TranscribeFile.java

import java.io.File;
import java.io.FileInputStream;
import java.io.InputStream;
import java.io.OutputStream;
import java.io.FileOutputStream;
import java.io.PrintStream;

import edu.cmu.sphinx.api.Configuration;
import edu.cmu.sphinx.api.SpeechResult;
import edu.cmu.sphinx.api.StreamSpeechRecognizer;
import edu.cmu.sphinx.decoder.adaptation.Stats;
import edu.cmu.sphinx.decoder.adaptation.Transform;
import edu.cmu.sphinx.result.Result;
import edu.cmu.sphinx.decoder.search.Token;

public class TranscribeFile {
   public static void showUsage() {
      System.out.println("Usage: java com.example.TranscribeFile <input file> [<output file>]");
   }

   //returns a string representing <time> in the format
   //"hours:minutes:seconds.milliseconds"
   //
   //argument: <time> is the time in milliseconds
   public static String millisecondsToHMS(long time) {
      long h,m,s,mill;
      mill = time % 1000;
      time = (time - mill)/1000;
      s = time % 60;
      time = (time - s)/60;
      m = time % 60;
      time = (time - m)/60;
      h = time;

      return String.format("%d:%02d:%02d.%04d", h, m, s, mill);
   }

   public static void main(String[] args) throws Exception {
      boolean verbose = false;
      String in_file = null;
      String out_file = null;
      PrintStream out = null;

      //Parse options
      try {
         if (args.length < 1)
            throw new Exception("Too few arguments.");
         if (args.length > 2)
            throw new Exception("Too many arguments.");

         in_file = args[0];
         if (args.length >= 2)
            out_file = args[1];
      } catch (Exception e) {
         System.out.format("Error: %s\n", e.getMessage());
         showUsage();
         System.exit(1);
      }

      if (out_file != null)
         out = new PrintStream(new FileOutputStream(out_file));
      else
         out = System.out;
      Configuration configuration = new Configuration();

      //configuration.setAcousticModelPath("resource:/edu/cmu/sphinx/models/en-us/en-us");
      configuration.setAcousticModelPath("file:/Users/johnny/Downloads/cmusphinx-en-us-ptm-5.2");
      configuration.setDictionaryPath("resource:/edu/cmu/sphinx/models/en-us/cmudict-en-us.dict");
      //configuration.setDictionaryPath("file:/Users/johnny/Downloads/cmudict-0.7b.dict");
      //configuration.setLanguageModelPath("resource:/edu/cmu/sphinx/models/en-us/en-us.lm.bin");
      configuration.setLanguageModelPath("file:/Users/johnny/Downloads/en-70k-0.2-pruned.lm");

      StreamSpeechRecognizer recognizer = new StreamSpeechRecognizer(configuration);
      InputStream stream;

      //SpeechResult result;
      stream = new FileInputStream(new File(in_file));
      recognizer.startRecognition(stream);
      SpeechResult result;
      Token token;
      long time;
      while ((result = recognizer.getResult()) != null) {
         token = result.getResult().getBestToken();
         if (token != null) {
            time = token.getCollectTime();
            out.format("%s: %s\n", millisecondsToHMS(time), token.getWordPathNoFiller()); //print result together with timecode
         } else {
            out.println("");
         }
      }
      recognizer.stopRecognition();
      stream.close();
      out.close();
   }
}