今天跟大家分享一篇语音相关的论文合集：cs.SD语音6篇，eess.AS音频处理8篇。
【1】 The SpeakIn Speaker Verification System for Far-Field Speaker Verification Challenge 2022
标题：面向2022远场说话人确认挑战赛的说话人确认系统
链接：https://arxiv.org/abs/2209.11625
作者：Yu Zheng,Jinghan Peng,Yihao Chen,Yajun Zhang,Jialong Wang,Min Liu,Minqiang Xu机构：SpeakIn Technologies Co. Ltd.备注：5 pages. arXiv admin note: text overlap with arXiv:2209.10846摘要：本文介绍了SpeakIn团队提交给2022年远场说话人确认挑战赛（FFSVC 2022）任务1和任务2的说话人确认（SV）系统。该挑战的SV任务集中于全监督远场说话人确认（任务1）和半监督远场说话人确认（任务2）问题。在任务1中，我们使用VoxCeleb和FFSVC 2020数据集作为训练数据集。对于任务2，我们仅使用VoxCeleb数据集作为训练集。基于ResNet和基于RepVGG的体系结构就是为应对这一挑战而开发的。采用全局统计池结构和MQMHA池结构对帧级特征进行跨时间聚合，得到话语级特征表示。我们采用AM-Softmax和AAM-Softmax对得到的嵌入进行分类。我们创新性地提出了一种分阶段迁移学习方法。在预训练阶段，我们保留了说话人的权值，在这一阶段没有正样本进行训练。然后，在第二阶段，我们使用正样本和负样本对这些权重进行微调。与传统的迁移学习策略相比，该策略能更好地提高模型的性能。采用Sub-Mean和AS-Norm后端方法来解决域不匹配问题。在融合阶段，Task 1融合了三个模型，Task 2融合了两个模型。在FFSVC 2022排行榜上，我们提交的EER为3.0049%，任务1中相应的minDCF为0.2938。在任务2中，EER和最小DCF分别为6.2060%和0.5232。我们的方法带来了出色的性能，在两项挑战任务中均名列第一。摘要：This paper describes speaker verification (SV) systems submitted by the SpeakIn team to the Task 1 and Task 2 of the Far-Field Speaker Verification Challenge 2022 (FFSVC2022). SV tasks of the challenge focus on the problem of fully supervised far-field speaker verification (Task 1) and semi-supervised far-field speaker verification (Task 2). In Task 1, we used the VoxCeleb and FFSVC2020 datasets as train datasets. And for Task 2, we only used the VoxCeleb dataset as train set. The ResNet-based and RepVGG-based architectures were developed for this challenge. Global statistic pooling structure and MQMHA pooling structure were used to aggregate the frame-level features across time to obtain utterance-level representation. We adopted AM-Softmax and AAM-Softmax to classify the resulting embeddings. We innovatively propose a staged transfer learning method. In the pre-training stage we reserve the speaker weights, and there are no positive samples to train them in this stage. Then we fine-tune these weights with both positive and negative samples in the second stage. Compared with the traditional transfer learning strategy, this strategy can better improve the model performance. The Sub-Mean and AS-Norm backend methods were used to solve the problem of domain mismatch. In the fusion stage, three models were fused in Task1 and two models were fused in Task2. On the FFSVC2022 leaderboard, the EER of our submission is 3.0049% and the corresponding minDCF is 0.2938 in Task1. In Task2, EER and minDCF are 6.2060% and 0.5232 respectively. Our approach leads to excellent performance and ranks 1st in both challenge tasks.
