今日论文合集：cs.SD语音7篇，eess.AS音频处理9篇。
cs.SD语音

【1】 A Universal Identity Backdoor Attack against Speaker Verification based on Siamese Network标题：基于暹罗网络的针对说话人确认的通用身份后门攻击链接：http://arxiv.org/pdf/2303.16031v1

作者：Haodong Zhao,Wei Du,Junjie Guo,Gongshen Liu机构：School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai, China备注：Accepted by the Interspeech 2022. The first two authors contributed equally to this work摘要：说话人确认已被广泛应用于许多认证场景中。然而，说话人确认的训练模型需要大量的数据和计算能力，用户往往会使用不可信的第三方数据或直接部署第三方模型，这可能会带来安全隐患。在本文中，我们提出了一个后门攻击上述情况。具体来说，对于说话人验证系统中的连体网络，我们尝试在模型中植入一个通用的身份，可以模拟任何注册的说话人，并通过验证。因此攻击者不需要知道受害者，这使得攻击更加灵活和隐蔽。此外，我们设计并比较了三种选择攻击者话语的方式和两种方式的中毒训练的GE2E损失函数在不同的情况下。在TIMIT和Voxceleb1数据集上的实验结果表明，该方法在保证正常验证精度的同时，能够获得较高的攻击成功率。我们的工作揭示了说话人确认系统的脆弱性，为进一步提高系统的鲁棒性提供了新的视角。摘要：Speaker verification has been widely used in many authentication scenarios. However, training models for speaker verification requires large amounts of data and computing power, so users often use untrustworthy third-party data or deploy third-party models directly, which may create security risks. In this paper, we propose a backdoor attack for the above scenario. Specifically, for the Siamese network in the speaker verification system, we try to implant a universal identity in the model that can simulate any enrolled speaker and pass the verification. So the attacker does not need to know the victim, which makes the attack more flexible and stealthy. In addition, we design and compare three ways of selecting attacker utterances and two ways of poisoned training for the GE2E loss function in different scenarios. The results on the TIMIT and Voxceleb1 datasets show that our approach can achieve a high attack success rate while guaranteeing the normal verification accuracy. Our work reveals the vulnerability of the speaker verification system and provides a new perspective to further improve the robustness of the system.

【2】 Egocentric Auditory Attention Localization in Conversations标题：对话中以自我为中心的听觉注意定位链接：http://arxiv.org/pdf/2303.16024v1

作者：Fiona Ryan,Hao Jiang,Abhinav Shukla,James M. Rehg,Vamsi Krishna Ithapu机构：Georgia Institute of Technology, Meta Reality Labs Research摘要：在嘈杂的谈话环境中，例如晚宴，人们通常表现出选择性听觉注意力，或者专注于特定发言者而忽略其他人的能力。识别某人在对话中听谁说话对于开发能够理解社会行为的技术和能够通过放大特定声源来增强人类听力的设备至关重要。计算机视觉和音频研究社区在识别场景中的声源和扬声器方面取得了长足的进步。在这项工作中，我们采取了更进一步的专注于问题的本地化听觉注意力的目标，以自我为中心的视频，或检测谁在相机佩戴者的视野，他们正在听。为了解决新的和具有挑战性的选择性听觉注意力定位问题，我们提出了一种端到端的深度学习方法，该方法使用以自我为中心的视频和多通道音频来预测相机佩戴者的听觉注意力的热图。我们的方法利用时空视听功能和整体推理的场景进行预测，并优于一组具有挑战性的多说话人对话数据集的基线。项目页面：https: fkryan.github.io saal。摘要：In a noisy conversation environment such as a dinner party, people often exhibit selective auditory attention, or the ability to focus on a particular speaker while tuning out others. Recognizing who somebody is listening to in a conversation is essential for developing technologies that can understand social behavior and devices that can augment human hearing by amplifying particular sound sources. The computer vision and audio research communities have made great strides towards recognizing sound sources and speakers in scenes. In this work, we take a step further by focusing on the problem of localizing auditory attention targets in egocentric video, or detecting who in a camera wearer's field of view they are listening to. To tackle the new and challenging Selective Auditory Attention Localization problem, we propose an end-to-end deep learning approach that uses egocentric video and multichannel audio to predict the heatmap of the camera wearer's auditory attention. Our approach leverages spatiotemporal audiovisual features and holistic reasoning about the scene to make predictions, and outperforms a set of baselines on a challenging multi-speaker conversation dataset. Project page: https: fkryan.github.io saal.

