AclNet: efficient end-to-end audio classification CNN
This work addresses efficient audio classification for energy-efficient platforms, though it is incremental as it builds on existing CNN methods with optimizations.
The paper tackled audio classification by proposing AclNet, an efficient end-to-end CNN, achieving state-of-the-art performance with 85.65% accuracy on the ESC-50 corpus and demonstrating high accuracy at reduced computational complexity, such as 81.75% accuracy with only 155k parameters.
We propose an efficient end-to-end convolutional neural network architecture, AclNet, for audio classification. When trained with our data augmentation and regularization, we achieved state-of-the-art performance on the ESC-50 corpus with 85:65% accuracy. Our network allows configurations such that memory and compute requirements are drastically reduced, and a tradeoff analysis of accuracy and complexity is presented. The analysis shows high accuracy at significantly reduced computational complexity compared to existing solutions. For example, a configuration with only 155k parameters and 49:3 million multiply-adds per second is 81:75%, exceeding human accuracy of 81:3%. This improved efficiency can enable always-on inference in energy-efficient platforms.