【2】 Synthetic Voice Spoofing Detection Based On Online Hard Example Mining
标题：基于在线硬例挖掘的合成语音欺骗检测
链接：https://arxiv.org/abs/2209.11585
作者：Ruohua Zhou,Chenlei Hu,Qiuchen Yu,Yuxuan Du摘要：自动说话人确认欺骗（ASVspoof）挑战系列对于增强欺骗考虑和对抗措施的发展至关重要。虽然最新的ASVspoof 2019验证结果表明，该模型对大多数攻击具有显著的识别能力，但对部分攻击的识别效果仍然较差。提出了一种检测未知语音欺骗攻击的在线硬样本挖掘算法。OHEM用于克服数据集中简单样本和硬样本之间的不平衡。所提出的系统在ASVspoof 2019挑战逻辑访问场景的评估集上提供了0.77%的等错误率（EER）。摘要：The automatic speaker verification spoofing (ASVspoof) challenge series is crucial for enhancing the spoofing consideration and the countermeasures growth. Although the recent ASVspoof 2019 validation results indicate the significant capability to identify most attacks, the model's recognition effect is still poor for some attacks. This paper presents the Online Hard Example Mining (OHEM) algorithm for detecting unknown voice spoofing attacks. The OHEM is utilized to overcome the imbalance between simple and hard samples in the dataset. The presented system provides an equal error rate (EER) of 0.77% on the ASVspoof 2019 Challenge logical access scenario's evaluation set.
【3】 An artificial neural network-based system for detecting machine failures using tiny sound data: A case study
标题：基于人工神经网络的微小声音故障检测系统的实例研究
链接：https://arxiv.org/abs/2209.11527
作者：Thanh Tran,Sebastian Bader,Jan Lundgren机构：∗ STC Research Centre, Department of Electronics Design, Mid Sweden University备注：8 pages, 9 figures, conference摘要：为了推动基于深度学习的机器故障检测系统的研究，我们给出了一个基于微小声音数据集的案例研究。我们的案例研究调查了一种变分自动编码器（VAE），用于增强来自Valmet AB的小型钻孔声音数据集。Valmet数据集包含134种声音，分为两类：“异常”和“正常”记录来自Valmet AB公司的钻孔机，该公司位于瑞典Sundsvall，为生物燃料生产提供设备和工艺。使用深度学习模型来检测如此小的声音数据集上的失败演练通常是不成功的。我们采用VAE通过从原始声音合成新声音来增加微小数据集中的声音数量。通过将这些合成声音与原始声音组合来创建扩充数据集。我们使用了通带频率为1000 Hz的高通滤波器和通带频率为22\kern0.16667em000 Hz的低通滤波器，在将声音转换为Mel声谱图之前，对增强数据集中的声音进行预处理。然后使用这些Mel谱图训练预训练的2D-CNN Alexnet。当与使用原始微小声音数据集来训练预训练的Alexnet相比时，使用增强的声音数据集将CNN模型的分类结果提高了6.62%（在增强的数据集上训练时为94.12%，而在原始数据集上训练时为87.5%）。摘要：In an effort to advocate the research for a deep learning-based machine failure detection system, we present a case study of our proposed system based on a tiny sound dataset. Our case study investigates a variational autoencoder (VAE) for augmenting a small drill sound dataset from Valmet AB. A Valmet dataset contains 134 sounds that have been divided into two categories: "Anomaly" and "Normal" recorded from a drilling machine in Valmet AB, a company in Sundsvall, Sweden that supplies equipment and processes for the production of biofuels. Using deep learning models to detect failure drills on such a small sound dataset is typically unsuccessful. We employed a VAE to increase the number of sounds in the tiny dataset by synthesizing new sounds from original sounds. The augmented dataset was created by combining these synthesized sounds with the original sounds. We used a high-pass filter with a passband frequency of 1000 Hz and a low-pass filter with a passband frequency of 22\kern 0.16667em000 Hz to pre-process sounds in the augmented dataset before transforming them to Mel spectrograms. The pre-trained 2D-CNN Alexnet was then trained using these Mel spectrograms. When compared to using the original tiny sound dataset to train pre-trained Alexnet, using the augmented sound dataset enhanced the CNN model's classification results by 6.62\%(94.12\% when trained on the augmented dataset versus 87.5\% when trained on the original dataset).
