今日论文合集：cs.SD语音6篇，eess.AS音频处理7篇。
cs.SD语音

【1】 MusicFace: Music-driven Expressive Singing Face Synthesis标题：MusicFace：音乐驱动的表现力唱面合成链接：http://arxiv.org/pdf/2303.14044v1

作者：Pengfei Liu,Wenjin Deng,Hengda Li,Jintai Wang,Yinglin Zheng,Yiwei Ding,Xiaohu Guo,Ming Zeng机构：Zeng,(�), © The Author(s)备注：Accepted to CVMJ摘要：如何在音乐信号的驱动下合成一个生动逼真的歌唱人脸仍然是一个有趣而富有挑战性的问题。在本文中，我们提出了一种方法，这项任务与自然运动的嘴唇，面部表情，头部姿势，眼睛的状态。针对音乐音频信号中人声和背景音乐混合信息的耦合性，设计了一种解耦融合策略。我们首先将输入的音乐音频分解为人声流和背景音乐流。由于两个流的输入信号和面部表情，头部运动和眼睛状态的动态之间的隐式和复杂的相关性，我们建模与注意力方案，其中两个流的效果无缝融合它们的关系。此外，为了提高所产生的结果的表现力，我们建议分解的头部运动生成为速度生成和方向生成，并分解的眼睛状态生成为短时间眨眼生成和长时间闭眼生成分别建模。我们还建立了一个新的SingingFace数据集，以支持这项任务的训练和评估，并促进未来的工作。大量的实验和用户研究表明，我们提出的方法是能够合成生动的歌唱脸，这是优于国家的最先进的方法定性和定量。摘要：It is still an interesting and challenging problem to synthesize a vivid and realistic singing face driven by music signal. In this paper, we present a method for this task with natural motions of the lip, facial expression, head pose, and eye states. Due to the coupling of the mixed information of human voice and background music in common signals of music audio, we design a decouple-and-fuse strategy to tackle the challenge. We first decompose the input music audio into human voice stream and background music stream. Due to the implicit and complicated correlation between the two-stream input signals and the dynamics of the facial expressions, head motions and eye states, we model their relationship with an attention scheme, where the effects of the two streams are fused seamlessly. Furthermore, to improve the expressiveness of the generated results, we propose to decompose head movements generation into speed generation and direction generation, and decompose eye states generation into the short-time eye blinking generation and the long-time eye closing generation to model them separately. We also build a novel SingingFace Dataset to support the training and evaluation of this task, and to facilitate future works on this topic. Extensive experiments and user study show that our proposed method is capable of synthesizing vivid singing face, which is better than state-of-the-art methods qualitatively and quantitatively.

【2】 Overview of the ICASSP 2023 General Meeting Understanding and Generation Challenge (MUG)标题：ICASSP 2023大会理解与生成挑战(MUG)综述链接：http://arxiv.org/pdf/2303.13932v1

作者：Qinglin Zhang,Chong Deng,Jiaqing Liu,Hai Yu,Qian Chen,Wen Wang,Zhijie Yan,Jinglin Liu,Yi Ren,Zhou Zhao机构：Speech Lab of DAMO Academy, Alibaba Group , Zhejiang University备注：Paper accepted to the 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2023), Rhodes, Greece摘要：ICASSP2023 General Meeting Understanding and Generation Challenge（MUG）专注于促进对会议记录的广泛口语处理（SLP）研究，因为SLP应用对于提高用户在会议中掌握重要信息的效率至关重要。MUG包括五个轨道，包括主题分割，主题级和会话级提取摘要，主题标题生成，关键词提取和动作项检测。为了促进MUG，我们构建并发布了一个大规模的会议数据集，AliMeeting4MUG语料库。摘要：ICASSP2023 General Meeting Understanding and Generation Challenge (MUG) focuses on prompting a wide range of spoken language processing (SLP) research on meeting transcripts, as SLP applications are critical to improve users' efficiency in grasping important information in meetings. MUG includes five tracks, including topic segmentation, topic-level and session-level extractive summarization, topic title generation, keyphrase extraction, and action item detection. To facilitate MUG, we construct and release a large-scale meeting dataset, the AliMeeting4MUG Corpus.