【3】 Cluster-Guided Unsupervised Domain Adaptation for Deep Speaker Embedding标题：基于簇引导的无监督域自适应深度说话人嵌入链接：http://arxiv.org/pdf/2303.15944v1

作者：Haiquan Mao,Feng Hong,Man-wai Mak机构：TheHongKong Polytechnic University摘要：最近的研究表明，伪标签可以有助于无监督域自适应（UDA）的说话人确认。受使用现有分类器标记未标记数据进行再训练的自训练策略的启发，我们提出了一个聚类引导的UDA框架，该框架通过聚类标记目标域数据，并将标记的源域数据和伪标记的目标域数据相结合来训练说话人嵌入网络。为了提高聚类质量，我们训练一个专门用于聚类的说话人嵌入网络，通过最小化对比中心损失。其目标是减少嵌入和其分配的聚类中心之间的距离，同时扩大嵌入和其他聚类中心之间的距离。使用VoxCeleb 2作为源域和CN-Celeb 1作为目标域，我们证明了所提出的方法可以在CN-Celeb 1评估集上实现8.10%的等错误率（EER），而无需使用来自目标域的任何标签。该结果比监督基线高出39.6%，是该语料库上最先进的UDA性能。摘要：Recent studies have shown that pseudo labels can contribute to unsupervised domain adaptation (UDA) for speaker verification. Inspired by the self-training strategies that use an existing classifier to label the unlabeled data for retraining, we propose a cluster-guided UDA framework that labels the target domain data by clustering and combines the labeled source domain data and pseudo-labeled target domain data to train a speaker embedding network. To improve the cluster quality, we train a speaker embedding network dedicated for clustering by minimizing the contrastive center loss. The goal is to reduce the distance between an embedding and its assigned cluster center while enlarging the distance between the embedding and the other cluster centers. Using VoxCeleb2 as the source domain and CN-Celeb1 as the target domain, we demonstrate that the proposed method can achieve an equal error rate (EER) of 8.10% on the CN-Celeb1 evaluation set without using any labels from the target domain. This result outperforms the supervised baseline by 39.6% and is the state-of-the-art UDA performance on this corpus.

【4】 TransAudio: Towards the Transferable Adversarial Audio Attack via Learning Contextualized Perturbations标题：TransAudio：通过学习上下文扰动实现可转移的对抗性音频攻击链接：http://arxiv.org/pdf/2303.15940v1

作者：Qi Gege,Yuefeng Chen,Xiaofeng Mao,Yao Zhu,Binyuan Hui,Xiaodan Li,Rong Zhang,Hui Xue机构：Alibaba Group, Hangzhou, China ,Zhejiang University, Hangzhou, China摘要：在针对自动语音识别（ASR）系统的基于传输的攻击中，攻击无法访问目标模型的架构和参数。现有的攻击方法主要在具有受限语音命令的语音助理场景中进行研究，从而禁止其适用于更一般的ASR相关应用。为了解决这一挑战，我们提出了一种新的上下文攻击删除，插入和替换对抗行为，即TransAudio，实现任意字级攻击的基础上提出的两阶段框架。为了增强攻击的可移植性，我们进一步引入了一个音频分数匹配优化策略来规范训练过程，从而减少了敌对样本对代理模型的过拟合。大量的实验和分析证明了TransAudio对开源ASR模型和商业API的有效性。摘要：In a transfer-based attack against Automatic Speech Recognition (ASR) systems, attacks are unable to access the architecture and parameters of the target model. Existing attack methods are mostly investigated in voice assistant scenarios with restricted voice commands, prohibiting their applicability to more general ASR related applications. To tackle this challenge, we propose a novel contextualized attack with deletion, insertion, and substitution adversarial behaviors, namely TransAudio, which achieves arbitrary word-level attacks based on the proposed two-stage framework. To strengthen the attack transferability, we further introduce an audio score-matching optimization strategy to regularize the training process, which mitigates adversarial example over-fitting to the surrogate model. Extensive experiments and analysis demonstrate the effectiveness of TransAudio against open-source ASR models and commercial APIs.

