{"title":"A closer look at deep learning neural networks with low-level spectral periodicity features","authors":"Bob L. Sturm, Corey Kereliuk, A. Pikrakis","doi":"10.1109/CIP.2014.6844511","DOIUrl":null,"url":null,"abstract":"Systems built using deep learning neural networks trained on low-level spectral periodicity features (DeSPerF) reproduced the most “ground truth” of the systems submitted to the MIREX 2013 task, “Audio Latin Genre Classification.” To answer why this was the case, we take a closer look at the behavior of a DeSPerF system we create and evaluate using the benchmark dataset BALLROOM. We find through time stretching that this DeSPerF system appears to obtain a high figure of merit on the task of music genre recognition because of a confounding of tempo with “ground truth” in BALLROOM. This observation motivates several predictions.","PeriodicalId":117669,"journal":{"name":"2014 4th International Workshop on Cognitive Information Processing (CIP)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2014-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"11","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2014 4th International Workshop on Cognitive Information Processing (CIP)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CIP.2014.6844511","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 11
Abstract
Systems built using deep learning neural networks trained on low-level spectral periodicity features (DeSPerF) reproduced the most “ground truth” of the systems submitted to the MIREX 2013 task, “Audio Latin Genre Classification.” To answer why this was the case, we take a closer look at the behavior of a DeSPerF system we create and evaluate using the benchmark dataset BALLROOM. We find through time stretching that this DeSPerF system appears to obtain a high figure of merit on the task of music genre recognition because of a confounding of tempo with “ground truth” in BALLROOM. This observation motivates several predictions.
使用深度学习神经网络构建的系统,经过低水平频谱周期性特征(DeSPerF)的训练,再现了提交给MIREX 2013任务“Audio Latin Genre Classification”的系统的最“基本事实”。为了回答为什么会出现这种情况,我们仔细研究了我们使用基准数据集BALLROOM创建和评估的一个DeSPerF系统的行为。我们发现,通过时间的延伸,这种DeSPerF系统似乎在音乐类型识别的任务上获得了很高的价值,因为在舞厅中节奏与“地面真理”的混淆。这一观察结果引发了一些预测。