【3】 Wave-U-Net Discriminator: Fast and Lightweight Discriminator for Generative Adversarial Network-Based Speech Synthesis标题：Wave-U-net鉴别器：基于生成性对抗性网络语音合成的快速轻量级鉴别器链接：http://arxiv.org/pdf/2303.13909v1

作者：Takuhiro Kaneko,Hirokazu Kameoka,Kou Tanaka,Shogo Seki机构：NTT Communication Science Laboratories, NTT Corporation, Japan备注：Accepted to ICASSP 2023. Project page: this https URL摘要：在语音合成中，生成式对抗网络（GAN）通过最小最大博弈来训练生成器（语音合成器）和鉴别器，被广泛用于提高语音质量。鉴别器的集合通常用于最近的神经声码器（例如，HiFi-GAN）和端到端文本到语音（TTS）系统（例如，VITS）以从多个视角仔细检查波形。这样的鉴别器允许合成语音充分地接近真实语音;然而，根据鉴别器数量的增加，它们需要增加模型大小和计算时间。或者，本研究提出了一种Wave-U-Net鉴别器，这是一种具有Wave-U-Net架构的单一但富有表现力的鉴别器。此鉴别器是唯一的;它可以以与输入信号相同的分辨率以逐样本的方式评估波形，同时通过具有跳跃连接的编码器和解码器来提取多级特征。这种结构为生成器提供了足够丰富的信息，以使合成语音与真实语音紧密匹配。在实验中，所提出的想法被应用到一个代表性的神经声码器（HiFi-GAN）和一个端到端的TTS系统（VITS）。结果表明，所提出的模型可以实现相当的语音质量与2.31倍的速度和14.5倍的轻量级鉴别器时，用于HiFi-GAN和1.90倍的速度和9.62倍的轻量级鉴别器时，用于VITS。音频样本可从www.example.com获得https: www.kecl.ntt.co.jp people kaneko.takuhiro projects waveunetd 。摘要：In speech synthesis, a generative adversarial network (GAN), training a generator (speech synthesizer) and a discriminator in a min-max game, is widely used to improve speech quality. An ensemble of discriminators is commonly used in recent neural vocoders (e.g., HiFi-GAN) and end-to-end text-to-speech (TTS) systems (e.g., VITS) to scrutinize waveforms from multiple perspectives. Such discriminators allow synthesized speech to adequately approach real speech; however, they require an increase in the model size and computation time according to the increase in the number of discriminators. Alternatively, this study proposes a Wave-U-Net discriminator, which is a single but expressive discriminator with Wave-U-Net architecture. This discriminator is unique; it can assess a waveform in a sample-wise manner with the same resolution as the input signal, while extracting multilevel features via an encoder and decoder with skip connections. This architecture provides a generator with sufficiently rich information for the synthesized speech to be closely matched to the real speech. During the experiments, the proposed ideas were applied to a representative neural vocoder (HiFi-GAN) and an end-to-end TTS system (VITS). The results demonstrate that the proposed models can achieve comparable speech quality with a 2.31 times faster and 14.5 times more lightweight discriminator when used in HiFi-GAN and a 1.90 times faster and 9.62 times more lightweight discriminator when used in VITS. Audio samples are available at https: www.kecl.ntt.co.jp people kaneko.takuhiro projects waveunetd .

【4】 Symbolic Music Structure Analysis with Graph Representations and Changepoint Detection Methods标题：基于图形表示和变点检测的符号音乐结构分析链接：http://arxiv.org/pdf/2303.13881v1