【5】 Adaptive Background Music for a Fighting Game: A Multi-Instrument Volume Modulation Approach标题：格斗游戏的自适应背景音乐：一种多乐器音量调制方法链接：http://arxiv.org/pdf/2303.15734v1

作者：Ibrahim Khan,Thai Van Nguyen,Chollakorn Nimpattanavong,Ruck Thawonmas机构：Graduate School of Information Science and Engineering, Ritsumeikan University, Kusatsu, Shiga, Japan, College of Information Science and Engineering备注：This paper under review is made available for participants of DareFightingICE Competition (this https URL) and readers interested in relevant areas摘要：本文介绍了我们通过添加自适应背景音乐来增强DareFightingICE中背景音乐（BGM）的工作。自适应BGM由五种不同的乐器组成，演奏一首名为“G弦上的空气”的古典音乐作品。“BGM通过改变乐器的音量进行调整。每个乐器都连接到游戏的不同元素。然后，我们通过使用仅使用音频作为输入的深度强化学习AI（盲DL AI）来运行实验以评估自适应BGM。结果表明，与没有自适应BGM的播放相比，在播放自适应BGM时，盲DL AI的性能得到改善。摘要：This paper presents our work to enhance the background music (BGM) in DareFightingICE by adding an adaptive BGM. The adaptive BGM consists of five different instruments playing a classical music piece called "Air on G-String." The BGM adapts by changing the volume of the instruments. Each instrument is connected to a different element of the game. We then run experiments to evaluate the adaptive BGM by using a deep reinforcement learning AI that only uses audio as input (Blind DL AI). The results show that the performance of the Blind DL AI improves while playing with the adaptive BGM as compared to playing without the adaptive BGM.

【6】 Translate the Beauty in Songs: Jointly Learning to Align Melody and Translate Lyrics标题：翻译歌曲之美：共同学习对齐旋律和翻译歌词链接：http://arxiv.org/pdf/2303.15705v1

作者：Chengxi Li,Kai Fan,Jiajun Bu,Boxing Chen,Zhongqiang Huang,Zhi Yu机构：Zhejiang University, Alibaba DAMO Academy备注：13 pages摘要：歌曲翻译既需要翻译歌词，也需要调整音乐音符，以便最终的诗句可以唱出伴随的旋律，这是一个具有挑战性的问题，在翻译过程的不同方面引起了一些兴趣。在本文中，我们提出了歌词旋律翻译与自适应分组（LTAG），一个整体的解决方案，自动歌曲翻译，共同建模歌词翻译和歌词旋律对齐。它是一种新颖的编码器-解码器框架，可以同时翻译的源歌词，并确定对齐的注意，在每个解码步骤通过自适应注意分组模块的数量。为了解决数据稀缺问题，我们委托了少量专门为此任务注释的训练数据，并通过反向翻译使用了大量增强数据。在一个英汉歌曲翻译数据集上进行的实验表明了该模型在自动和人工评价中的有效性。摘要：Song translation requires both translation of lyrics and alignment of music notes so that the resulting verse can be sung to the accompanying melody, which is a challenging problem that has attracted some interests in different aspects of the translation process. In this paper, we propose Lyrics-Melody Translation with Adaptive Grouping (LTAG), a holistic solution to automatic song translation by jointly modeling lyrics translation and lyrics-melody alignment. It is a novel encoder-decoder framework that can simultaneously translate the source lyrics and determine the number of aligned notes at each decoding step through an adaptive note grouping module. To address data scarcity, we commissioned a small amount of training data annotated specifically for this task and used large amounts of augmented data through back-translation. Experiments conducted on an English-Chinese song translation data set show the effectiveness of our model in both automatic and human evaluation.

