In this paper, we use data augmentation to improve performance of deep neural network (DNN) embeddings for speaker recognition. The DNN, which is trained to discriminate between speakers, maps variable-length utterances to fixed-dimensional embeddings that we call x-vectors. Prior studies have found that embeddings leverage large-scale training datasets better than i-vectors. However, it can be challenging to collect substantial quantities of labeled data for training. We use data augmentation, consisting of added noise and reverberation, as an inexpensive method to multiply the amount of training data and improve robustness. The x-vectors are compared with i-vector baselines on Speakers in the Wild and NIST SRE 2016 Cantonese. We find that while augmentation is beneficial in the PLDA classifier, it is not helpful in the i-vector extractor. However, the x-vector DNN effectively exploits data augmentation, due to its supervised training. As a result, the x-vectors achieve superior performance on the evaluation datasets.

我们介绍BERTPHONE，一个在大型语音上培训的变压器编码器，输出可以用于扬声器和语言识别的语音感知的上下文表示向量。这是通过对两个目标的培训来实现的：首先是通过调整伯特对连续领域的启发，涉及掩蔽输入框架的跨度并重建用于声学表示学习的整个序列;其次，由ASR的瓶颈特征成功的启发是应用于音素标签的序列级CTC损失，用于语音表示学习。我们预留了两种BERTPHONE型号（一个在FISHER上，一个在TED-lium上），并用它们用作两个任务的X-Vector-Sique DNN中的特征提取器。我们达到最先进的$ C _ {\ TEXT {AVG}} $ 6.16就具有挑战性的LRE07 3SEC封闭式语言识别任务。在Fisher和VoxceleB扬声器识别任务上，我们在培训BertPhone向量而不是MFCC时，我们看到扬声器EER的相对减少18％。通常，BERTPHONE在同一数据上优于先前的语音预制方法。我们在https://github.com/awslabs/speech -representations释放我们的代码和模型。

在这项工作中，我们对情感和压力环境中的文本独立扬声器验证性能进行了实证对比研究。这项工作结合了浅架构的深层模型，导致新的混合分类器。利用了四种不同的混合模型：深神经网络隐藏式马尔可夫模型（DNN-HMM），深神经网络 - 高斯混合模型（DNN-GMM），高斯混合模型 - 深神经网络（GMM-DNN）和隐藏的马尔可夫模型-Deep神经网络（HMM-DNN）。所有模型都基于新颖的实施架构。比较研究使用了三个不同的语音数据集：私人阿拉伯数据集和两个公共英语数据库，即在模拟和实际压力下的演讲（Susas）和情感语音和歌曲（Ravdess）的ryerson视听数据库。上述混合模型的测试结果表明，所提出的HMM-DNN利用情绪和压力环境中的验证性能。结果还表明，HMM-DNN在曲线（AUC）评估度量下的相同错误率（eer）和面积方面优于所有其他混合模型。基于三个数据集的平均所产生的验证系统分别基于HMM-DNN，DNN-HMM，DNN-GMM和GMM-DNN产生7.19％，16.85％，11.51％和11.90％的eERs。此外，我们发现，与两个谈话环境中的所有其他混合模型相比，DNN-GMM模型展示了最少的计算复杂性。相反，HMM-DNN模型需要最多的培训时间。调查结果还证明了EER和AUC值在比较平均情绪和压力表演时依赖于数据库。

Speaker embedding extractors significantly influence the performance of clustering-based speaker diarisation systems. Conventionally, only one embedding is extracted from each speech segment. However, because of the sliding window approach, a segment easily includes two or more speakers owing to speaker change points. This study proposes a novel embedding extractor architecture, referred to as a high-resolution embedding extractor (HEE), which extracts multiple high-resolution embeddings from each speech segment. Hee consists of a feature-map extractor and an enhancer, where the enhancer with the self-attention mechanism is the key to success. The enhancer of HEE replaces the aggregation process; instead of a global pooling layer, the enhancer combines relative information to each frame via attention leveraging the global context. Extracted dense frame-level embeddings can each represent a speaker. Thus, multiple speakers can be represented by different frame-level features in each segment. We also propose an artificially generating mixture data training framework to train the proposed HEE. Through experiments on five evaluation sets, including four public datasets, the proposed HEE demonstrates at least 10% improvement on each evaluation set, except for one dataset, which we analyse that rapid speaker changes less exist.

