mahendrathapa / sarangi-ii Goto Github PK

LabelNotFoundException is thrown when predicting the labels of a single song.

This is because, in the predict() method, we are calling DatasetUtil.getSongwiseDataset() which tries to find the label for the song returned too.

A possible solution is to add a method in DatasetUtil that only returns the dataset of LearningDataset.

As it stands, the length of each feature is not clear.

Either we have to remember it by heart or try out ANN which points out the difference in expected and input vector size. SVM doesn't even do that (it runs silently).

An Enum of the features with a length attribute should do the job.