【7】 Spatial Active Noise Control Method Based On Sound Field Interpolation From Reference Microphone SignalsEnvironmental Noise标题：基于参考麦克风信号声场内插的空间有源噪声控制方法链接：http://arxiv.org/pdf/2303.16021v1

作者：Kazuyuki Arikawa,Shoichi Koyama,Hiroshi Saruwatari机构：The University of Tokyo,-,-, Hongo, Bunkyo-ku, Tokyo ,-, Japan备注：Accepted to International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2023摘要：提出了一种基于参考传声器信号声场插值的空间有源噪声控制方法。在大多数当前的空间ANC方法中，需要足够数量的误差麦克风来减少目标区域上的噪声，因为声场是从误差麦克风信号估计的。然而，在实际应用中，优选地，误差麦克风的数量尽可能小，以在目标区域中为ANC用户保留空间。我们建议从参考麦克风，这是通常放置在目标区域之外，而不是误差麦克风的声场插值。我们推导出一个固定的过滤器的基础上的核岭回归声场插值空间降噪。此外，为了补偿估计误差，我们将所提出的固定滤波器与基于使用误差麦克风信号的控制滤波器的过渡的多通道ANC相结合。数值实验结果表明，区域噪声可以充分降低所提出的方法，即使在错误的麦克风的数量是特别小。摘要：A spatial active noise control (ANC) method based on the interpolation of a sound field from reference microphone signals is proposed. In most current spatial ANC methods, a sufficient number of error microphones are required to reduce noise over the target region because the sound field is estimated from error microphone signals. However, in practical applications, it is preferable that the number of error microphones is as small as possible to keep a space in the target region for ANC users. We propose to interpolate the sound field from reference microphones, which are normally placed outside the target region, instead of the error microphones. We derive a fixed filter for spatial noise reduction on the basis of the kernel ridge regression for sound field interpolation. Furthermore, to compensate for estimation errors, we combine the proposed fixed filter with multichannel ANC based on a transition of the control filter using the error microphone signals. Numerical experimental results indicate that regional noise can be sufficiently reduced by the proposed methods even when the number of error microphones is particularly small.
eess.AS音频处理

【1】 Spatial Active Noise Control Method Based On Sound Field Interpolation From Reference Microphone SignalsEnvironmental Noise标题：基于参考麦克风信号声场内插的空间有源噪声控制方法链接：http://arxiv.org/pdf/2303.16021v1

*与cs.SD语音【7】为同一篇

作者：Kazuyuki Arikawa,Shoichi Koyama,Hiroshi Saruwatari机构：The University of Tokyo,-,-, Hongo, Bunkyo-ku, Tokyo ,-, Japan备注：Accepted to International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2023摘要：提出了一种基于参考传声器信号声场插值的空间有源噪声控制方法。在大多数当前的空间ANC方法中，需要足够数量的误差麦克风来减少目标区域上的噪声，因为声场是从误差麦克风信号估计的。然而，在实际应用中，优选地，误差麦克风的数量尽可能小，以在目标区域中为ANC用户保留空间。我们建议从参考麦克风，这是通常放置在目标区域之外，而不是误差麦克风的声场插值。我们推导出一个固定的过滤器的基础上的核岭回归声场插值空间降噪。此外，为了补偿估计误差，我们将所提出的固定滤波器与基于使用误差麦克风信号的控制滤波器的过渡的多通道ANC相结合。数值实验结果表明，区域噪声可以充分降低所提出的方法，即使在错误的麦克风的数量是特别小。摘要：A spatial active noise control (ANC) method based on the interpolation of a sound field from reference microphone signals is proposed. In most current spatial ANC methods, a sufficient number of error microphones are required to reduce noise over the target region because the sound field is estimated from error microphone signals. However, in practical applications, it is preferable that the number of error microphones is as small as possible to keep a space in the target region for ANC users. We propose to interpolate the sound field from reference microphones, which are normally placed outside the target region, instead of the error microphones. We derive a fixed filter for spatial noise reduction on the basis of the kernel ridge regression for sound field interpolation. Furthermore, to compensate for estimation errors, we combine the proposed fixed filter with multichannel ANC based on a transition of the control filter using the error microphone signals. Numerical experimental results indicate that regional noise can be sufficiently reduced by the proposed methods even when the number of error microphones is particularly small.

