今日论文合集：cs.SD语音2篇，eess.AS音频处理3篇。
cs.SD语音

【1】 On the Impact of Voice Anonymization on Speech-Based COVID-19 Detection标题：语音匿名化对基于语音的新冠肺炎检测的影响链接：http://arxiv.org/pdf/2304.02181v1

作者：Yi Zhu,Mohamed Imoussaïne-Aïkous,Carolyn Côté-Lussier,Tiago H. Falk机构：University of Qu´ebec备注：11 pages, 10 figures摘要：随着深度学习的进步，基于语音的应用正在蓬勃发展，从个人助理，情感计算到远程疾病诊断。由于语音包含语言和副语言信息（例如，音调、语调、语速、响度），对语音匿名化以保护说话者隐私和身份的兴趣日益增长。语音隐私的挑战已经出现在过去的几年里，重点已经放在删除扬声器的身份，同时保持语言内容完好无损。然而，对于情感计算和疾病监测应用，副语言内容可能更关键。不幸的是，匿名化可能对这些系统产生的影响在很大程度上仍然是未知的。在本文中，我们填补了这一空白，并专注于一个特定的健康监测应用程序：基于语音的COVID-19诊断我们使用三个公共数据集测试了两种流行的匿名化方法及其对五种不同的最先进的COVID-19诊断系统的影响。我们验证了匿名化方法的有效性，比较了它们的计算复杂性，并量化了不同测试场景对数据集内和跨数据集条件的影响。最后，我们展示了匿名化作为数据增强工具的好处，以帮助恢复匿名数据所导致的COVID-19诊断准确性损失。摘要：With advances seen in deep learning, voice-based applications are burgeoning, ranging from personal assistants, affective computing, to remote disease diagnostics. As the voice contains both linguistic and paralinguistic information (e.g., vocal pitch, intonation, speech rate, loudness), there is growing interest in voice anonymization to preserve speaker privacy and identity. Voice privacy challenges have emerged over the last few years and focus has been placed on removing speaker identity while keeping linguistic content intact. For affective computing and disease monitoring applications, however, the paralinguistic content may be more critical. Unfortunately, the effects that anonymization may have on these systems are still largely unknown. In this paper, we fill this gap and focus on one particular health monitoring application: speech-based COVID-19 diagnosis. We test two popular anonymization methods and their impact on five different state-of-the-art COVID-19 diagnostic systems using three public datasets. We validate the effectiveness of the anonymization methods, compare their computational complexity, and quantify the impact across different testing scenarios for both within- and across-dataset conditions. Lastly, we show the benefits of anonymization as a data augmentation tool to help recover some of the COVID-19 diagnostic accuracy loss seen with anonymized data.

【2】 Pac-HuBERT: Self-Supervised Music Source Separation via Primitive Auditory Clustering and Hidden-Unit BERT标题：PAC-Hubert：基于原始听觉聚类和隐元BERT的自监督音乐源分离链接：http://arxiv.org/pdf/2304.02160v1

作者：Ke Chen,Gordon Wichern,François G. Germain,Jonathan Le Roux机构：Mitsubishi Electric Research Laboratories (MERL), Cambridge, MA, USA, �University of California San Diego (UCSD), La Jolla, CA, USA备注：5 pages, 2 figures, 3 tables摘要：尽管音乐源分离研究取得了进展，但少量公开可用的干净源数据仍然是性能的恒定限制因素。因此，自我监督学习的最新进展为通过利用未标记的音乐数据来改进分离模型提供了一个尚未探索的机会。在本文中，我们提出了一个自我监督的学习框架，音乐源分离的启发HuBERT语音表示模型。我们首先调查的潜在影响，原来的HuBERT模型插入到著名的Demucs V2时域分离模型架构的适应版本。然后，我们提出了一个时间-频率域的自监督模型，Pac-HuBERT（原始听觉聚类HuBERT），我们后来使用的Res-U-Net解码器的源分离相结合。Pac-HuBERT使用音乐的原始听觉特征作为无监督聚类标签，使用Free Music Archive（FMA）数据集初始化自监督预训练过程。由此产生的框架在MusDB 18测试集上实现了比原始Demucs V2和Res-U-Net模型更好的源失真比（SDR）性能。我们进一步证明了它可以用少量的监督数据提高性能。最终，我们提出的框架是一个有效的解决方案，有限的清洁源数据的音乐源分离的挑战。摘要：In spite of the progress in music source separation research, the small amount of publicly-available clean source data remains a constant limiting factor for performance. Thus, recent advances in self-supervised learning present a largely-unexplored opportunity for improving separation models by leveraging unlabelled music data. In this paper, we propose a self-supervised learning framework for music source separation inspired by the HuBERT speech representation model. We first investigate the potential impact of the original HuBERT model by inserting an adapted version of it into the well-known Demucs V2 time-domain separation model architecture. We then propose a time-frequency-domain self-supervised model, Pac-HuBERT (for primitive auditory clustering HuBERT), that we later use in combination with a Res-U-Net decoder for source separation. Pac-HuBERT uses primitive auditory features of music as unsupervised clustering labels to initialize the self-supervised pretraining process using the Free Music Archive (FMA) dataset. The resulting framework achieves better source-to-distortion ratio (SDR) performance on the MusDB18 test set than the original Demucs V2 and Res-U-Net models. We further demonstrate that it can boost performance with small amounts of supervised data. Ultimately, our proposed framework is an effective solution to the challenge of limited clean source data for music source separation.
eess.AS音频处理

