A Low-Compexity Deep Learning FrameworkFor Acoustic Scene Classification (original) (raw)
Related papers
A Convolutional Neural Network Approach for Acoustic Scene Classification
—This paper presents a novel application of convo-lutional neural networks (CNNs) for the task of acoustic scene classification (ASC). We here propose the use of a CNN trained to classify short sequences of audio, represented by their log-mel spectrogram. We also introduce a training method that can be used under particular circumstances in order to make full use of small datasets. The proposed system is tested and evaluated on three different ASC datasets and compared to other state-of-the-art systems which competed in the " Detection and Classification of Acoustic Scenes and Events " (DCASE) challenges held in 2016 1 and 2013. The best accuracy scores obtained by our system on the DCASE 2016 datasets are 79.0% (development) and 86.2% (evaluation), which constitute a 6.4% and 9% improvements with respect to the baseline system. Finally, when tested on the DCASE 2013 evaluation dataset, the proposed system manages to reach a 77.0% accuracy, improving by 1% the challenge winner's score.
ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Acoustic Scene Classification (ASC) is one of the core research problems in the field of Computational Sound Scene Analysis. In this work, we present SubSpectralNet, a novel model which captures discriminative features by incorporating frequency band-level differences to model soundscapes. Using mel-spectrograms, we propose the idea of using band-wise crops of the input time-frequency representations and train a convolutional neural network (CNN) on the same. We also propose a modification in the training method for more efficient learning of the CNN models. We first give a motivation for using sub-spectrograms by giving intuitive and statistical analyses and finally we develop a sub-spectrogram based CNN architecture for ASC. The system is evaluated on the public ASC development dataset provided for the "Detection and Classification of Acoustic Scenes and Events" (DCASE) 2018 Challenge. Our best model achieves an improvement of +14% in terms of classification accuracy with respect to the DCASE 2018 baseline system. Code and figures are available at https:
Robust acoustic scene classification using a multi-spectrogram encoder-decoder framework
Digital Signal Processing, 2021
This article proposes an encoder-decoder network model for Acoustic Scene Classification (ASC), the task of identifying the scene of an audio recording from its acoustic signature. We make use of multiple low-level spectrogram features at the front-end, transformed into higher level features through a well-trained CNN-DNN front-end encoder. The high level features and their combination (via a trained feature combiner) are then fed into different decoder models comprising random forest regression, DNNs and a mixture of experts, for back-end classification. We report extensive experiments to evaluate the accuracy of this framework for various ASC datasets, including LITIS Rouen and IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events (DCASE) 2016 Task 1, 2017 Task 1, 2018 Tasks 1A & 1B and 2019 Tasks 1A & 1B. The experimental results highlight two main contributions; the first is an effective method for high-level feature extraction from multi-spectrogram input via the novel C-DNN architecture encoder network, and the second is the proposed decoder which enables the framework to achieve competitive results on various datasets. The fact that a single framework is highly competitive for several different challenges is an indicator of its robustness for performing general ASC tasks.
A Robust Framework for Acoustic Scene Classification
Interspeech 2019, 2019
Acoustic scene classification (ASC) using front-end timefrequency features and back-end neural network classifiers has demonstrated good performance in recent years. However a profusion of systems has arisen to suit different tasks and datasets, utilising different feature and classifier types. This paper aims at a robust framework that can explore and utilise a range of different time-frequency features and neural networks, either singly or merged, to achieve good classification performance. In particular, we exploit three different types of frontend time-frequency feature; log energy Mel filter, Gammatone filter and constant Q transform. At the back-end we evaluate effective a two-stage model that exploits a Convolutional Neural Network for pre-trained feature extraction, followed by Deep Neural Network classifiers as a post-trained feature adaptation model and classifier. We also explore the use of a data augmentation technique for these features that effectively generates a variety of intermediate data, reinforcing model learning abilities, particularly for marginal cases. We assess performance on the DCASE2016 dataset, demonstrating good classification accuracies exceeding 90%, significantly outperforming the DCASE2016 baseline and highly competitive compared to state-of-the-art systems.
URBAN SOUND CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORKS FOR DCASE 2020 CHALLENGE
This technical report describes our system proposed for Task 5-Urban Sound Tagging. The system has a core architecture based on Convolutional Neural Networks. This neural network uses log melspectrogram features as input and this input is processed by two CNN layers. The output of the convolutional stack is processed by several fully connected layers plus an output layer to produce the classification decision. Spatiotemporal context data is also available and we propose a multi-input architecture, with two input branches that are merged for the final processing. The spatiotemporal context information is processed by an additional neural network of 2 fully connected layers. Its output is merged with the output of the CNN stack and the resulting data is fed to the fully connected output block. In this report, we describe the proposed models in detail and compare them to the baseline approach using the provided development datasets. Finally, we present the results obtained with the validation split from the dataset.
