• Language: US English
  • Vocabulary: ~20K words (WSJ TCB20ONP)
  • Acoustic Model: GMM (HTK-trained)
Text speech duration recognition time recognition speed