【1】 Zero-shot domain adaptation of anomalous samples for semi-supervised anomaly detection标题：用于半监督异常检测的异常样本零射频域自适应链接：http://arxiv.org/pdf/2304.02221v1

作者：Tomoya Nishida,Takashi Endo,Yohei Kawaguchi机构：Research and Development Group, Hitachi, Ltd., -, Higashi-koigakubo, Kokubunji-shi, Tokyo ,-, Japan摘要：半监督异常检测（SSAD）是一个任务，其中正常数据和有限数量的异常数据可用于训练。在实际情况中，SSAD方法遭受适应域移位，因为异常数据不太可能在训练阶段中可用于目标域。为了解决这个问题，我们提出了一个域自适应方法SSAD没有异常数据可用于目标域。首先，我们引入一个域对抗网络的变分自动编码器为基础的SSAD模型，以获得域不变的潜变量。由于解码器不能仅从域不变潜变量重建原始数据，因此我们将解码器限制在域标签上。为了补偿目标域的缺失异常数据，我们引入了一个基于重要性采样的加权损失函数，它近似于理想的损失函数。实验结果表明，当目标域没有异常数据时，该方法能够使SSAD模型适应目标域。摘要：Semi-supervised anomaly detection~(SSAD) is a task where normal data and a limited number of anomalous data are available for training. In practical situations, SSAD methods suffer adapting to domain shifts, since anomalous data are unlikely to be available for the target domain in the training phase. To solve this problem, we propose a domain adaptation method for SSAD where no anomalous data are available for the target domain. First, we introduce a domain-adversarial network to a variational auto-encoder-based SSAD model to obtain domain-invariant latent variables. Since the decoder cannot reconstruct the original data solely from domain-invariant latent variables, we conditioned the decoder on the domain label. To compensate for the missing anomalous data of the target domain, we introduce an importance sampling-based weighted loss function that approximates the ideal loss function. Experimental results indicate that the proposed method helps adapt SSAD models to the target domain when no anomalous data are available for the target domain.

【2】 On the Impact of Voice Anonymization on Speech-Based COVID-19 Detection标题：语音匿名化对基于语音的新冠肺炎检测的影响链接：http://arxiv.org/pdf/2304.02181v1