【2】 AD-YOLO: You Look ONly Once in Training Multiple Sound Event Localization and Detection标题：Ad-YOLO：在训练多个声音事件定位和检测中只看一次链接：http://arxiv.org/pdf/2303.15703v1

作者：Jin Sob Kim,Hyun Joon Park,Wooseok Shin,Sung Won Han机构：School of Industrial and Management Engineering, Korea University, Seoul, Republic of Korea备注：5 pages, 3 figures, accepted for publication in IEEE ICASSP 2023摘要：声事件定位与检测（SELD）将声事件的识别与相应的到达方向（DOA）相结合。最近，已经采用面向事件的轨迹输出格式来解决这个问题;然而，在未知的复调环境中，它们对现实世界问题的概括仍然有限。为了解决这个问题，我们提出了一个基于角距离的多SELD（AD-YOLO），这是一个适应的“你只看一次”算法SELD。AD-YOLO格式允许模型通过将类责任分配给DOA预测来学习声音发生位置敏感性。因此，该格式使模型能够处理复调问题，而不管声音重叠的数量。我们使用四个SELD客观指标在DCASE 2020-2022挑战任务3数据集上评估了AD-YOLO。实验结果表明，AD-YOLO取得了出色的整体性能，也完成了类同质复调环境中的鲁棒性。摘要：Sound event localization and detection (SELD) combines the identification of sound events with the corresponding directions of arrival (DOA). Recently, event-oriented track output formats have been adopted to solve this problem; however, they still have limited generalization toward real-world problems in an unknown polyphony environment. To address the issue, we proposed an angular-distance-based multiple SELD (AD-YOLO), which is an adaptation of the "You Look Only Once" algorithm for SELD. The AD-YOLO format allows the model to learn sound occurrences location-sensitively by assigning class responsibility to DOA predictions. Hence, the format enables the model to handle the polyphony problem, regardless of the number of sound overlaps. We evaluated AD-YOLO on DCASE 2020-2022 challenge Task 3 datasets using four SELD objective metrics. The experimental results show that AD-YOLO achieved outstanding performance overall and also accomplished robustness in class-homogeneous polyphony environments.

【3】 Unsupervised Pre-Training For Data-Efficient Text-to-Speech On Low Resource Languages标题：低资源语言数据高效文语转换的无监督预训练链接：http://arxiv.org/pdf/2303.15669v1

作者：Seongyeon Park,Myungseo Song,Bohyung Kim,Tae-Hyun Oh机构：CNAI, Seoul, Korea, Dept. of EE and GSAI, POSTECH, Pohang, Korea, Institute for Convergence Research and Education in Advanced Technology, Yonsei University, Seoul, Korea备注：ICASSP 2023摘要：None摘要：Neural text-to-speech (TTS) models can synthesize natural human speech when trained on large amounts of transcribed speech. However, collecting such large-scale transcribed data is expensive. This paper proposes an unsupervised pre-training method for a sequence-to-sequence TTS model by leveraging large untranscribed speech data. With our pre-training, we can remarkably reduce the amount of paired transcribed data required to train the model for the target downstream TTS task. The main idea is to pre-train the model to reconstruct de-warped mel-spectrograms from warped ones, which may allow the model to learn proper temporal assignment relation between input and output sequences. In addition, we propose a data augmentation method that further improves the data efficiency in fine-tuning. We empirically demonstrate the effectiveness of our proposed method in low-resource language scenarios, achieving outstanding performance compared to competing methods. The code and audio samples are available at: https: github.com cnaigithub SpeechDewarping

【4】 A Universal Identity Backdoor Attack against Speaker Verification based on Siamese Network标题：基于暹罗网络的针对说话人确认的通用身份后门攻击链接：http://arxiv.org/pdf/2303.16031v1

