检索结果-内蒙古大学图书馆

25th Interspeech Conference

作者： Dinkel, Heinrich Yan, Zhiyong Wang, Yongqing Zhang, Junbo Wang, Yujun Wang, Bin Xiaomi Corp AI Lab Beijing Peoples R China

Despite progress in audio classification, a generalization gap remains between speech and other sound domains, such as environmental sounds and music. Models trained for speech tasks often fail to perform well on environmental or musical audio tasks, and vice versa. While self-supervised (SSL) audio representations offer an alternative, there has been limited exploration of scaling both model and dataset sizes for SSL-based general audio classification. We introduce Dasheng, a simple SSL audio encoder, based on the efficient masked autoencoder framework. Trained with 1.2 billion parameters on 272,356 hours of diverse audio, Dasheng obtains significant performance gains on the HEAR benchmark. It outperforms previous works on CREMA-D, LibriCount, Speech Commands, VoxLingua, and competes well in music and environment classification. Dasheng features inherently contain rich speech, music, and environmental information, as shown in nearest-neighbor classification experiments.

关键词： audio classification General audio feature Transformer Masked auto encoder

来源：评论

学校读者我要写书评

暂无评论

Self-Supervised Learning-Based General Fine-tuning Framework For audio classification and Event Detection

Self-Supervised Learning-Based General Fine-tuning Framework...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Sun, Yanjie Xu, Kele Dou, Yong Gao, Tian Natl Univ Def Technol Changsha Peoples R China iFlytek Res Hefei Peoples R China

ISBN: (纸本)9798350390155;9798350390162

Recently, self-supervised learning (SSL) has made remarkable progress in signal representation and has become a de facto solution for different audio processing tasks. Generally, the SSL consists of the foundation pre-training and downstream fine-tuning phases. However, fine-tuning frameworks may lack universality due to the distinct learning paradigms and model designs employed in audio signal processing tasks. Furthermore, the varying degrees of dataset labeling across different tasks challenge unifying a fine-tuning framework. To address these issues, we propose vec2task, a cross-task general fine-tuning framework based on the SSL pre-trained model. It employs a semantic-aware module and an alternating training strategy, enabling the framework to generalize across various audio signal processing tasks. Additionally, the framework employs automatic audio augmentation strategies, eliminating the requirement for individually tailored algorithms to improve task performance. Experimental validations of the vec2task framework outperformed previous methods in audio classification and event detection tasks, showcasing its generalization ability across tasks.

关键词： audio classification audio event detection Semantic-aware Self-supervised learning Fine-tuning

来源：评论

学校读者我要写书评

暂无评论

MULTI-LABEL ZERO-SHOT audio classification WITH TEMPORAL ATTENTION 18

MULTI-LABEL ZERO-SHOT AUDIO CLASSIFICATION WITH TEMPORAL ATT...

引用

18th International Workshop on Acoustic Signal Enhancement (IWAENC)

作者： Dogan, Duygu Xie, Huang Heittola, Toni Virtanen, Tuomas Tampere Univ Unit Comp Sci Tampere Finland

ISBN: (纸本)9798350361865;9798350361858

Zero-shot learning models are capable of classifying new classes by transferring knowledge from the seen classes using auxiliary information. While most of the existing zero-shot learning methods focused on single-label classification tasks, the present study introduces a method to perform multi-label zero-shot audio classification. To address the challenge of classifying multi-label sounds while generalizing to unseen classes, we adapt temporal attention. The temporal attention mechanism assigns importance weights to different audio segments based on their acoustic and semantic compatibility, thus enabling the model to capture the varying dominance of different sound classes within an audio sample by focusing on the segments most relevant for each class. This leads to more accurate multi-label zero-shot classification than methods employing temporally aggregated acoustic features without weighting, which treat all audio segments equally. We evaluate our approach on a subset of audioSet against a zero-shot model using uniformly aggregated acoustic features, a zero-rule baseline, and the proposed method in the supervised scenario. Our results show that temporal attention enhances the zero-shot audio classification performance in multi-label scenario.