【4】 UniKW-AT: Unified Keyword Spotting and Audio Tagging
标题：UniKW-AT：统一的关键字检测和音频标注
链接：https://arxiv.org/abs/2209.11377
作者：Heinrich Dinkel,Yongqing Wang,Zhiyong Yan,Junbo Zhang,Yujun Wang机构：Xiaomi Corperation, Beijing, China备注：Accepted in Interspeech2022摘要：在音频研究界和业界，关键词识别（KWS）和音频标注（AT）被视为两个不同的任务和研究领域。但是，从技术角度看，这两项任务是相同的：它们为一些固定大小输入音频段预测标签（KWS中的关键字，AT中的声音事件）。这项工作提出UniKW-AT：联合培训KWS和AT的初步方法。UniKW-AT增强了KWS的噪声鲁棒性，同时还能够预测特定的声音事件，并支持对声音事件的有条件唤醒。我们的方法用描述关键字的存在的附加标签来扩展AT管道。在Google Speech Commands V1（GSCV 1）和平衡的Audioset（AS）数据集上进行了实验。所提出的MobileNetV 2模型在GSCV 1数据集上实现了97.53%的准确率，在AS评估集上实现了33.4的mAP。此外，我们还表明，在真实世界的KWS数据集上可以观察到显著的噪声鲁棒性增益，大大优于标准KWS方法。我们的研究表明，KWS和AT可以合并成一个框架，而不会显著降低性能。摘要：Within the audio research community and the industry, keyword spotting (KWS) and audio tagging (AT) are seen as two distinct tasks and research fields. However, from a technical point of view, both of these tasks are identical: they predict a label (keyword in KWS, sound event in AT) for some fixed-sized input audio segment. This work proposes UniKW-AT: An initial approach for jointly training both KWS and AT. UniKW-AT enhances the noise-robustness for KWS, while also being able to predict specific sound events and enabling conditional wake-ups on sound events. Our approach extends the AT pipeline with additional labels describing the presence of a keyword. Experiments are conducted on the Google Speech Commands V1 (GSCV1) and the balanced Audioset (AS) datasets. The proposed MobileNetV2 model achieves an accuracy of 97.53% on the GSCV1 dataset and an mAP of 33.4 on the AS evaluation set. Further, we show that significant noise-robustness gains can be observed on a real-world KWS dataset, greatly outperforming standard KWS approaches. Our study shows that KWS and AT can be merged into a single framework without significant performance degradation.
【5】 The Microsoft System for VoxCeleb Speaker Recognition Challenge 2022
标题：微软VoxCeleb说话人识别挑战赛2022
链接：https://arxiv.org/abs/2209.11266
作者：Gang Liu,Tianyan Zhou,Yong Zhao,Yu Wu,Zhuo Chen,Yao Qian,Jian Wu机构：Microsoft Corporation, One Microsoft Way, Redmond, WA , USA备注：3 pages, 3 tables, VoxSRC2022摘要：在本报告中，我们描述了我们为2022年VoxCeleb说话人识别挑战赛（VoxSRC-22）第2轨道提交的系统。我们融合了各种性能良好的模型，从监督模型到自监督学习（SSL）预训练模型。使用VoxCeleb-2 dev数据训练的监督模型由ECAPA-TDNN和Res 2Net组成，具有非常深的结构。SSL预训练模型wav 2 vec和wavLM使用高达百万小时的大规模未标记语音数据进行训练。这些模型与ECAPA-TDNN级联，并以监督方式进一步微调，以提取说话人表示。所有13个模型都经过分数归一化和校准，然后融合到提交的系统中。我们还探讨了校准阶段的音频质量指标，如持续时间、SNR、T60和MOS。在VoxSRC-22评估集上，最佳提交系统的minDCF达到0.073，EER达到1.436%。摘要：In this report, we describe our submitted system for track 2 of the VoxCeleb Speaker Recognition Challenge 2022 (VoxSRC-22). We fuse a variety of good-performing models ranging from supervised models to self-supervised learning(SSL) pre-trained models. The supervised models, trained using VoxCeleb-2 dev data, consist of ECAPA-TDNN and Res2Net in a very deep structure. The SSL pre-trained models, wav2vec and wavLM, are trained using large scale unlabeled speech data up to million hours. These models are cascaded with ECAPA-TDNN and further fine-tuned in a supervised fashion to extract the speaker representations. All 13 models are applied with score normalization and calibration and then fused into the the submitted system. We also explore the audio quality measures in the calibration stage such as duration, SNR, T60, and MOS. The best submitted system achieves 0.073 in minDCF and 1.436% in EER on the VoxSRC-22 evaluation set.
