Towards Cross-Lingual Audio Abuse Detection in 

Low-Resource Settings with Few-Shot Learning