Prepare and Curate training and test data for transfer learning

[technosaby]

Use techniques to curate sound effects labels from text/transcripts and use them as training data ? Need to try as suggested by prof Urhig. Reference: https://www.kaggle.com/datasets/amandam1/bridgerton-transcripts-cast