【6】 The Kriston AI System for the VoxCeleb Speaker Recognition Challenge 2022
标题：VoxCeleb说话人识别挑战赛2022的Kriston AI系统
链接：https://arxiv.org/abs/2209.11433
作者：Qutang Cai,Guoqiang Hong,Zhijian Ye,Ximin Li,Haizhou Li机构：Kriston AI Lab, China, The Chinese University of Hong Kong, Shenzhen, China, Department of Electrical and Computer Engineering, National University of Singapore, Singapore备注：System description of VoxSRC 2022: track 1, 2 and 4摘要：本技术报告描述了我们的系统在2022年VoxCeleb说话人识别挑战赛（VoxSRC-22）中的第1、2和4轨道。通过组合几个ResNet变体，我们提交的轨道1获得了0：090的minDCF和1：401%的EER。通过进一步合并三个微调的预训练模型，我们提交的轨道2实现了0：072的minDCF和1：119%的EER。对于音轨4，我们的系统包括语音活动检测（VAD）、说话人嵌入提取、凝聚层次聚类（AHC）以及基于贝叶斯隐马尔可夫模型的重聚类步骤和重叠语音检测和处理。我们提交的轨道4实现了4.86%的日记错误率（DER）。所有参赛作品均在相应曲目中排名第二。摘要：This technical report describes our system for track 1, 2 and 4 of the VoxCeleb Speaker Recognition Challenge 2022 (VoxSRC-22). By combining several ResNet variants, our submission for track 1 attained a minDCF of 0:090 with EER 1:401%. By further incorporating three fine-tuned pre-trained models, our submission for track 2 achieved a minDCF of 0:072 with EER 1:119%. For track 4, our system consisted of voice activity detection (VAD), speaker embedding extraction, agglomerative hierarchical clustering (AHC) followed by a re-clustering step based on a Bayesian hidden Markov model and overlapped speech detection and handling. Our submission for track 4 achieved a diarisation error rate (DER) of 4.86%. The submissions all ranked the 2nd places for the corresponding tracks.
【1】 Stereo InSE-NET: Stereo Audio Quality Predictor Transfer Learned from Mono InSE-NET
标题：立体声Inse-Net：从单声道Inse-Net学习的立体声音质预测器转移
链接：https://arxiv.org/abs/2209.11666
作者：Arijit Biswas,Guanxin Jiang机构：Dolby Germany GmbH, N¨urnberg, Germany备注：Accepted to 153rd Audio Engineering Society (AES), New York, NY, USA, October 2022摘要：自动编码音频质量预测器通常被设计用于评估单个声道而不考虑任何空间方面。借助InSE-NET [1]，我们演示了如何使用深度神经网络（DNN）模拟最先进的编码音频质量指标（ViSQOL-v3 [2]），并随后对其进行改进-完全使用编程生成的数据。在本研究中，我们采取步骤来建立一个基于DNN的编码立体声音频质量预测器，我们提出了一个扩展的InSE-NET来处理立体声信号。该设计通过用左、右、中和侧声道调节模型来考虑立体声/空间方面;我们将模型命名为StereoInSE-NET。通过从预训练的单声道InSE-NET中转移选定的权重，并使用真实和综合增强听力测试进行再训练，我们证明了与最新的ViSQOL-v3 [3]相比，Pearson和Spearman Rank相关系数分别显著提高了12%和6%。摘要：Automatic coded audio quality predictors are typically designed for evaluating single channels without considering any spatial aspects. With InSE-NET [1], we demonstrated mimicking a state-of-the-art coded audio quality metric (ViSQOL-v3 [2]) with deep neural networks (DNN) and subsequently improving it - completely with programmatically generated data. In this study, we take steps towards building a DNN-based coded stereo audio quality predictor and we propose an extension of the InSE-NET for handling stereo signals. The design considers stereo/spatial aspects by conditioning the model with left, right, mid, and side channels; and we name our model Stereo InSE-NET. By transferring selected weights from the pre-trained mono InSE-NET and retraining with both real and synthetically augmented listening tests, we demonstrate a significant improvement of 12% and 6% of Pearson and Spearman Rank correlation coefficient, respectively, over the latest ViSQOL-v3 [3].
