Basic usage of the CMU Sphinx voice recognition toolkit in JRuby

transcriber.rb

require "java"
require "benchmark"

# To install and build CMU Sphinx, from the directory where you wish to run this script:
#
# svn co https://svn.code.sf.net/p/cmusphinx/code/trunk/sphinx4
# cd sphinx4
# ant
#

$CLASSPATH << "sphinx4/bld"

import "edu.cmu.sphinx.frontend.util.AudioFileDataSource"
import "edu.cmu.sphinx.recognizer.Recognizer"
import "edu.cmu.sphinx.result.Result"
import "edu.cmu.sphinx.util.props.ConfigurationManager"

import "javax.sound.sampled.UnsupportedAudioFileException"
import "java.io.IOException"
import "java.net.URL"

class Transcriber

  # Transcribe an audio file
  # @param [String] audio_file_name
  def self.run(audio_file_name)
    transcriber = new(audio_file_name)
    transcriber.run
  end

  # @param [String] audio_file_name The name of an audio file to transcribe
  def initialize(audio_file_name)
    @audio_file_name = audio_file_name
    configure
  end

  # Run the transcriber
  # @return [Array<String>] Transcribed text
  def run
    @results ||= []
    while !(result = @recognizer.recognize).nil? do
      string = result.get_best_result_no_filler
      @results << string if string != ""
    end
    @results
  end

  private

  # Configure the transcriber
  def configure
    load_configuration
    configure_recognizer
    load_audio
  end

  # Load the CMU Sphinx configuration XML file
  def load_configuration
    @class_loader = JRuby.runtime.get_jruby_class_loader
    Java::java.lang.Thread.current_thread.set_context_class_loader(@class_loader)

    resource = @class_loader.get_resource("sphinx4/bld/edu/cmu/sphinx/demo/transcriber/config.xml")
    @configuration = ConfigurationManager.new(resource)
  end

  # Configure the CMU Sphinx recognizer object
  def configure_recognizer
    @recognizer = @configuration.lookup("recognizer")
    @recognizer.allocate
  end

  # Load the audio file
  def load_audio
    audio_url = @class_loader.get_resource(@audio_file_name)

    data_source = @configuration.lookup("audioFileDataSource")
    data_source.set_audio_file(audio_url, nil)
  end

end

# Run
output = []

benchmark = Benchmark.measure do 
  output = Transcriber.run("test.wav")
end

puts output
puts benchmark