phonemes in the entire language of the task under consideration. The graph is typically composed of the HMMs of sound units concatenated in a guided manner, as specified by the grammar of the task."

There are a lot of Java Based Speech Recognition Engines, you can find them here:

http://ocvolume.sourceforge.net/links.php.

In this post we will select one of the famous open source Java speech recognition software and identify simple steps to use it.

Sphinx-4

Sphinx-4 is a state-of-the-art speech recognition system written entirely in the JavaTM programming language. It was created via a joint collaboration between the Sphinx group at Carnegie Mellon University, Sun Microsystems Laboratories, Mitsubishi Electric Research Labs (MERL), and Hewlett Packard (HP), with contributions from the University of California at Santa Cruz (UCSC) and the Massachusetts Institute of Technology (MIT).

Sphinx-4 started out as a port of Sphinx-3 to the Java programming language, but evolved into a recognizer designed to be much more flexible than Sphinx-3, thus becoming an excellent platform for speech research.


1) Download the software:

Here is the download link:

http://sourceforge.net/projects/cmusphinx/files/sphinx4/1.0%20beta6/sphinx4-1.0beta6-bin.zip/download.

2) Create new Java project:

Add to class-path 3 jars from the lib folder:
jsapi.jar (if not exist jsapi.exe will extract it on windows or jsapi.sh will extract it)
sphinx4.jar (APIs Jar)
WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz.jar (this is the jar containing the speech)


3) Create a new class to write the logic on it: e.g. ListenToMe.java
Let the package name osa.ora

4) Add the file dict.gram to the same package: osa.ora
It should looks like:
#JSGF V1.0;
grammar hello;
public <greet>= (Good morning | Hello | Hi | Welcome) ( Osama | Paul | Philip | Rita | Will );

The structure is to write the alternative words between () and separated by | and give them tag name to describe these words (greet here).

The grammar specs in this URL:
http://www.w3.org/TR/jsgf/
A quick tutorial in this URL:
http://puneetk.com/basics-of-java-speech-grammar-format-jsgf


5) Add the config file myconfig.xml to the same package: osa.ora
You can get the file structure from the jar in bin folder HelloWorld.jar
in the location "HelloWorld.jar\edu\cmu\sphinx\demo\helloworld\helloworld.config.xml"

You need to edit the following entry:
<component name="jsgfGrammar" type="edu.cmu.sphinx.jsgf.JSGFGrammar">
<property name="dictionary" value="dictionary"/>
<property name="grammarLocation"
value="resource:/osa/ora/"/>
<property name="grammarName" value="dict"/>
<property name="logMath" value="logMath"/>
</component>

6) Write the following code in the created class:

import edu.cmu.sphinx.frontend.util.Microphone;
import edu.cmu.sphinx.recognizer.Recognizer;
import edu.cmu.sphinx.result.Result;
import edu.cmu.sphinx.util.props.ConfigurationManager;

public static void main(String args[]){
ConfigurationManager cm = new ConfigurationManager(ListenToMe.class.getResource("myconfig.xml"));
Recognizer recognizer = (Recognizer)cm.lookup("recognizer");
recognizer.allocate();
Microphone microphone = (Microphone)cm.lookup("microphone");
if(!microphone.startRecording()){
System.out.println("Cannot start microphone.");
recognizer.deallocate();
System.exit(1);
}
System.out.println("Say: (Good morning | Hello | Hi | Welcome) ( Osama | Paul | Philip | Rita | Will )");
do {
System.out.println("Start speaking. Press Ctrl-C to quit.\n");
Result result = recognizer.recognize();
if(result != null) {
String resultText = result.getBestFinalResultNoFiller();
System.out.println((new StringBuilder()).append("You said: ").append(resultText).append('\n').toString());
} else {
System.out.println("I can't hear what you said.\n");
}
} while(true);
}


As you can see you can use ConfigurationManager to re-set the grammarName to dict1 , dict2 ,..etc, you can also have multiple configurations.

For more details about the usage you can refer to this URL:

http://cmusphinx.sourceforge.net/sphinx4/

You may use Interactive4J implementation direct to achieve dynamic voice recognition.. Find more details here.