【2】 MMS-MSG: A Multi-purpose Multi-Speaker Mixture Signal Generator
标题：MMS-MSG：一种多功能多说话人混合信号发生器
链接：https://arxiv.org/abs/2209.11494
作者：Tobias Cord-Landwehr,Thilo von Neumann,Christoph Boeddeker,Reinhold Haeb-Umbach机构：Paderborn University, Department of Communications Engineering, Paderborn, Germany备注：Accepted at IWAENC 2022摘要：语音增强的范围已经从单一的、独立的任务的整体观点改变到复杂的会话语音记录的联合处理。这些单一任务的训练和评估需要合成数据，并访问尽可能接近评估场景的中间信号。由于这样的数据通常是不可用的，许多工作改为使用专门的数据库来训练每个系统组件，例如WSJ 0-mix用于源分离。我们提出了一种多用途多说话者混合信号发生器（MMS-MSG），用于基于任何语音语料库来生成各种语音混合信号，范围从经典的无回声混合（例如，WSJ 0-mix）优于混响混合（例如，SMS-WSJ）转换为会议式数据。其高度模块化和灵活的结构允许模拟不同的环境和动态混合，同时允许轻松扩展和修改，以生成新的场景和混合类型。这些会议可用于原型设计、评估或培训目的。我们提供了基于WSJ语料库的示例评估数据和会议基线结果。此外，我们通过使用MMS-MSG为LibriCSS数据库提供训练数据，展示了在现实场景中的有用性。摘要：The scope of speech enhancement has changed from a monolithic view of single, independent tasks, to a joint processing of complex conversational speech recordings. Training and evaluation of these single tasks requires synthetic data with access to intermediate signals that is as close as possible to the evaluation scenario. As such data often is not available, many works instead use specialized databases for the training of each system component, e.g WSJ0-mix for source separation. We present a Multi-purpose Multi-Speaker Mixture Signal Generator (MMS-MSG) for generating a variety of speech mixture signals based on any speech corpus, ranging from classical anechoic mixtures (e.g., WSJ0-mix) over reverberant mixtures (e.g., SMS-WSJ) to meeting-style data. Its highly modular and flexible structure allows for the simulation of diverse environments and dynamic mixing, while simultaneously enabling an easy extension and modification to generate new scenarios and mixture types. These meetings can be used for prototyping, evaluation, or training purposes. We provide example evaluation data and baseline results for meetings based on the WSJ corpus. Further, we demonstrate the usefulness for realistic scenarios by using MMS-MSG to provide training data for the LibriCSS database.
【3】 The Kriston AI System for the VoxCeleb Speaker Recognition Challenge 2022
标题：VoxCeleb说话人识别挑战赛2022的Kriston AI系统
链接：https://arxiv.org/abs/2209.11433
* 与cs.SD语音【6】为同一篇
作者：Qutang Cai,Guoqiang Hong,Zhijian Ye,Ximin Li,Haizhou Li机构：Kriston AI Lab, China, The Chinese University of Hong Kong, Shenzhen, China, Department of Electrical and Computer Engineering, National University of Singapore, Singapore备注：System description of VoxSRC 2022: track 1, 2 and 4摘要：本技术报告描述了我们的系统在2022年VoxCeleb说话人识别挑战赛（VoxSRC-22）中的第1、2和4轨道。通过组合几个ResNet变体，我们提交的轨道1获得了0：090的minDCF和1：401%的EER。通过进一步合并三个微调的预训练模型，我们提交的轨道2实现了0：072的minDCF和1：119%的EER。对于音轨4，我们的系统包括语音活动检测（VAD）、说话人嵌入提取、凝聚层次聚类（AHC）以及基于贝叶斯隐马尔可夫模型的重聚类步骤和重叠语音检测和处理。我们提交的轨道4实现了4.86%的日记错误率（DER）。所有参赛作品均在相应曲目中排名第二。摘要：This technical report describes our system for track 1, 2 and 4 of the VoxCeleb Speaker Recognition Challenge 2022 (VoxSRC-22). By combining several ResNet variants, our submission for track 1 attained a minDCF of 0:090 with EER 1:401%. By further incorporating three fine-tuned pre-trained models, our submission for track 2 achieved a minDCF of 0:072 with EER 1:119%. For track 4, our system consisted of voice activity detection (VAD), speaker embedding extraction, agglomerative hierarchical clustering (AHC) followed by a re-clustering step based on a Bayesian hidden Markov model and overlapped speech detection and handling. Our submission for track 4 achieved a diarisation error rate (DER) of 4.86%. The submissions all ranked the 2nd places for the corresponding tracks.