作者：Carlos Hernandez-Olivan,Sonia Rubio Llamas,Jose R. Beltran机构：Department of ElectronicEngineering and Communications, University of Zaragoza摘要：音乐结构分析是音乐信息检索中的一个开放性研究课题。在过去，已经有一些作品，试图分割音乐到音频和符号域，然而，在不同层次的音乐结构的识别和分割仍然是一个开放的研究问题，在这一领域。在这项工作中，我们提出了三种方法，其中两种是新颖的基于图的算法，旨在通过其形式或结构来分割符号音乐：Norm、G-PELT和G-Window。我们用两个具有不同形式或结构的公共数据集进行了消融研究，以便比较这些方法，改变它们的参数值并比较针对不同音乐风格的性能。我们发现，用图表示的符号音乐编码和计算从图中获得的邻接矩阵的新颖性很好地表示了符号音乐作品的结构，而不需要从中提取特征。我们能够用在线无监督的变化点检测方法检测边界，F_1为0。在我们用于测试我们的方法的一个公共数据集中，对于1 bar的公差为5640。我们还提供了在不同层次的结构，高，中，低的算法的性能结果，以显示所提出的方法的参数必须根据水平进行调整。我们在musicaiz（一个开源python包）中添加了性能最好的方法及其每个结构级别的参数，以促进这项工作的可重复性和可用性。我们希望这些方法可以用来改进其他的MIR任务，如有结构的音乐生成、音乐分类或基调变化检测。摘要：Music Structure Analysis is an open research task in Music Information Retrieval (MIR). In the past, there have been several works that attempt to segment music into the audio and symbolic domains, however, the identification and segmentation of the music structure at different levels is still an open research problem in this area. In this work we propose three methods, two of which are novel graph-based algorithms that aim to segment symbolic music by its form or structure: Norm, G-PELT and G-Window. We performed an ablation study with two public datasets that have different forms or structures in order to compare such methods varying their parameter values and comparing the performance against different music styles. We have found that encoding symbolic music with graph representations and computing the novelty of Adjacency Matrices obtained from graphs represent the structure of symbolic music pieces well without the need to extract features from it. We are able to detect the boundaries with an online unsupervised changepoint detection method with a F_1 of 0.5640 for a 1 bar tolerance in one of the public datasets that we used for testing our methods. We also provide the performance results of the algorithms at different levels of structure, high, medium and low, to show how the parameters of the proposed methods have to be adjusted depending on the level. We added the best performing method with its parameters for each structure level to musicaiz, an open source python package, to facilitate the reproducibility and usability of this work. We hope that this methods could be used to improve other MIR tasks such as music generation with structure, music classification or key changes detection.

【5】 In-depth analysis of music structure as a self-organized network标题：音乐结构作为一个自组织网络的深入分析链接：http://arxiv.org/pdf/2303.13631v1

作者：Ping-Rui Tsai,Yen-Ting Chou,Nathan-Christopher Wang,Hui-Ling Chen,Hong-Yue Huang,Zih-Jia Luo,Tzay-Ming Hong机构：Department of Physics, National Tsing Hua University, Hsinchu , Taiwan, R.O.C, College of Pharmacy, University of Michigan, Ann Arbor, MI , U.S.A., Department of Chinese Literature, National Tsing Hua University, Hsinchu , Taiwan, R.O.C and备注：5 pages摘要：自然语言中的词语不仅传递信息，而且随着文明的发展和人类的迁徙而演变。音乐也是如此。为了理解音乐背后的复杂结构，我们引入了一种名为Essential Element Network（EEN）的算法，将音频编码为文本。该网络通过计算尺度、时间和体积之间的相关性来获得。优化EEN生成的频率和排名的聚类系数的Zipfs定律，使我们能够产生和视为词的语义关系。我们将这些编码词映射到尺度-时间空间，这有助于我们系统地组织音乐深层结构中的句法。我们的算法提供了对音乐背后复杂网络的精确描述，而不是其他深度学习方法的黑盒性质。因此，通过这些过程积累的经验和属性不仅可以为自然语言处理（NLP）的应用提供一种新的方法，而且可以更容易，更客观地分析音乐的演变和发展。摘要：Words in a natural language not only transmit information but also evolve with the development of civilization and human migration. The same is true for music. To understand the complex structure behind the music, we introduced an algorithm called the Essential Element Network (EEN) to encode the audio into text. The network is obtained by calculating the correlations between scales, time, and volume. Optimizing EEN to generate Zipfs law for the frequency and rank of the clustering coefficient enables us to generate and regard the semantic relationships as words. We map these encoded words into the scale-temporal space, which helps us organize systematically the syntax in the deep structure of music. Our algorithm provides precise descriptions of the complex network behind the music, as opposed to the black-box nature of other deep learning approaches. As a result, the experience and properties accumulated through these processes can offer not only a new approach to the applications of Natural Language Processing (NLP) but also an easier and more objective way to analyze the evolution and development of music.