关节特征本质上是声信号失真的不变，并且已成功地纳入了为正常语音设计的自动语音识别（ASR）系统。它们在非典型任务领域（例如老年人和跨语言的言语无序）的实际应用通常受到从目标扬声器收集此类专家数据的困难。本文介绍了一种跨域和跨语性A2A反演方法，该方法利用了A2A模型中24小时TAL Corpus的平行音频，视觉和超声舌成像（UTI）数据，然后进行交叉训练和交叉训练。语言适用于两种语言的三个数据集：英语dementiabank pitt和antonese JCCOCC MOCA老年演讲Corpora;以及英语Torgo违反语音数据，以产生基于UTI的发音特征。 Experiments conducted on three tasks suggested incorporating the generated articulatory features consistently outperformed the baseline hybrid TDNN and Conformer based end-to-end systems constructed using acoustic features only by statistically significant word error rate or character error rate reductions up to 2.64%, 1.92% and数据增强和说话者适应后，绝对4.17％，7.89％和13.28％相对1.21％。

扬声器日流是一个标签音频或视频录制的任务，与扬声器身份或短暂的任务标记对应于扬声器标识的类，以识别“谁谈到何时发表讲话”。在早期，对MultiSpeaker录音的语音识别开发了扬声器日益衰退算法，以使扬声器自适应处理能够实现扬声器自适应处理。这些算法还将自己的价值作为独立应用程序随着时间的推移，为诸如音频检索等下游任务提供特定于扬声器的核算。最近，随着深度学习技术的出现，这在讲话应用领域的研究和实践中引起了革命性的变化，对扬声器日益改善已经进行了快速进步。在本文中，我们不仅审查了扬声器日益改善技术的历史发展，而且还审查了神经扬声器日益改善方法的最新进步。此外，我们讨论了扬声器日复速度系统如何与语音识别应用相结合，以及最近深度学习的激增是如何引领联合建模这两个组件互相互补的方式。通过考虑这种令人兴奋的技术趋势，我们认为本文对社区提供了有价值的贡献，以通过巩固具有神经方法的最新发展，从而促进更有效的扬声器日益改善进一步进展。

端到端模型在自动语音识别中快速更换传统的混合模型。变压器，基于机器翻译任务的自我关注的序列到序列模型，在用于自动语音识别时已经给出了有希望的结果。本文探讨了在培训基于变压器的模型的同时在编码器输入时结合扬声器信息的不同方式，以提高其语音识别性能。我们以每个扬声器的扬声器嵌入形式呈现扬声器信息。我们使用两种类型的扬声器嵌入进行实验：在我们以前的工作中提出的X-Vectors和新颖的S-Vectors。我们向两个数据集报告结果a）肉kel讲座数据库和b）librispeech 500小时分割。NPTEL是一个开源电子学习门户，提供来自印度顶级大学的讲座。通过我们将扬声器嵌入的方法集成到模型中，我们通过基线获得了基线的错误率的改进。

