Bi-LSTM Model for Speech Recognition