Dataset length: 40 min (from speech)

Extraction: Mangio-Crepe

Batch Size: 7

Epoch: 250