关键词： multi-label zero-shot learning audio classification audio tagging temporal attention

来源：评论

学校读者我要写书评

暂无评论

Enhancing Zero-shot audio classification using Sound Attribute Knowledge from Large Language Models 25

Enhancing Zero-shot Audio Classification using Sound Attribu...

引用

25th Interspeech Conference

作者： Xu, Xuenan Zhang, Pingyue Yang, Ming Zhang, Ji Wu, Mengyue Shanghai Jiao Tong Univ MoE Key Lab Artificial Intelligence X LANCE Lab Shanghai Peoples R China Alibaba Grp Inst Intelligent Comp Hangzhou Peoples R China

Zero-shot audio classification aims to recognize and classify a sound class that the model has never seen during training. This paper presents a novel approach for zero-shot audio classification using automatically generated sound attribute descriptions. We propose a list of sound attributes and leverage large language model's domain knowledge to generate detailed attribute descriptions for each class. In contrast to previous works that primarily relied on class labels or simple descriptions, our method focuses on multi-dimensional innate auditory attributes, capturing different characteristics of sound classes. Additionally, we incorporate a contrastive learning approach to enhance zero-shot learning from textual labels. We validate the effectiveness of our method on VGGSound and audioSet(1). Our results demonstrate a substantial improvement in zero-shot classification accuracy. Ablation results show robust performance enhancement, regardless of the model architecture.

关键词： zero-shot learning audio classification sound attribute large language model audio-text contrastive learning

来源：评论

学校读者我要写书评

暂无评论

Few-shot audio classification using Contrastive Training 32

Few-shot Audio Classification using Contrastive Training

引用

32nd IEEE Signal Processing and Communications Applications Conference (SIU)

作者： Cigdem, Enes Furkan Keles, Hacer Yalim Hacettepe Univ Ankara Turkiye SESTEK Istanbul Turkiye

ISBN: (纸本)9798350388978;9798350388961

In this study, the issue of few-shot audio classification in scenarios with limited labeled data is addressed, and experiments conducted on the GSC and ESC-50 audio datasets are presented. The study examines three experimental setups structured around training scenarios with 5, 10, and 15 samples. In all these experiments, accuracy values obtained using 1 and 5 audio recordings per instance for 5-class situations are compared. Trainings were conducted with three different loss optimizations, and the effects of simple feature transformations on classification performance for each training were also assessed. The findings indicate that these feature transformations enhance classification accuracy in both datasets. Notably, the hybrid approach, which combines simultaneous contrastive loss with few sample cross-entropy loss, achieved the highest classification performance in the fine-tuned scenario. In this context, tests conducted with 5 samples for 5 classes yielded success rates ranging between 86% and 91% in ESC-50 dataset, 91% and 95% in GSC dataset, depending on the number of samples used in training.

关键词： Few-shot Learning Contrastive Training Nonepisodic training audio classification Neural Speech Embedding

来源：评论

学校读者我要写书评

暂无评论

Secure audio classification for Voice Assistants: A Multi-Party Homomorphic Approach 16

Secure Audio Classification for Voice Assistants: A Multi-Pa...

引用

16th International Conference on Human System Interaction (HSI)

作者： Heya, Tasnia Ashrafi Serwadda, Abdul Texas Tech Univ Lubbock TX 79409 USA

ISBN: (纸本)9798350362923;9798350362916

The widespread adoption of smart speaker technologies like Amazon Echo and Google Home has significantly embedded them into our everyday lives. These devices offer advanced features, including emergency services and smart home capabilities, through environmental sound source detection. However, they also face cybersecurity risks such as backdoor and adversarial attacks, which exploit server-side vulnerabilities. In our experiment, we initially conducted a user survey (n=97) to get an overview of user perspectives on security concerns related to voice assistants. To counteract these threats, our study investigates a secure multi-party homomorphic neural network for audio classification, aiming to safeguard against such threats by processing encrypted audio data. We collected data from 20 homes, extracted time-series features, and encrypted these for input into the Homomorphic Neural Network (HNN), leading to encrypted predictions. The model demonstrated a high accuracy of 93.18% in identifying various audio objects (11 types in this study). This research not only assesses accuracy but also contrasts the multi-party homomorphic method with conventional neural networks across various performance indicators, highlighting the efficiencies, and potential challenges of using encrypted models.

