SpeechRecognitionAlgorithmsUsingWeightedFinite-StateTransducers