An empirical comparison of machine learning techniques for chant classification

A comparative evaluation of applying algorithms is presented, based on Hidden Markov Models and Artificial Neural Networks, in order to create a corpus of chants through vocal music recognition. The sound signal sequences of singing performances were captured, a training dataset was created based on the extracted sound features of the captured sound signal sequences and finally, the music recognition system was trained using the mentioned algorithms. Finally, the music recognition was performed and a score of successfully recognized of hymn performances was calculated by utilizing the cross - validation statistical method Jackknife. The results of the evaluation revealed that HMM algorithm is more efficient than ANN in order to train a machine learning system for chanting recognition. The findings can be used to build and/or improve the performance of machine learning systems for monophonic singing recognition.