Improving Audio Recognition With Randomized Area Ratio Patch Masking: A Data Augmentation Perspective
Sign Up to like & getrecommendations! Published in 2024 at "IEEE Access"
DOI: 10.1109/access.2024.3475508
Abstract: In audio recognition, improving the accuracy and generalizability of Pretrained Audio Neural Networks (PANNs) remains challenging. This study introduces Randomized Area Ratio Patch Masking (RARPM), a novel data augmentation technique that applies random patches with… read more here.
Keywords: randomized area; audio recognition; area ratio; data augmentation ... See more keywords