While efficient architectures and a plethora of augmentations for end-to-end image classification tasks have been suggested and heavily investigated, state-of-the-art techniques for audio classifications still rely on numerous representations of the audio signal together with large architectures, fine-tuned from large datasets. By utilizing the inherited lightweight nature of audio and novel audio augmentations, we were able to present an efficient end-to-end network with strong generalization ability. Experiments on a variety of sound classification sets demonstrate the effectiveness and robustness of our approach, by achieving state-of-the-art results in various settings. Public code will be available.
%0 Generic
%1 Gazneli.25.04.2022
%A Gazneli, Avi
%A Zimerman, Gadi
%A Ridnik, Tal
%A Sharir, Gilad
%A Noy, Asaf
%D 2022
%K from:lukasbarth imported ma_ss22_ts
%T End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network
%U https://arxiv.org/pdf/2204.11479
%X While efficient architectures and a plethora of augmentations for end-to-end image classification tasks have been suggested and heavily investigated, state-of-the-art techniques for audio classifications still rely on numerous representations of the audio signal together with large architectures, fine-tuned from large datasets. By utilizing the inherited lightweight nature of audio and novel audio augmentations, we were able to present an efficient end-to-end network with strong generalization ability. Experiments on a variety of sound classification sets demonstrate the effectiveness and robustness of our approach, by achieving state-of-the-art results in various settings. Public code will be available.
@misc{Gazneli.25.04.2022,
abstract = {While efficient architectures and a plethora of augmentations for end-to-end image classification tasks have been suggested and heavily investigated, state-of-the-art techniques for audio classifications still rely on numerous representations of the audio signal together with large architectures, fine-tuned from large datasets. By utilizing the inherited lightweight nature of audio and novel audio augmentations, we were able to present an efficient end-to-end network with strong generalization ability. Experiments on a variety of sound classification sets demonstrate the effectiveness and robustness of our approach, by achieving state-of-the-art results in various settings. Public code will be available.},
added-at = {2022-07-05T18:31:53.000+0200},
author = {Gazneli, Avi and Zimerman, Gadi and Ridnik, Tal and Sharir, Gilad and Noy, Asaf},
biburl = {https://www.bibsonomy.org/bibtex/2067be862a7296dec6f25c9b5e604f603/lukasbarth},
file = {Gazneli, Zimerman et al. 25.04.2022 - End-to-End Audio Strikes Back:Attachments/Gazneli, Zimerman et al. 25.04.2022 - End-to-End Audio Strikes Back.pdf:application/pdf},
interhash = {815025a660becba2a7847523ceb9641d},
intrahash = {067be862a7296dec6f25c9b5e604f603},
keywords = {from:lukasbarth imported ma_ss22_ts},
timestamp = {2022-07-05T18:32:19.000+0200},
title = {End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network},
url = {https://arxiv.org/pdf/2204.11479},
year = 2022
}