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】 Dual-Attention Neural Transducers for Efficient Wake Word Spotting in Speech Recognition标题：双注意神经传感器用于语音识别中的有效尾音检测链接：http://arxiv.org/pdf/2304.01905v1

作者：Saumya Y. Sahai,Jing Liu,Thejaswi Muniyappa,Kanthashree M. Sathyendra,Anastasios Alexandridis,Grant P. Strimel,Ross McGowan,Ariya Rastrow,Feng-Ju Chang,Athanasios Mouchtaris,Siegfried Kunzmann机构：Amazon Alexa AI, USA备注：Accepted in ICASSP 2023摘要：我们提出了双注意力神经偏置，一个架构，旨在提高唤醒词（WW）的识别和改善语音识别任务的推理时间延迟。该架构通过利用WW定位来选择针对输入音频帧执行其注意力网络的哪个分支，从而实现其运行时计算路径的动态切换。通过这种方法，我们有效地提高了WW定位精度，同时节省了由浮点运算（FLOPs）定义的运行时计算成本。使用内部去识别数据集，我们证明了所提出的双注意力网络可以减少计算成本$90 %$WW音频帧，只有$1 %$的参数的数量增加。与基线相比，此架构将WW F1分数相对提高了16 %$，并将通用罕见字错误率相对提高了3 %$。摘要：We present dual-attention neural biasing, an architecture designed to boost Wake Words (WW) recognition and improve inference time latency on speech recognition tasks. This architecture enables a dynamic switch for its runtime compute paths by exploiting WW spotting to select which branch of its attention networks to execute for an input audio frame. With this approach, we effectively improve WW spotting accuracy while saving runtime compute cost as defined by floating point operations (FLOPs). Using an in-house de-identified dataset, we demonstrate that the proposed dual-attention network can reduce the compute cost by $90 %$ for WW audio frames, with only $1 %$ increase in the number of parameters. This architecture improves WW F1 score by $16 %$ relative and improves generic rare word error rate by $3 %$ relative compared to the baselines.

【2】 Independent Vector Extraction Constrained on Manifold of Half-Length Filters标题：半长滤波流形上的独立向量提取链接：http://arxiv.org/pdf/2304.01778v1

作者：Zbyněk Koldovský,Jaroslav Čmejla,Tülay Adalı,Stephen O'Regan机构：Technical University of Liberec, †Department of Electrical and Computer EngineeringUniversity of Maryland, which means that the separated sources have different ordersThis work was supported by the department of the Navy摘要：独立向量分析（Independent Vector Analysis，IVA）是独立分量分析（Independent Component Analysis，ICA）的一种流行的扩展，用于一组瞬时线性混合信号的联合分离，直接应用于频域说话人分离或提取。通过混合矩阵来参数化混合物，每个混合物一个矩阵。这意味着IVA混合模型不考虑跨混合物 频率的参数之间的任何关系。分离联合进行，只有通过源模型，其中考虑到跨混合物的源的统计依赖性。在本文中，我们提出了一个混合模型的联合盲源提取的混合模型参数的频率连接。这是通过约束的可行参数集的半长度分离滤波器，这在频域说话人提取有一个明确的解释和应用的流形。摘要：Independent Vector Analysis (IVA) is a popular extension of Independent Component Analysis (ICA) for joint separation of a set of instantaneous linear mixtures, with a direct application in frequency-domain speaker separation or extraction. The mixtures are parameterized by mixing matrices, one matrix per mixture. This means that the IVA mixing model does not account for any relationships between parameters across the mixtures frequencies. The separation proceeds jointly only through the source model, where statistical dependencies of sources across the mixtures are taken into account. In this paper, we propose a mixing model for joint blind source extraction where the mixing model parameters are linked across the frequencies. This is achieved by constraining the set of feasible parameters to the manifold of half-length separating filters, which has a clear interpretation and application in frequency-domain speaker extraction.
eess.AS音频处理

【1】 Independent Vector Extraction Constrained on Manifold of Half-Length Filters标题：半长滤波流形上的独立向量提取链接：http://arxiv.org/pdf/2304.01778v1

*与cs.SD语音【2】为同一篇

作者：Zbyněk Koldovský,Jaroslav Čmejla,Tülay Adalı,Stephen O'Regan机构：Technical University of Liberec, †Department of Electrical and Computer EngineeringUniversity of Maryland, which means that the separated sources have different ordersThis work was supported by the department of the Navy摘要：独立向量分析（Independent Vector Analysis，IVA）是独立分量分析（Independent Component Analysis，ICA）的一种流行的扩展，用于一组瞬时线性混合信号的联合分离，直接应用于频域说话人分离或提取。通过混合矩阵来参数化混合物，每个混合物一个矩阵。这意味着IVA混合模型不考虑跨混合物 频率的参数之间的任何关系。分离联合进行，只有通过源模型，其中考虑到跨混合物的源的统计依赖性。在本文中，我们提出了一个混合模型的联合盲源提取的混合模型参数的频率连接。这是通过约束的可行参数集的半长度分离滤波器，这在频域说话人提取有一个明确的解释和应用的流形。摘要：Independent Vector Analysis (IVA) is a popular extension of Independent Component Analysis (ICA) for joint separation of a set of instantaneous linear mixtures, with a direct application in frequency-domain speaker separation or extraction. The mixtures are parameterized by mixing matrices, one matrix per mixture. This means that the IVA mixing model does not account for any relationships between parameters across the mixtures frequencies. The separation proceeds jointly only through the source model, where statistical dependencies of sources across the mixtures are taken into account. In this paper, we propose a mixing model for joint blind source extraction where the mixing model parameters are linked across the frequencies. This is achieved by constraining the set of feasible parameters to the manifold of half-length separating filters, which has a clear interpretation and application in frequency-domain speaker extraction.