1-D CNN based Acoustic Scene Classification via Reducing Layer-wise Dimensionality
arXiv (Cornell University), 2022
This paper presents an alternate representation framework to commonly used time-frequency representation for acoustic scene classification (ASC). A raw audio signal is represented using a pre-trained convolutional neural network (CNN) using its various intermediate layers. The study assumes that the representations obtained from the intermediate layers lie in low-dimensions intrinsically. To obtain low-dimensional embeddings, principal component analysis is performed, and the study analyzes that only a few principal components are significant. However, the appropriate number of significant components are not known. To address this, an automatic dictionary learning framework is utilized that approximates the underlying subspace. Further, the low-dimensional embeddings are aggregated in a late-fusion manner in the ensemble framework to incorporate hierarchical information learned at various intermediate layers. The experimental evaluation is performed on publicly available DCASE 2017 and 2018 ASC datasets on a pre-trained 1-D CNN, SoundNet. Empirically, it is observed that deeper layers show more compression ratio 1 than others. At 70% compression ratio across different datasets, the performance is similar to that obtained without performing any dimensionality reduction. The proposed framework outperforms the time-frequency representation based methods.
Acoustic Scene Analysis and Classification Using Densenet Convolutional Neural Network
In this paper we present an account of state-of the-art in Acoustic Scene Classification (ASC), the task of environmental scenario classification through the sounds they produce. Our work aims to classify 50 different outdoor and indoor scenario using environmental sounds. We use a dataset ESC-50 from the IEEE challenge on Detection and Classification of Acoustic Scenes and Events (DCASE). In this we propose to use 2000 different environmental audio recordings. In this method the raw audio data is converted into Mel-spectrogram and other characteristics like Tonnetz, Chroma and MFCC. The generated Mel-spectrogram is fed as an input to neural network for training. Our model follows structure of neural network in the form of convolution and pooling. With a focus on real time environmental classification and to overcome the problem of low generalization in the model, the paper introduced augmentation to achieve modified noise based audio by adding gaussian white noise. Active researche...
Sound Context Classification based on Joint Learning Model and Multi-Spectrogram Features
International Journal of Computing
This article presents a deep learning framework applied for Acoustic Scene Classification (ASC), the task of classifying different environments from the sounds they produce. To successfully develop the framework, we firstly carry out a comprehensive analysis of spectrogram representation extracted from sound scene input, then propose the best multi-spectrogram combination for front-end feature extraction. In terms of back-end classification, we propose a novel joint learning model using a parallel architecture of Convolutional Neural Network (CNN) and Convolutional Recurrent Neural Network (C-RNN), which is able to learn efficiently both spatial features and temporal sequences of a spectrogram input. The experimental results have proved our proposed framework general and robust for ASC tasks by three main contributions. Firstly, the most effective spectrogram combination is indicated for specific datasets that none of publication previously analyzed. Secondly, our joint learning arc...
Interspeech 2017, 2017
Enabling smart devices to infer about the environment using audio signals has been one of the several long-standing challenges in machine listening. The availability of public-domain datasets, e.g., Detection and Classification of Acoustic Scenes and Events (DCASE) 2016, enabled researchers to compare various algorithms on standard predefined tasks. Most of the current best performing individual acoustic scene classification systems utilize different spectrogram image based features with a Convolutional Neural Network (CNN) architecture. In this study, we first analyze the performance of a state-of-theart CNN system for different auditory image and spectrogram features, including Mel-scaled, logarithmically scaled, linearly scaled filterbank spectrograms, and Stabilized Auditory Image (SAI) features. Next, we benchmark an MFCC based Gaussian Mixture Model (GMM) SuperVector (SV) system for acoustic scene classification. Finally, we utilize the activations from the final layer of the CNN to form a SuperVector (SV) and use them as feature vectors for a Probabilistic Linear Discriminative Analysis (PLDA) classifier. Experimental evaluation on the DCASE 2016 database demonstrates the effectiveness of the proposed CNN-SV approach compared to conventional CNNs with a fully connected softmax output layer. Score fusion of individual systems provides up to 7% relative improvement in overall accuracy compared to the CNN baseline system.
2020 Fourth International Conference On Intelligent Computing in Data Sciences (ICDS), 2020
There are many sounds all around us and our brain can easily and clearly identify them. Furthermore, our brain processes the received sound signals continuously and provides us with relevant environmental knowledge. Although not up to the level of accuracy of the brain, there are some smart devices which can extract necessary information from an audio signal, with the help of different algorithms. And as the days pass by more, more research is being conducted to ensure that accuracy level of this information extraction increases. Over the years several models like the CNN, ANN, RCNN and many machine learning techniques have been adopted to classify sound accurately and these have shown promising results in the recent years in distinguishing spectra-temporal pictures. For our research purpose, we are using seven features which are Chromagram, Mel-spectrogram, Spectral contrast, Tonnetz, MFCC, Chroma CENS and Chroma cqt.We have employed two models for the classification process of audio signals which are LSTM and CNN and the dataset used for the research is the UrbanSound8K. The novelty of the research lies in showing that the LSTM shows a better result in classification accuracy compared to CNN, when the MFCC feature is used. Furthermore, we have augmented the UrbanSound8K dataset to ensure that the accuracy of the LSTM is higher than the CNN in case of both the original dataset as well as the augmented one. Moreover, we have tested the accuracy of the models based on the features used. This has been done by using each of the features separately on each of the models, in addition to the two forms of feature stacking that we have performed. The first form of feature stacking contains the features Chromagram, Mel-spectrogram, Spectral contrast, Tonnetz, MFCC, while the second form of feature stacking contains MFCC, Melspectrogram, Chroma cqt and Chroma stft. Likewise, we have stacked features using different combinations to expand our research.In such a way it was possible, with our LSTM model, to reach an accuracy of 98.80%, which is state-of-the-art performance.