【4】 Isolation performance metrics for personal sound zone reproduction systems
标题：个人声区重现系统的隔离性能指标
链接：https://arxiv.org/abs/2209.11296
作者：Yue Qiao,Léo Guadagnin,Edgar Choueiri机构：),D Audio and Applied Acoustics Laboratory, Princeton University, Princeton, New Jersey, USA, )FOCAL-JMlab, La Talaudiere, France摘要：介绍了两种用于评价个人声区（PSZ）系统的隔离性能指标,,区间隔离（IZI）和程序间隔离（IPI）。与常用的声学对比度度量相比，IZI和IPI被推广到多声道音频，并分别量化了声音区域和音频节目的隔离度。这两个度量被示出为通常不可互换，并且适合于不同的场景，诸如生成暗区（IZI）或最小化音频对音频干扰（IPI）。最后给出了两个自由场仿真的例子，说明了IZI和IPI在评价PSZ在不同渲染模式下的性能和PSZ鲁棒性方面的应用。摘要：Two isolation performance metrics, Inter-Zone Isolation (IZI) and Inter-Program Isolation (IPI), are introduced for evaluating Personal Sound Zone (PSZ) systems. Compared to the commonly-used Acoustic Contrast metric, IZI and IPI are generalized for multichannel audio, and quantify the isolation of sound zones and of audio programs, respectively. The two metrics are shown to be generally non-interchangeable and suitable for different scenarios, such as generating dark zones (IZI) or minimizing audio-on-audio interference (IPI). Furthermore, two examples with free-field simulations are presented and demonstrate the applications of IZI and IPI in evaluating PSZ performance in different rendering modes and PSZ robustness.
【5】 The SpeakIn Speaker Verification System for Far-Field Speaker Verification Challenge 2022
标题：面向2022远场说话人确认挑战赛的说话人确认系统
链接：https://arxiv.org/abs/2209.11625
* 与cs.SD语音【1】为同一篇
作者：Yu Zheng,Jinghan Peng,Yihao Chen,Yajun Zhang,Jialong Wang,Min Liu,Minqiang Xu机构：SpeakIn Technologies Co. Ltd.备注：5 pages. arXiv admin note: text overlap with arXiv:2209.10846摘要：本文介绍了SpeakIn团队提交给2022年远场说话人确认挑战赛（FFSVC 2022）任务1和任务2的说话人确认（SV）系统。该挑战的SV任务集中于全监督远场说话人确认（任务1）和半监督远场说话人确认（任务2）问题。在任务1中，我们使用VoxCeleb和FFSVC 2020数据集作为训练数据集。对于任务2，我们仅使用VoxCeleb数据集作为训练集。基于ResNet和基于RepVGG的体系结构就是为应对这一挑战而开发的。采用全局统计池结构和MQMHA池结构对帧级特征进行跨时间聚合，得到话语级特征表示。我们采用AM-Softmax和AAM-Softmax对得到的嵌入进行分类。我们创新性地提出了一种分阶段迁移学习方法。在预训练阶段，我们保留了说话人的权值，在这一阶段没有正样本进行训练。然后，在第二阶段，我们使用正样本和负样本对这些权重进行微调。与传统的迁移学习策略相比，该策略能更好地提高模型的性能。采用Sub-Mean和AS-Norm后端方法来解决域不匹配问题。在融合阶段，Task 1融合了三个模型，Task 2融合了两个模型。在FFSVC 2022排行榜上，我们提交的EER为3.0049%，任务1中相应的minDCF为0.2938。在任务2中，EER和最小DCF分别为6.2060%和0.5232。我们的方法带来了出色的性能，在两项挑战任务中均名列第一。摘要：This paper describes speaker verification (SV) systems submitted by the SpeakIn team to the Task 1 and Task 2 of the Far-Field Speaker Verification Challenge 2022 (FFSVC2022). SV tasks of the challenge focus on the problem of fully supervised far-field speaker verification (Task 1) and semi-supervised far-field speaker verification (Task 2). In Task 1, we used the VoxCeleb and FFSVC2020 datasets as train datasets. And for Task 2, we only used the VoxCeleb dataset as train set. The ResNet-based and RepVGG-based architectures were developed for this challenge. Global statistic pooling structure and MQMHA pooling structure were used to aggregate the frame-level features across time to obtain utterance-level representation. We adopted AM-Softmax and AAM-Softmax to classify the resulting embeddings. We innovatively propose a staged