【2】 TorchAudio-Squim: Reference-less Speech Quality and Intelligibility measures in TorchAudio标题：TorchAudio-Squim：TorchAudio中的无参考语音质量和清晰度测量链接：http://arxiv.org/pdf/2304.01448v1

作者：Anurag Kumar,Ke Tan,Zhaoheng Ni,Pranay Manocha,Xiaohui Zhang,Ethan Henderson,Buye Xu机构：Meta Reality Labs Research, Meta, Princeton University备注：ICASSP 2023摘要：测量语音信号的质量和可懂度通常是开发语音处理系统的关键步骤。为了实现这一点，已经开发了在不同假设下测量质量和可理解性的各种度量。通过本文，我们介绍了使用深度神经网络来估计这些已知指标的工具和一组模型。这些模型可以在成熟的TorchAudio库中使用，TorchAudio库是PyTorch深度学习框架中的核心音频和语音处理库。我们将其称为TorchAudio-Squim，TorchAudio-Speech QUality and Intelligibility Measures。更具体地说，在当前版本的TorchAudio-squim中，我们建立并发布了用于估计客观指标中的PESQ、STOI和SI-SDR以及主观指标中的MOS的模型。我们开发了一种新的客观度量估计的方法，并使用最近开发的主观度量估计的方法。这些模型以“无参考”的方式运行，即它们不需要相应的干净语音作为语音评估的参考。考虑到在现实世界的情况下，干净的语音和主观评价的费力过程的不可用性，这种易于使用的工具将大大有利于语音处理的研究和开发。摘要：Measuring quality and intelligibility of a speech signal is usually a critical step in development of speech processing systems. To enable this, a variety of metrics to measure quality and intelligibility under different assumptions have been developed. Through this paper, we introduce tools and a set of models to estimate such known metrics using deep neural networks. These models are made available in the well-established TorchAudio library, the core audio and speech processing library within the PyTorch deep learning framework. We refer to it as TorchAudio-Squim, TorchAudio-Speech QUality and Intelligibility Measures. More specifically, in the current version of TorchAudio-squim, we establish and release models for estimating PESQ, STOI and SI-SDR among objective metrics and MOS among subjective metrics. We develop a novel approach for objective metric estimation and use a recently developed approach for subjective metric estimation. These models operate in a reference-less" manner, that is they do not require the corresponding clean speech as reference for speech assessment. Given the unavailability of clean speech and the effortful process of subjective evaluation in real-world situations, such easy-to-use tools would greatly benefit speech processing research and development.

【3】 Dual-Attention Neural Transducers for Efficient Wake Word Spotting in Speech Recognition标题：双注意神经传感器用于语音识别中的有效尾音检测链接：http://arxiv.org/pdf/2304.01905v1

*与cs.SD语音【1】为同一篇

作者：Saumya Y. Sahai,Jing Liu,Thejaswi Muniyappa,Kanthashree M. Sathyendra,Anastasios Alexandridis,Grant P. Strimel,Ross McGowan,Ariya Rastrow,Feng-Ju Chang,Athanasios Mouchtaris,Siegfried Kunzmann机构：Amazon Alexa AI, USA备注：Accepted in ICASSP 2023摘要：我们提出了双注意力神经偏置，一个架构，旨在提高唤醒词（WW）的识别和改善语音识别任务的推理时间延迟。该架构通过利用WW定位来选择针对输入音频帧执行其注意力网络的哪个分支，从而实现其运行时计算路径的动态切换。通过这种方法，我们有效地提高了WW定位精度，同时节省了由浮点运算（FLOPs）定义的运行时计算成本。使用内部去识别数据集，我们证明了所提出的双注意力网络可以减少计算成本$90 %$WW音频帧，只有$1 %$的参数的数量增加。与基线相比，此架构将WW F1分数相对提高了16 %$，并将通用罕见字错误率相对提高了3 %$。摘要：We present dual-attention neural biasing, an architecture designed to boost Wake Words (WW) recognition and improve inference time latency on speech recognition tasks. This architecture enables a dynamic switch for its runtime compute paths by exploiting WW spotting to select which branch of its attention networks to execute for an input audio frame. With this approach, we effectively improve WW spotting accuracy while saving runtime compute cost as defined by floating point operations (FLOPs). Using an in-house de-identified dataset, we demonstrate that the proposed dual-attention network can reduce the compute cost by $90 %$ for WW audio frames, with only $1 %$ increase in the number of parameters. This architecture improves WW F1 score by $16 %$ relative and improves generic rare word error rate by $3 %$ relative compared to the baselines.

机器翻译，仅供参考