1 paper
Emiliano Acevedo, MartÃn Rocamora, Magdalena Fuentes
Audio-text models are widely used in zero-shot environmental sound classification as they alleviate the need for annotated data. However, we show that their performance severely dr…