I want to split my audio dataset into train and test. can datasetLoader do this for me?

1

I just started coding in Pytorch. I have converted my wav files into text using glob library. But now I want to split that text file into train and test. Actually the dataset is very small and imbalanced. To be more clear it has 7 classes in file name only. But different classes have different samples like 100, 50 etc. Not How to split it into Train and Test
can anybody help plz

2

If you want to randomly split your Dataset, you could use torch.utils.data.random_split.
Alternatively, if you want to apply a stratified split, you could use sklearn.model_selection.train_test_split and pass the targets as the stratify argument.

1 Like
3

Thank you so much sir