transfer learning method. In the pre-training stage we reserve the speaker weights, and there are no positive samples to train them in this stage. Then we fine-tune these weights with both positive and negative samples in the second stage. Compared with the traditional transfer learning strategy, this strategy can better improve the model performance. The Sub-Mean and AS-Norm backend methods were used to solve the problem of domain mismatch. In the fusion stage, three models were fused in Task1 and two models were fused in Task2. On the FFSVC2022 leaderboard, the EER of our submission is 3.0049% and the corresponding minDCF is 0.2938 in Task1. In Task2, EER and minDCF are 6.2060% and 0.5232 respectively. Our approach leads to excellent performance and ranks 1st in both challenge tasks.
【6】 Synthetic Voice Spoofing Detection Based On Online Hard Example Mining
标题：基于在线硬例挖掘的合成语音欺骗检测
链接：https://arxiv.org/abs/2209.11585
* 与cs.SD语音【2】为同一篇
作者：Ruohua Zhou,Chenlei Hu,Qiuchen Yu,Yuxuan Du摘要：自动说话人确认欺骗（ASVspoof）挑战系列对于增强欺骗考虑和对抗措施的发展至关重要。虽然最新的ASVspoof 2019验证结果表明，该模型对大多数攻击具有显著的识别能力，但对部分攻击的识别效果仍然较差。提出了一种检测未知语音欺骗攻击的在线硬样本挖掘算法。OHEM用于克服数据集中简单样本和硬样本之间的不平衡。所提出的系统在ASVspoof 2019挑战逻辑访问场景的评估集上提供了0.77%的等错误率（EER）。摘要：The automatic speaker verification spoofing (ASVspoof) challenge series is crucial for enhancing the spoofing consideration and the countermeasures growth. Although the recent ASVspoof 2019 validation results indicate the significant capability to identify most attacks, the model's recognition effect is still poor for some attacks. This paper presents the Online Hard Example Mining (OHEM) algorithm for detecting unknown voice spoofing attacks. The OHEM is utilized to overcome the imbalance between simple and hard samples in the dataset. The presented system provides an equal error rate (EER) of 0.77% on the ASVspoof 2019 Challenge logical access scenario's evaluation set.
【7】 An artificial neural network-based system for detecting machine failures using tiny sound data: A case study
标题：基于人工神经网络的微小声音故障检测系统的实例研究
链接：https://arxiv.org/abs/2209.11527
* 与cs.SD语音【3】为同一篇
作者：Thanh Tran,Sebastian Bader,Jan Lundgren机构：∗ STC Research Centre, Department of Electronics Design, Mid Sweden University备注：8 pages, 9 figures, conference摘要：为了推动基于深度学习的机器故障检测系统的研究，我们给出了一个基于微小声音数据集的案例研究。我们的案例研究调查了一种变分自动编码器（VAE），用于增强来自Valmet AB的小型钻孔声音数据集。Valmet数据集包含134种声音，分为两类：“异常”和“正常”记录来自Valmet AB公司的钻孔机，该公司位于瑞典Sundsvall，为生物燃料生产提供设备和工艺。使用深度学习模型来检测如此小的声音数据集上的失败演练通常是不成功的。我们采用VAE通过从原始声音合成新声音来增加微小数据集中的声音数量。通过将这些合成声音与原始声音组合来创建扩充数据集。我们使用了通带频率为1000 Hz的高通滤波器和通带频率为22\kern0.16667em000 Hz的低通滤波器，在将声音转换为Mel声谱图之前，对增强数据集中的声音进行预处理。然后使用这些Mel谱图训练预训练的2D-CNN Alexnet。当与使用原始微小声音数据集来训练预训练的Alexnet相比时，使用增强的声音数据集将CNN模型的分类结果提高了6.62%（在增强的数据集上训练时为94.12%，而在原始数据集上训练时为87.5%）。