【6】 Enhancing Unsupervised Speech Recognition with Diffusion GANs标题：利用扩散GANS增强无监督语音识别链接：http://arxiv.org/pdf/2303.13559v1

作者：Xianchao Wu机构：NVIDIA备注：5 pages, 1 figure, accepted by ICASSP 2023摘要：我们通过扩散GAN增强了无监督自动语音识别（ASR）的香草对抗训练方法。我们的模型（1）将各种强度的实例噪声注入到生成器的输出和未标记的参考文本中，这些参考文本是从具有长度约束的预训练音素语言模型中采样的，（2）要求扩散时间步相关的鉴别器将它们分开，以及（3）反向传播梯度以更新生成器。单词 音素错误率比较与wav 2 vec-U Librisepeech（3.1%的测试清洁和5.6%的测试其他），TIMIT和MLS数据集，表明我们的增强策略有效地工作。摘要：We enhance the vanilla adversarial training method for unsupervised Automatic Speech Recognition (ASR) by a diffusion-GAN. Our model (1) injects instance noises of various intensities to the generator's output and unlabeled reference text which are sampled from pretrained phoneme language models with a length constraint, (2) asks diffusion timestep-dependent discriminators to separate them, and (3) back-propagates the gradients to update the generator. Word phoneme error rate comparisons with wav2vec-U under Librispeech (3.1% for test-clean and 5.6% for test-other), TIMIT and MLS datasets, show that our enhancement strategies work effectively.
eess.AS音频处理

【1】 Attention-based Speech Enhancement Using Human Quality Perception Modelling标题：基于人类质量感知模型的注意力语音增强链接：http://arxiv.org/pdf/2303.13685v1

作者：Khandokar Md. Nayem,Donald S. Williamson机构：Williamson was with the Department of Computer Science atIndiana University, but he is now with the Department of Computer Scienceand Engineering备注：11 pages, 4 figures, 3 tables, submitted in journal TASLP 2023摘要：感知启发的目标函数，例如语音质量的感知评估（PESQ）、信号失真比（SDR）和短时客观可懂度（STOI），最近已被用于优化基于深度学习的语音增强算法的性能。然而，这些目标函数并不总是与收听者对感知质量的评估强烈相关，因此利用这些措施进行优化通常会导致在现实世界场景中的较差性能。在这项工作中，我们提出了一种基于注意力的增强方法，使用学习语音嵌入向量的平均意见得分（MOS）预测模型和语音增强模块，以共同提高嘈杂的语音。MOS预测模型直接从音频信号估计由人类收听者评估的语音质量的感知MOS。增强模块还采用量化语言模型，该量化语言模型强制执行频谱约束以获得更好的语音真实性和性能。我们使用在日常环境中捕获的真实世界的噪声语音数据来训练模型，并使用看不见的语料库来测试它。结果表明，我们提出的方法显着优于其他方法，优化与客观的措施，其中预测的质量分数与人类的判断密切相关。摘要：Perceptually-inspired objective functions such as the perceptual evaluation of speech quality (PESQ), signal-to-distortion ratio (SDR), and short-time objective intelligibility (STOI), have recently been used to optimize performance of deep-learning-based speech enhancement algorithms. These objective functions, however, do not always strongly correlate with a listener's assessment of perceptual quality, so optimizing with these measures often results in poorer performance in real-world scenarios. In this work, we propose an attention-based enhancement approach that uses learned speech embedding vectors from a mean-opinion score (MOS) prediction model and a speech enhancement module to jointly enhance noisy speech. The MOS prediction model estimates the perceptual MOS of speech quality, as assessed by human listeners, directly from the audio signal. The enhancement module also employs a quantized language model that enforces spectral constraints for better speech realism and performance. We train the model using real-world noisy speech data that has been captured in everyday environments and test it using unseen corpora. The results show that our proposed approach significantly outperforms other approaches that are optimized with objective measures, where the predicted quality scores strongly correlate with human judgments.

【2】 MusicFace: Music-driven Expressive Singing Face Synthesis标题：MusicFace：音乐驱动的表现力唱面合成链接：http://arxiv.org/pdf/2303.14044v1