*与cs.SD语音【1】为同一篇

作者：Haodong Zhao,Wei Du,Junjie Guo,Gongshen Liu机构：School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai, China备注：Accepted by the Interspeech 2022. The first two authors contributed equally to this work摘要：说话人确认已被广泛应用于许多认证场景中。然而，说话人确认的训练模型需要大量的数据和计算能力，用户往往会使用不可信的第三方数据或直接部署第三方模型，这可能会带来安全隐患。在本文中，我们提出了一个后门攻击上述情况。具体来说，对于说话人验证系统中的连体网络，我们尝试在模型中植入一个通用的身份，可以模拟任何注册的说话人，并通过验证。因此攻击者不需要知道受害者，这使得攻击更加灵活和隐蔽。此外，我们设计并比较了三种选择攻击者话语的方式和两种方式的中毒训练的GE2E损失函数在不同的情况下。在TIMIT和Voxceleb1数据集上的实验结果表明，该方法在保证正常验证精度的同时，能够获得较高的攻击成功率。我们的工作揭示了说话人确认系统的脆弱性，为进一步提高系统的鲁棒性提供了新的视角。摘要：Speaker verification has been widely used in many authentication scenarios. However, training models for speaker verification requires large amounts of data and computing power, so users often use untrustworthy third-party data or deploy third-party models directly, which may create security risks. In this paper, we propose a backdoor attack for the above scenario. Specifically, for the Siamese network in the speaker verification system, we try to implant a universal identity in the model that can simulate any enrolled speaker and pass the verification. So the attacker does not need to know the victim, which makes the attack more flexible and stealthy. In addition, we design and compare three ways of selecting attacker utterances and two ways of poisoned training for the GE2E loss function in different scenarios. The results on the TIMIT and Voxceleb1 datasets show that our approach can achieve a high attack success rate while guaranteeing the normal verification accuracy. Our work reveals the vulnerability of the speaker verification system and provides a new perspective to further improve the robustness of the system.

【5】 Egocentric Auditory Attention Localization in Conversations标题：对话中以自我为中心的听觉注意定位链接：http://arxiv.org/pdf/2303.16024v1

*与cs.SD语音【2】为同一篇

作者：Fiona Ryan,Hao Jiang,Abhinav Shukla,James M. Rehg,Vamsi Krishna Ithapu机构：Georgia Institute of Technology, Meta Reality Labs Research摘要：在嘈杂的谈话环境中，例如晚宴，人们通常表现出选择性听觉注意力，或者专注于特定发言者而忽略其他人的能力。识别某人在对话中听谁说话对于开发能够理解社会行为的技术和能够通过放大特定声源来增强人类听力的设备至关重要。计算机视觉和音频研究社区在识别场景中的声源和扬声器方面取得了长足的进步。在这项工作中，我们采取了更进一步的专注于问题的本地化听觉注意力的目标，以自我为中心的视频，或检测谁在相机佩戴者的视野，他们正在听。为了解决新的和具有挑战性的选择性听觉注意力定位问题，我们提出了一种端到端的深度学习方法，该方法使用以自我为中心的视频和多通道音频来预测相机佩戴者的听觉注意力的热图。我们的方法利用时空视听功能和整体推理的场景进行预测，并优于一组具有挑战性的多说话人对话数据集的基线。项目页面：https: fkryan.github.io saal。摘要：In a noisy conversation environment such as a dinner party, people often exhibit selective auditory attention, or the ability to focus on a particular speaker while tuning out others. Recognizing who somebody is listening to in a conversation is essential for developing technologies that can understand social behavior and devices that can augment human hearing by amplifying particular sound sources. The computer vision and audio research communities have made great strides towards recognizing sound sources and speakers in scenes. In this work, we take a step further by focusing on the problem of localizing auditory attention targets in egocentric video, or detecting who in a camera wearer's field of view they are listening to. To tackle the new and challenging Selective Auditory Attention Localization problem, we propose an end-to-end deep learning approach that uses egocentric video and multichannel audio to predict the heatmap of the camera wearer's auditory attention. Our approach leverages spatiotemporal audiovisual features and holistic reasoning about the scene to make predictions, and outperforms a set of baselines on a challenging multi-speaker conversation dataset. Project page: https: fkryan.github.io saal.

