Korean Singing Voice Synthesis Based on an LSTM Recurrent Neural Network