*与cs.SD语音【1】为同一篇

作者：Pengfei Liu,Wenjin Deng,Hengda Li,Jintai Wang,Yinglin Zheng,Yiwei Ding,Xiaohu Guo,Ming Zeng机构：Zeng,(�), © The Author(s)备注：Accepted to CVMJ摘要：如何在音乐信号的驱动下合成一个生动逼真的歌唱人脸仍然是一个有趣而富有挑战性的问题。在本文中，我们提出了一种方法，这项任务与自然运动的嘴唇，面部表情，头部姿势，眼睛的状态。针对音乐音频信号中人声和背景音乐混合信息的耦合性，设计了一种解耦融合策略。我们首先将输入的音乐音频分解为人声流和背景音乐流。由于两个流的输入信号和面部表情，头部运动和眼睛状态的动态之间的隐式和复杂的相关性，我们建模与注意力方案，其中两个流的效果无缝融合它们的关系。此外，为了提高所产生的结果的表现力，我们建议分解的头部运动生成为速度生成和方向生成，并分解的眼睛状态生成为短时间眨眼生成和长时间闭眼生成分别建模。我们还建立了一个新的SingingFace数据集，以支持这项任务的训练和评估，并促进未来的工作。大量的实验和用户研究表明，我们提出的方法是能够合成生动的歌唱脸，这是优于国家的最先进的方法定性和定量。摘要：It is still an interesting and challenging problem to synthesize a vivid and realistic singing face driven by music signal. In this paper, we present a method for this task with natural motions of the lip, facial expression, head pose, and eye states. Due to the coupling of the mixed information of human voice and background music in common signals of music audio, we design a decouple-and-fuse strategy to tackle the challenge. We first decompose the input music audio into human voice stream and background music stream. Due to the implicit and complicated correlation between the two-stream input signals and the dynamics of the facial expressions, head motions and eye states, we model their relationship with an attention scheme, where the effects of the two streams are fused seamlessly. Furthermore, to improve the expressiveness of the generated results, we propose to decompose head movements generation into speed generation and direction generation, and decompose eye states generation into the short-time eye blinking generation and the long-time eye closing generation to model them separately. We also build a novel SingingFace Dataset to support the training and evaluation of this task, and to facilitate future works on this topic. Extensive experiments and user study show that our proposed method is capable of synthesizing vivid singing face, which is better than state-of-the-art methods qualitatively and quantitatively.

【3】 Overview of the ICASSP 2023 General Meeting Understanding and Generation Challenge (MUG)标题：ICASSP 2023大会理解与生成挑战(MUG)综述链接：http://arxiv.org/pdf/2303.13932v1

*与cs.SD语音【2】为同一篇

作者：Qinglin Zhang,Chong Deng,Jiaqing Liu,Hai Yu,Qian Chen,Wen Wang,Zhijie Yan,Jinglin Liu,Yi Ren,Zhou Zhao机构：Speech Lab of DAMO Academy, Alibaba Group , Zhejiang University备注：Paper accepted to the 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2023), Rhodes, Greece摘要：ICASSP2023 General Meeting Understanding and Generation Challenge（MUG）专注于促进对会议记录的广泛口语处理（SLP）研究，因为SLP应用对于提高用户在会议中掌握重要信息的效率至关重要。MUG包括五个轨道，包括主题分割，主题级和会话级提取摘要，主题标题生成，关键词提取和动作项检测。为了促进MUG，我们构建并发布了一个大规模的会议数据集，AliMeeting4MUG语料库。摘要：ICASSP2023 General Meeting Understanding and Generation Challenge (MUG) focuses on prompting a wide range of spoken language processing (SLP) research on meeting transcripts, as SLP applications are critical to improve users' efficiency in grasping important information in meetings. MUG includes five tracks, including topic segmentation, topic-level and session-level extractive summarization, topic title generation, keyphrase extraction, and action item detection. To facilitate MUG, we construct and release a large-scale meeting dataset, the AliMeeting4MUG Corpus.

【4】 Wave-U-Net Discriminator: Fast and Lightweight Discriminator for Generative Adversarial Network-Based Speech Synthesis标题：Wave-U-net鉴别器：基于生成性对抗性网络语音合成的快速轻量级鉴别器链接：http://arxiv.org/pdf/2303.13909v1