【6】 Cluster-Guided Unsupervised Domain Adaptation for Deep Speaker Embedding标题：基于簇引导的无监督域自适应深度说话人嵌入链接：http://arxiv.org/pdf/2303.15944v1

*与cs.SD语音【3】为同一篇

作者：Haiquan Mao,Feng Hong,Man-wai Mak机构：TheHongKong Polytechnic University摘要：最近的研究表明，伪标签可以有助于无监督域自适应（UDA）的说话人确认。受使用现有分类器标记未标记数据进行再训练的自训练策略的启发，我们提出了一个聚类引导的UDA框架，该框架通过聚类标记目标域数据，并将标记的源域数据和伪标记的目标域数据相结合来训练说话人嵌入网络。为了提高聚类质量，我们训练一个专门用于聚类的说话人嵌入网络，通过最小化对比中心损失。其目标是减少嵌入和其分配的聚类中心之间的距离，同时扩大嵌入和其他聚类中心之间的距离。使用VoxCeleb 2作为源域和CN-Celeb 1作为目标域，我们证明了所提出的方法可以在CN-Celeb 1评估集上实现8.10%的等错误率（EER），而无需使用来自目标域的任何标签。该结果比监督基线高出39.6%，是该语料库上最先进的UDA性能。摘要：Recent studies have shown that pseudo labels can contribute to unsupervised domain adaptation (UDA) for speaker verification. Inspired by the self-training strategies that use an existing classifier to label the unlabeled data for retraining, we propose a cluster-guided UDA framework that labels the target domain data by clustering and combines the labeled source domain data and pseudo-labeled target domain data to train a speaker embedding network. To improve the cluster quality, we train a speaker embedding network dedicated for clustering by minimizing the contrastive center loss. The goal is to reduce the distance between an embedding and its assigned cluster center while enlarging the distance between the embedding and the other cluster centers. Using VoxCeleb2 as the source domain and CN-Celeb1 as the target domain, we demonstrate that the proposed method can achieve an equal error rate (EER) of 8.10% on the CN-Celeb1 evaluation set without using any labels from the target domain. This result outperforms the supervised baseline by 39.6% and is the state-of-the-art UDA performance on this corpus.

【7】 TransAudio: Towards the Transferable Adversarial Audio Attack via Learning Contextualized Perturbations标题：TransAudio：通过学习上下文扰动实现可转移的对抗性音频攻击链接：http://arxiv.org/pdf/2303.15940v1

*与cs.SD语音【4】为同一篇

作者：Qi Gege,Yuefeng Chen,Xiaofeng Mao,Yao Zhu,Binyuan Hui,Xiaodan Li,Rong Zhang,Hui Xue机构：Alibaba Group, Hangzhou, China ,Zhejiang University, Hangzhou, China摘要：在针对自动语音识别（ASR）系统的基于传输的攻击中，攻击无法访问目标模型的架构和参数。现有的攻击方法主要在具有受限语音命令的语音助理场景中进行研究，从而禁止其适用于更一般的ASR相关应用。为了解决这一挑战，我们提出了一种新的上下文攻击删除，插入和替换对抗行为，即TransAudio，实现任意字级攻击的基础上提出的两阶段框架。为了增强攻击的可移植性，我们进一步引入了一个音频分数匹配优化策略来规范训练过程，从而减少了敌对样本对代理模型的过拟合。大量的实验和分析证明了TransAudio对开源ASR模型和商业API的有效性。摘要：In a transfer-based attack against Automatic Speech Recognition (ASR) systems, attacks are unable to access the architecture and parameters of the target model. Existing attack methods are mostly investigated in voice assistant scenarios with restricted voice commands, prohibiting their applicability to more general ASR related applications. To tackle this challenge, we propose a novel contextualized attack with deletion, insertion, and substitution adversarial behaviors, namely TransAudio, which achieves arbitrary word-level attacks based on the proposed two-stage framework. To strengthen the attack transferability, we further introduce an audio score-matching optimization strategy to regularize the training process, which mitigates adversarial example over-fitting to the surrogate model. Extensive experiments and analysis demonstrate the effectiveness of TransAudio against open-source ASR models and commercial APIs.