*与cs.SD语音【1】为同一篇

作者：Yi Zhu,Mohamed Imoussaïne-Aïkous,Carolyn Côté-Lussier,Tiago H. Falk机构：University of Qu´ebec备注：11 pages, 10 figures摘要：随着深度学习的进步，基于语音的应用正在蓬勃发展，从个人助理，情感计算到远程疾病诊断。由于语音包含语言和副语言信息（例如，音调、语调、语速、响度），对语音匿名化以保护说话者隐私和身份的兴趣日益增长。语音隐私的挑战已经出现在过去的几年里，重点已经放在删除扬声器的身份，同时保持语言内容完好无损。然而，对于情感计算和疾病监测应用，副语言内容可能更关键。不幸的是，匿名化可能对这些系统产生的影响在很大程度上仍然是未知的。在本文中，我们填补了这一空白，并专注于一个特定的健康监测应用程序：基于语音的COVID-19诊断我们使用三个公共数据集测试了两种流行的匿名化方法及其对五种不同的最先进的COVID-19诊断系统的影响。我们验证了匿名化方法的有效性，比较了它们的计算复杂性，并量化了不同测试场景对数据集内和跨数据集条件的影响。最后，我们展示了匿名化作为数据增强工具的好处，以帮助恢复匿名数据所导致的COVID-19诊断准确性损失。摘要：With advances seen in deep learning, voice-based applications are burgeoning, ranging from personal assistants, affective computing, to remote disease diagnostics. As the voice contains both linguistic and paralinguistic information (e.g., vocal pitch, intonation, speech rate, loudness), there is growing interest in voice anonymization to preserve speaker privacy and identity. Voice privacy challenges have emerged over the last few years and focus has been placed on removing speaker identity while keeping linguistic content intact. For affective computing and disease monitoring applications, however, the paralinguistic content may be more critical. Unfortunately, the effects that anonymization may have on these systems are still largely unknown. In this paper, we fill this gap and focus on one particular health monitoring application: speech-based COVID-19 diagnosis. We test two popular anonymization methods and their impact on five different state-of-the-art COVID-19 diagnostic systems using three public datasets. We validate the effectiveness of the anonymization methods, compare their computational complexity, and quantify the impact across different testing scenarios for both within- and across-dataset conditions. Lastly, we show the benefits of anonymization as a data augmentation tool to help recover some of the COVID-19 diagnostic accuracy loss seen with anonymized data.

【3】 Pac-HuBERT: Self-Supervised Music Source Separation via Primitive Auditory Clustering and Hidden-Unit BERT标题：PAC-Hubert：基于原始听觉聚类和隐元BERT的自监督音乐源分离链接：http://arxiv.org/pdf/2304.02160v1

*与cs.SD语音【2】为同一篇

作者：Ke Chen,Gordon Wichern,François G. Germain,Jonathan Le Roux机构：Mitsubishi Electric Research Laboratories (MERL), Cambridge, MA, USA, �University of California San Diego (UCSD), La Jolla, CA, USA备注：5 pages, 2 figures, 3 tables摘要：尽管音乐源分离研究取得了进展，但少量公开可用的干净源数据仍然是性能的恒定限制因素。因此，自我监督学习的最新进展为通过利用未标记的音乐数据来改进分离模型提供了一个尚未探索的机会。在本文中，我们提出了一个自我监督的学习框架，音乐源分离的启发HuBERT语音表示模型。我们首先调查的潜在影响，原来的HuBERT模型插入到著名的Demucs V2时域分离模型架构的适应版本。然后，我们提出了一个时间-频率域的自监督模型，Pac-HuBERT（原始听觉聚类HuBERT），我们后来使用的Res-U-Net解码器的源分离相结合。Pac-HuBERT使用音乐的原始听觉特征作为无监督聚类标签，使用Free Music Archive（FMA）数据集初始化自监督预训练过程。由此产生的框架在MusDB 18测试集上实现了比原始Demucs V2和Res-U-Net模型更好的源失真比（SDR）性能。我们进一步证明了它可以用少量的监督数据提高性能。最终，我们提出的框架是一个有效的解决方案，有限的清洁源数据的音乐源分离的挑战。摘要：In spite of the progress in music source separation research, the small amount of publicly-available clean source data remains a constant limiting factor for performance. Thus, recent advances in self-supervised learning present a largely-unexplored opportunity for improving separation models by leveraging unlabelled music data. In this paper, we propose a self-supervised learning framework for music source separation inspired by the HuBERT speech representation model. We first investigate the potential impact of the original HuBERT model by inserting an adapted version of it into the well-known Demucs V2 time-domain separation model architecture. We then propose a time-frequency-domain self-supervised model, Pac-HuBERT (for primitive auditory clustering HuBERT), that we later use in combination with a Res-U-Net decoder for source separation. Pac-HuBERT uses primitive auditory features of music as unsupervised clustering labels to initialize the self-supervised pretraining process using the Free Music Archive (FMA) dataset. The resulting framework achieves better source-to-distortion ratio (SDR) performance on the MusDB18 test set than the original Demucs V2 and Res-U-Net models. We further demonstrate that it can boost performance with small amounts of supervised data. Ultimately, our proposed framework is an effective solution to the challenge of limited clean source data for music source separation.

机器翻译，仅供参考