*与cs.SD语音【3】为同一篇

作者：Takuhiro Kaneko,Hirokazu Kameoka,Kou Tanaka,Shogo Seki机构：NTT Communication Science Laboratories, NTT Corporation, Japan备注：Accepted to ICASSP 2023. Project page: this https URL摘要：在语音合成中，生成式对抗网络（GAN）通过最小最大博弈来训练生成器（语音合成器）和鉴别器，被广泛用于提高语音质量。鉴别器的集合通常用于最近的神经声码器（例如，HiFi-GAN）和端到端文本到语音（TTS）系统（例如，VITS）以从多个视角仔细检查波形。这样的鉴别器允许合成语音充分地接近真实语音;然而，根据鉴别器数量的增加，它们需要增加模型大小和计算时间。或者，本研究提出了一种Wave-U-Net鉴别器，这是一种具有Wave-U-Net架构的单一但富有表现力的鉴别器。此鉴别器是唯一的;它可以以与输入信号相同的分辨率以逐样本的方式评估波形，同时通过具有跳跃连接的编码器和解码器来提取多级特征。这种结构为生成器提供了足够丰富的信息，以使合成语音与真实语音紧密匹配。在实验中，所提出的想法被应用到一个代表性的神经声码器（HiFi-GAN）和一个端到端的TTS系统（VITS）。结果表明，所提出的模型可以实现相当的语音质量与2.31倍的速度和14.5倍的轻量级鉴别器时，用于HiFi-GAN和1.90倍的速度和9.62倍的轻量级鉴别器时，用于VITS。音频样本可从www.example.com获得https: www.kecl.ntt.co.jp people kaneko.takuhiro projects waveunetd 。摘要：In speech synthesis, a generative adversarial network (GAN), training a generator (speech synthesizer) and a discriminator in a min-max game, is widely used to improve speech quality. An ensemble of discriminators is commonly used in recent neural vocoders (e.g., HiFi-GAN) and end-to-end text-to-speech (TTS) systems (e.g., VITS) to scrutinize waveforms from multiple perspectives. Such discriminators allow synthesized speech to adequately approach real speech; however, they require an increase in the model size and computation time according to the increase in the number of discriminators. Alternatively, this study proposes a Wave-U-Net discriminator, which is a single but expressive discriminator with Wave-U-Net architecture. This discriminator is unique; it can assess a waveform in a sample-wise manner with the same resolution as the input signal, while extracting multilevel features via an encoder and decoder with skip connections. This architecture provides a generator with sufficiently rich information for the synthesized speech to be closely matched to the real speech. During the experiments, the proposed ideas were applied to a representative neural vocoder (HiFi-GAN) and an end-to-end TTS system (VITS). The results demonstrate that the proposed models can achieve comparable speech quality with a 2.31 times faster and 14.5 times more lightweight discriminator when used in HiFi-GAN and a 1.90 times faster and 9.62 times more lightweight discriminator when used in VITS. Audio samples are available at https: www.kecl.ntt.co.jp people kaneko.takuhiro projects waveunetd .

【5】 Symbolic Music Structure Analysis with Graph Representations and Changepoint Detection Methods标题：基于图形表示和变点检测的符号音乐结构分析链接：http://arxiv.org/pdf/2303.13881v1

*与cs.SD语音【4】为同一篇

作者：Carlos Hernandez-Olivan,Sonia Rubio Llamas,Jose R. Beltran机构：Department of ElectronicEngineering and Communications, University of Zaragoza摘要：音乐结构分析是音乐信息检索中的一个开放性研究课题。在过去，已经有一些作品，试图分割音乐到音频和符号域，然而，在不同层次的音乐结构的识别和分割仍然是一个开放的研究问题，在这一领域。在这项工作中，我们提出了三种方法，其中两种是新颖的基于图的算法，旨在通过其形式或结构来分割符号音乐：Norm、G-PELT和G-Window。我们用两个具有不同形式或结构的公共数据集进行了消融研究，以便比较这些方法，改变它们的参数值并比较针对不同音乐风格的性能。我们发现，用图表示的符号音乐编码和计算从图中获得的邻接矩阵的新颖性很好地表示了符号音乐作品的结构，而不需要从中提取特征。我们能够用在线无监督的变化点检测方法检测边界，F_1为0。在我们用于测试我们的方法的一个公共数据集中，对于1 bar的公差为5640。我们还提供了在不同层次的结构，高，中，低的算法的性能结果，以显示所提出的方法的参数必须根据水平进行调整。我们在musicaiz（一个开源python包）中添加了性能最好的方法及其每个结构级别的参数，以促进这项工作的可重复性和可用性。我们希望这些方法可以用来改进其他的MIR任务，如有结构的音乐生成、音乐分类或基调变化检测。摘要：Music Structure Analysis is an open research task in Music Information Retrieval (MIR). In the past, there have been several works that attempt to segment music into the audio and symbolic domains, however, the identification and segmentation of the music structure at different levels is still an open research problem in this area. In this work we propose three methods, two of which are novel graph-based algorithms that aim to segment symbolic music by its form or structure: Norm, G-PELT and G-Window. We performed an ablation study with two public datasets that have different forms or structures in order to compare such methods varying their parameter values and comparing the performance against different music styles. We have found that encoding symbolic music with graph representations and computing the novelty of Adjacency Matrices obtained from graphs represent the structure of symbolic music pieces well without the need to extract features from it. We are able to detect the boundaries with an online unsupervised changepoint detection method with a F_1 of 0.5640 for a 1 bar tolerance in one of the public datasets that we used for testing our methods. We also provide the performance results of the algorithms at different levels of structure, high, medium and low, to show how the parameters of the proposed methods have to be adjusted depending on the level. We added the best performing method with its parameters for each structure level to musicaiz, an open source python package, to facilitate the reproducibility and usability of this work. We hope that this methods could be used to improve other MIR tasks such as music generation with structure, music classification or key changes detection.