关键词： Homomorphic neural network Smart speakers Voice Assistants audio classification Multy-party computation

来源：评论

学校读者我要写书评

暂无评论

EPISODIC FINE-TUNING PROTOTYPICAL NETWORKS FOR OPTIMIZATION-BASED FEW-SHOT LEARNING: APPLICATION TO audio classification 34

EPISODIC FINE-TUNING PROTOTYPICAL NETWORKS FOR OPTIMIZATION-...

引用

34th International Workshop on Machine Learning for Signal Processing

作者： Zhuang, Xuanyu Peeters, Geoffroy Richard, Gael IP Paris Telecom Paris LTCI Paris France

ISBN: (纸本)9798350372267;9798350372250

The Prototypical Network (ProtoNet) has emerged as a popular choice in Few-shot Learning (FSL) scenarios due to its remarkable performance and straightforward implementation. Building upon such success, we first propose a simple (yet novel) method to fine-tune a ProtoNet on the (labeled) support set of the test episode of a C-way-K-shot test episode (without using the query set which is only used for evaluation). We then propose an algorithmic framework that combines ProtoNet with optimization-based FSL algorithms (MAML and Meta-Curvature) to work with such a fine-tuning method. Since optimization-based algorithms endow the target learner model with the ability to fast adaption to only a few samples, we utilize ProtoNet as the target model to enhance its fine-tuning performance with the help of a specifically designed episodic fine-tuning strategy. The experimental results confirm that our proposed models, MAML-Proto and MC-Proto, combined with our unique fine-tuning method, outperform regular ProtoNet by a large margin in few-shot audio classification tasks on the ESC-50 and Speech Commands v2 datasets. We note that although we have only applied our model to the audio domain, it is a general method and can be easily extended to other domains.

关键词： Few-shot learning audio classification Prototypical Network Model-Agnostic Meta-Learning Meta-Curvature

来源：评论

学校读者我要写书评

暂无评论

audio classification using grasshopper-ride optimization algorithm-based support vector machine

引用

IET CIRCUITS DEVICES & SYSTEMS 2021年第5期15卷 434-447页

作者： Singh, Suryabhan Pratap Jaiswal, Umesh Chandra Madan Mohan Malaviya Univ Technol Dept Comp Sci & Engn Gorakhpur Uttar Pradesh India Madan Mohan Malaviya Univ Technol Dept Informat Technol & Comp Applicat Gorakhpur Uttar Pradesh India

The accurate and robust detection of the audio has been widely grown as the speech technology in the area of audio forensics, speech recognition, and so on. However, in real time, it is a challenge to deal with the massive data arriving from distributed sources. Thus, the study introduces a method that effectively deals with the data from the distributed sources using the map-reduce framework (MRF). The map and reduce function in MRF aim at feature extraction and audio classification. The robust classification using the proposed grasshopper-ride optimization algorithm-based support vector machine (G-ROA-based SVM) uses the features, such as multiple kernel Mel frequency cepstral coefficients, spectral flux, spectral kurtosis, and delta-amplitude modulation spectrogram. The proposed G-ROA is the integration of ROA and grasshopper optimization algorithm in tuning the optimal weights of SVM and also, the kernel function in SVM is modified using the Gaussian radial basis function, Gaussian kernel, and polynomial kernels. The experimentation of the proposed method is done using two datasets, namely TUT sound event 2017 dataset and ESC dataset. TUT sound event 2017 dataset consists of eight audio recordings from a single acoustic scene. ESC dataset consists of three parts and 252,400 recordings. The analysis reveals that the proposed audio classification acquired the maximal accuracy of 0.96, minimal false alarm rate, and false rejection rate of 0.022 and 0.0119, respectively.