摘要：In an effort to advocate the research for a deep learning-based machine failure detection system, we present a case study of our proposed system based on a tiny sound dataset. Our case study investigates a variational autoencoder (VAE) for augmenting a small drill sound dataset from Valmet AB. A Valmet dataset contains 134 sounds that have been divided into two categories: "Anomaly" and "Normal" recorded from a drilling machine in Valmet AB, a company in Sundsvall, Sweden that supplies equipment and processes for the production of biofuels. Using deep learning models to detect failure drills on such a small sound dataset is typically unsuccessful. We employed a VAE to increase the number of sounds in the tiny dataset by synthesizing new sounds from original sounds. The augmented dataset was created by combining these synthesized sounds with the original sounds. We used a high-pass filter with a passband frequency of 1000 Hz and a low-pass filter with a passband frequency of 22\kern 0.16667em000 Hz to pre-process sounds in the augmented dataset before transforming them to Mel spectrograms. The pre-trained 2D-CNN Alexnet was then trained using these Mel spectrograms. When compared to using the original tiny sound dataset to train pre-trained Alexnet, using the augmented sound dataset enhanced the CNN model's classification results by 6.62\%(94.12\% when trained on the augmented dataset versus 87.5\% when trained on the original dataset).
【8】 UniKW-AT: Unified Keyword Spotting and Audio Tagging
标题：UniKW-AT：统一的关键字检测和音频标注
链接：https://arxiv.org/abs/2209.11377
* 与cs.SD语音【4】为同一篇
作者：Heinrich Dinkel,Yongqing Wang,Zhiyong Yan,Junbo Zhang,Yujun Wang机构：Xiaomi Corperation, Beijing, China备注：Accepted in Interspeech2022摘要：在音频研究界和业界，关键词识别（KWS）和音频标注（AT）被视为两个不同的任务和研究领域。但是，从技术角度看，这两项任务是相同的：它们为一些固定大小输入音频段预测标签（KWS中的关键字，AT中的声音事件）。这项工作提出UniKW-AT：联合培训KWS和AT的初步方法。UniKW-AT增强了KWS的噪声鲁棒性，同时还能够预测特定的声音事件，并支持对声音事件的有条件唤醒。我们的方法用描述关键字的存在的附加标签来扩展AT管道。在Google Speech Commands V1（GSCV 1）和平衡的Audioset（AS）数据集上进行了实验。所提出的MobileNetV 2模型在GSCV 1数据集上实现了97.53%的准确率，在AS评估集上实现了33.4的mAP。此外，我们还表明，在真实世界的KWS数据集上可以观察到显著的噪声鲁棒性增益，大大优于标准KWS方法。我们的研究表明，KWS和AT可以合并成一个框架，而不会显著降低性能。摘要：Within the audio research community and the industry, keyword spotting (KWS) and audio tagging (AT) are seen as two distinct tasks and research fields. However, from a technical point of view, both of these tasks are identical: they predict a label (keyword in KWS, sound event in AT) for some fixed-sized input audio segment. This work proposes UniKW-AT: An initial approach for jointly training both KWS and AT. UniKW-AT enhances the noise-robustness for KWS, while also being able to predict specific sound events and enabling conditional wake-ups on sound events. Our approach extends the AT pipeline with additional labels describing the presence of a keyword. Experiments are conducted on the Google Speech Commands V1 (GSCV1) and the balanced Audioset (AS) datasets. The proposed MobileNetV2 model achieves an accuracy of 97.53% on the GSCV1 dataset and an mAP of 33.4 on the AS evaluation set. Further, we show that significant noise-robustness gains can be observed on a real-world KWS dataset, greatly outperforming standard KWS approaches. Our study shows that KWS and AT can be merged into a single framework without significant performance degradation.
机器翻译，仅供参考
每日语音速递