Acoustic scene classification using convolutional neural network and multiple-width frequency-delta data augmentation

Abstract

In recent years, neural network approaches have shown superior performance to conventional hand-made features in numerous application areas. In particular, convolutional neural networks (ConvNets) exploit spatially local correlations across input data to improve the performance of audio processing tasks, such as speech recognition, musical chord recognition, and onset detection. Here we apply ConvNet to acoustic scene classification, and show that the error rate can be further decreased by using delta features in the frequency domain. We propose a multiplewidth frequency-delta (MWFD) data augmentation method that uses static mel-spectrogram and frequency-delta features as individual input examples. In addition, we describe a ConvNet output aggregation method designed for MWFD augmentation, folded mean aggregation, which combines output probabilities of static and MWFD features from the same analysis window using multiplication first, rather than taking an average of all output probabilities. We describe calculation results using the DCASE 2016 challenge dataset, which shows that ConvNet outperforms both of the baseline system with hand-crafted features and a deep neural network approach by around 7%. The performance was further improved (by 5.7%) using the MWFD augmentation together with folded mean aggregation. The system exhibited a classification accuracy of 0.831 when classifying 15 acoustic scenes.

A Two-Stage Approach to Device-Robust Acoustic Scene Classification

“…In recent years, the organizers of the Detection and Classification of Acoustic Scenes and Events (DCASE) challenge [1,2,3] provided both the benchmark data and a competitive platform to promote acoustic scene research and analysis. If we are to analyze top ASC systems reported in the challenge, we will find that most of them are built on the deep neural networks (DNNs) framework, and the key ingredient of their success is the use of convolutional layers [4,5,6,7,8]. Advanced deep learning techniques, such as attention mechanism [9,10], mix-up [11,12], generative adversarial network (GAN) and variational auto encoder (VAE) based data augmentation [13,14], and deep feature learning [15,16,17] can further enhance ASC results.…”

Relational Teacher Student Learning with Neural Label Embedding for Device Adaptation in Acoustic Scene Classification

“…In recent years, we have witnessed a great progress in the acoustic scene classification (ASC) task, as demonstrated by the high participation in the IEEE Detection and Classification of Acoustic Scenes and Events (DCASE) challenges [1,2,3]. Top ASC systems use deep neural networks (DNNs), and the main ingredient of their success is the application of deep convolutional neural networks (CNNs) [4,5,6,7,8,9]. Further boost in ASC performance is obtained with the introduction of advanced deep learning techniques, such as attention mechanism [10,11,12], mix-up [13,14], Generative Adversial Network (GAN) and Variational Auto Encoder (VAE) based data augmentation [15,16], and deep feature learning [17,18,19,20].…”

An Acoustic Segment Model Based Segment Unit Selection Approach to Acoustic Scene Classification with Partial Utterances

“…ASC has been an attracting research field for decades, and the IEEE Detection and Classification of Acoustic Scenes and Events (DCASE) challenge [1,2,3] provides the benchmark data and a competitive platform to promote sound scene research and analyses. In recent years, we have witnessed that the deep neural networks (DNNs) have gradually dominated the design of top ASC systems, and the main ingredient of their success is the application of deep convolutional neural networks (CNNs) [4,5,6,7]. Furthermore, with the use of advanced deep learning techniques, such as attention mechanism [8,9,10] and deep network based data augmentation [11,12,13], a further boost in ASC system performances can be obtained.…”