【6】 In-depth analysis of music structure as a self-organized network标题：音乐结构作为一个自组织网络的深入分析链接：http://arxiv.org/pdf/2303.13631v1

*与cs.SD语音【5】为同一篇

作者：Ping-Rui Tsai,Yen-Ting Chou,Nathan-Christopher Wang,Hui-Ling Chen,Hong-Yue Huang,Zih-Jia Luo,Tzay-Ming Hong机构：Department of Physics, National Tsing Hua University, Hsinchu , Taiwan, R.O.C, College of Pharmacy, University of Michigan, Ann Arbor, MI , U.S.A., Department of Chinese Literature, National Tsing Hua University, Hsinchu , Taiwan, R.O.C and备注：5 pages摘要：自然语言中的词语不仅传递信息，而且随着文明的发展和人类的迁徙而演变。音乐也是如此。为了理解音乐背后的复杂结构，我们引入了一种名为Essential Element Network（EEN）的算法，将音频编码为文本。该网络通过计算尺度、时间和体积之间的相关性来获得。优化EEN生成的频率和排名的聚类系数的Zipfs定律，使我们能够产生和视为词的语义关系。我们将这些编码词映射到尺度-时间空间，这有助于我们系统地组织音乐深层结构中的句法。我们的算法提供了对音乐背后复杂网络的精确描述，而不是其他深度学习方法的黑盒性质。因此，通过这些过程积累的经验和属性不仅可以为自然语言处理（NLP）的应用提供一种新的方法，而且可以更容易，更客观地分析音乐的演变和发展。摘要：Words in a natural language not only transmit information but also evolve with the development of civilization and human migration. The same is true for music. To understand the complex structure behind the music, we introduced an algorithm called the Essential Element Network (EEN) to encode the audio into text. The network is obtained by calculating the correlations between scales, time, and volume. Optimizing EEN to generate Zipfs law for the frequency and rank of the clustering coefficient enables us to generate and regard the semantic relationships as words. We map these encoded words into the scale-temporal space, which helps us organize systematically the syntax in the deep structure of music. Our algorithm provides precise descriptions of the complex network behind the music, as opposed to the black-box nature of other deep learning approaches. As a result, the experience and properties accumulated through these processes can offer not only a new approach to the applications of Natural Language Processing (NLP) but also an easier and more objective way to analyze the evolution and development of music.

【7】 Enhancing Unsupervised Speech Recognition with Diffusion GANs标题：利用扩散GANS增强无监督语音识别链接：http://arxiv.org/pdf/2303.13559v1

*与cs.SD语音【6】为同一篇

作者：Xianchao Wu机构：NVIDIA备注：5 pages, 1 figure, accepted by ICASSP 2023摘要：我们通过扩散GAN增强了无监督自动语音识别（ASR）的香草对抗训练方法。我们的模型（1）将各种强度的实例噪声注入到生成器的输出和未标记的参考文本中，这些参考文本是从具有长度约束的预训练音素语言模型中采样的，（2）要求扩散时间步相关的鉴别器将它们分开，以及（3）反向传播梯度以更新生成器。单词 音素错误率比较与wav 2 vec-U Librisepeech（3.1%的测试清洁和5.6%的测试其他），TIMIT和MLS数据集，表明我们的增强策略有效地工作。摘要：We enhance the vanilla adversarial training method for unsupervised Automatic Speech Recognition (ASR) by a diffusion-GAN. Our model (1) injects instance noises of various intensities to the generator's output and unlabeled reference text which are sampled from pretrained phoneme language models with a length constraint, (2) asks diffusion timestep-dependent discriminators to separate them, and (3) back-propagates the gradients to update the generator. Word phoneme error rate comparisons with wav2vec-U under Librispeech (3.1% for test-clean and 5.6% for test-other), TIMIT and MLS datasets, show that our enhancement strategies work effectively.

机器翻译，仅供参考