【8】 Adaptive Background Music for a Fighting Game: A Multi-Instrument Volume Modulation Approach标题：格斗游戏的自适应背景音乐：一种多乐器音量调制方法链接：http://arxiv.org/pdf/2303.15734v1

*与cs.SD语音【5】为同一篇

作者：Ibrahim Khan,Thai Van Nguyen,Chollakorn Nimpattanavong,Ruck Thawonmas机构：Graduate School of Information Science and Engineering, Ritsumeikan University, Kusatsu, Shiga, Japan, College of Information Science and Engineering备注：This paper under review is made available for participants of DareFightingICE Competition (this https URL) and readers interested in relevant areas摘要：本文介绍了我们通过添加自适应背景音乐来增强DareFightingICE中背景音乐（BGM）的工作。自适应BGM由五种不同的乐器组成，演奏一首名为“G弦上的空气”的古典音乐作品。“BGM通过改变乐器的音量进行调整。每个乐器都连接到游戏的不同元素。然后，我们通过使用仅使用音频作为输入的深度强化学习AI（盲DL AI）来运行实验以评估自适应BGM。结果表明，与没有自适应BGM的播放相比，在播放自适应BGM时，盲DL AI的性能得到改善。摘要：This paper presents our work to enhance the background music (BGM) in DareFightingICE by adding an adaptive BGM. The adaptive BGM consists of five different instruments playing a classical music piece called "Air on G-String." The BGM adapts by changing the volume of the instruments. Each instrument is connected to a different element of the game. We then run experiments to evaluate the adaptive BGM by using a deep reinforcement learning AI that only uses audio as input (Blind DL AI). The results show that the performance of the Blind DL AI improves while playing with the adaptive BGM as compared to playing without the adaptive BGM.

【9】 Translate the Beauty in Songs: Jointly Learning to Align Melody and Translate Lyrics标题：翻译歌曲之美：共同学习对齐旋律和翻译歌词链接：http://arxiv.org/pdf/2303.15705v1

*与cs.SD语音【6】为同一篇

作者：Chengxi Li,Kai Fan,Jiajun Bu,Boxing Chen,Zhongqiang Huang,Zhi Yu机构：Zhejiang University, Alibaba DAMO Academy备注：13 pages摘要：歌曲翻译既需要翻译歌词，也需要调整音乐音符，以便最终的诗句可以唱出伴随的旋律，这是一个具有挑战性的问题，在翻译过程的不同方面引起了一些兴趣。在本文中，我们提出了歌词旋律翻译与自适应分组（LTAG），一个整体的解决方案，自动歌曲翻译，共同建模歌词翻译和歌词旋律对齐。它是一种新颖的编码器-解码器框架，可以同时翻译的源歌词，并确定对齐的注意，在每个解码步骤通过自适应注意分组模块的数量。为了解决数据稀缺问题，我们委托了少量专门为此任务注释的训练数据，并通过反向翻译使用了大量增强数据。在一个英汉歌曲翻译数据集上进行的实验表明了该模型在自动和人工评价中的有效性。摘要：Song translation requires both translation of lyrics and alignment of music notes so that the resulting verse can be sung to the accompanying melody, which is a challenging problem that has attracted some interests in different aspects of the translation process. In this paper, we propose Lyrics-Melody Translation with Adaptive Grouping (LTAG), a holistic solution to automatic song translation by jointly modeling lyrics translation and lyrics-melody alignment. It is a novel encoder-decoder framework that can simultaneously translate the source lyrics and determine the number of aligned notes at each decoding step through an adaptive note grouping module. To address data scarcity, we commissioned a small amount of training data annotated specifically for this task and used large amounts of augmented data through back-translation. Experiments conducted on an English-Chinese song translation data set show the effectiveness of our model in both automatic and human evaluation.

机器翻译，仅供参考