关键词： audio signal processing feature extraction signal classification G-ROA-based SVM robust detection robust classification map-reduce framework accurate detection distributed sources speech recognition support vector machines radial basis function networks audio forensics audio recordings TUT sound event 2017 dataset grasshopper optimization algorithm audio recording audio classification grasshopper-ride optimization algorithm-based support vector machine multiple kernel Mel frequency cepstral coefficients cepstral analysis

来源：评论

学校读者我要写书评

暂无评论

Generate Unnoticeable Adversarial Examples on audio classification Models with Multi-perspective Objectives

Generate Unnoticeable Adversarial Examples on Audio Classifi...

引用

2024 International Conference on Computing, Communication, Cybersecurity and AI

作者： Ma, Bin Li, Weixun Li, Huifeng Sun, Guanghui Gao, Minghui Wu, Dexin Xu, Jian State Grid Hebei Elect Power Co Ltd Shijiazhuang Hebei Peoples R China State Grid Hebei Elect Power Res Inst Shijiazhuang Hebei Peoples R China NARI Grp Corp State Grid Elect Power Res Inst Nanjing Peoples R China Beijing Kedong Elect Power Control Syst Co Ltd Beijing Peoples R China Northeastern Univ Software Coll Shenyang Peoples R China

ISBN: (纸本)9783031744426;9783031744433

In recent years, there has been notable progress in the development of audio classification models, yet they remain vulnerable to adversarial attacks. While research into attack methodologies pertaining to images has experienced significant advancements, comparatively less attention has been directed towards attacks specifically targeting audio classification models. Furthermore, prevailing techniques for compromising audio models often yield adversarial examples that are readily identifiable due to discernible noise artifacts, thereby distinguishing them from clean samples. To address this disparity, we propose Multi-perspective Unnoticeable audio Attack (MU2A for short), an innovative framework designed for crafting imperceptible adversarial examples for both untargeted and targeted assaults on audio classification models. Concretely, we frame our attack strategy as an optimization dilemma and incorporate the Multi-Scale Spectral (MSS) score, Decibels score, and Mutual Information within the objective function. This formulation aims to minimize the disparity between clean audio signals and their adversarial counterparts across frequency and time domains, as well as probabilistic perspectives. Empirical evaluations demonstrate that the magnitude of the adversarial perturbation generated by MU2A represents only 0.625% of the clean example, rendering the resulting adversarial examples acoustically indistinguishable to human perception.

关键词： audio classification Adversarial Attack Deep Learning

来源：评论

学校读者我要写书评

暂无评论

LIGHT GATED MULTI MINI-PATCH EXTRACTOR FOR audio classification 49

LIGHT GATED MULTI MINI-PATCH EXTRACTOR FOR AUDIO CLASSIFICAT...

引用

49th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： He, Bo Zhang, Shiqi Wang, Xianrui Qiu, Zheng Takeuchi, Daiki Niizumi, Daisuke Harada, Noboru Makino, Shoji Waseda Univ Tokyo Japan NTT Corp Tokyo Japan

ISBN: (纸本)9798350374520;9798350374513

audio classification, which serves as a fundamental step for acoustic signal processing, has attracted a lot of researchers interest, and numerous audio classification neural networks have been proposed. In these networks, pooling modules, which compress audio features, are essential due to their computational capacity. However, compressing the signal will inevitably cause the loss of relevant information. To mitigate this issue, a large number of parameters should be used. In this paper, we present a novel pooling method called gated multi mini-patch extractor (GMME), in which multiple convolution layers are used to extract relevant information at different aspects, including time frames, pseudo-frequency bins, and global features. Besides, the gate mechanism is adopted to emphasize the correlation among the original features. Several simulations demonstrate that, compared to the baseline, our method can achieve comparable or slightly better performance with a significant reduction in model size.

关键词： audio classification feature extraction pooling module gated multi mini-patch extractor

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：