这项工作旨在自动评估儿童的语言发展是否适合年龄。经过验证的语音和语言测试用于此目的测试听觉记忆。在这项工作中，任务是确定是否正确说出了口语非单词。我们比较有动机来建模特定语言结构的不同方法：低水平特征（FFT），扬声器嵌入（ECAPA-TDNN），素化 - 动机的嵌入（WAV2VEC 2.0）和语音嵌入Senones（ASR ASR ACOSTIC模型）形式。每种方法都提供了类似VGG的5层CNN分类器的输入。我们还检查了每个非单词的适应性。使用来自口头非单词的不同幼儿园的录音进行了对拟议系统的评估。 ECAPA-TDNN和低级FFT特征不会明确模型语音信息； WAV2VEC2.0经过素数标签训练，我们的ASR声学模型包含（子）语音信息。我们发现，语音建模越颗粒状，达到的识别率就越高。在ASR声学模型特征上训练的最佳系统的精度为89.4％，在ROC（接收器操作特征）曲线（AUC）下的面积为0.923。与FFT-BASELINE相比，这对应于20.2％和AUC相对0.309的改善。

在本文中，我们提出了一种解决方案，以允许扬声器条件语音模型，例如VoiceFilter-Lite，以支持单个通过中的任意数量的注册用户。这是通过使用多个扬声器嵌入的注意机制来实现，以计算单个细小嵌入，然后将其用作模型的侧面输入。我们实现了多用户VoiceFilter-Lite并为三个任务进行了评估：（1）流自动语音识别（ASR）任务; （2）独立于文本的扬声器验证任务; （3）个性化关键级检测任务，其中ASR必须在嘈杂的环境中检测来自多个注册用户的关键次数。我们的实验表明，在最多四个注册的用户中，多用户VoiceFilter-Lite能够在具有重叠语音时显着降低语音识别和扬声器验证错误，而不会影响其他声学条件下的性能。这种细心的扬声器嵌入方法也可以轻松应用于其他扬声器条件模型，如个人VAD和个性化ASR。

最近深入学习的突破往往依靠代表学习和知识转移。近年来，开发了用于培养自动语音识别的无监督和自我监督的学习讲话技巧。迄今为止，大多数方法是特定于任务的，并且在特定任务的不同数据集或设置之间进行任务传输学习。反过来，学习任务 - 独立于转移学习的语音和交叉任务应用的代表仍然不那么常见。在这里，我们介绍了一个编码器捕获词级表示的跨任务传输学习。我们展示了预先训练的编码器在四个不同的语音和音频处理任务中的应用：（i）语音增强，（ii）语言识别，（iii）语音，噪声和音乐分类，和（iv）扬声器识别。在每项任务中，我们将跨任务转移学习方法的表现进行比较，以完成任务特定的基准。我们的结果表明，编码器通过预训练捕获的语音表示可在不同的语音处理任务和数据集中可转换。值得注意的是，即使是我们预先训练的编码器的简单应用也优于任务特定的方法，或者取决于任务。

本文介绍了在自动语音识别（ASR）的语境中的声学模型的新型深度学习架构，称为MixNet。除了在LSTM-HMM中的DNN-HMM和存储器单元中的完全连接层之外，该模型使用基于专家（MOE）的混合的两个附加层。在输入时操作的第一个Moe层基于预定义的广义语音类，并且在倒数第二层操作的第二层基于自动学习的声学类。在自然语音中，不同声学类的分布在分布中是不可避免的，这导致帧间错误分类。如果经过修改的传统架构，则预期ASR精度将改进，以使其更适合于占这种重叠。 MixNet正在开发牢记这一点。通过散点图进行的分析验证了MOE确实改善了转化为更好ASR精度的类之间的分离。实验在大型词汇ASR任务上进行，表明，与传统模型，即DNN和LSTM分别提供了13.6％和10.0％的单词误差速率，即使用SMBR标准训练。与用于电话分类的现有方法相比（由EIGEN等人），我们所提出的方法产生了显着的改善。

通过未计算的数据情况和缺乏本领域缺乏标准基准的动机，我们补充了我们以前的努力，并提出了一个专为培训和评估文本无关的多通道扬声器验证系统的全面语料库。还可以容易地用于DERE失去，去噪和语音增强的实验。我们通过利用VOXECEB数据集的清洁部分顶部的数据仿真来解决缺乏多通道训练数据的缺乏问题。开发和评估试验基于复杂的传统的声音，这些声音在复杂的环境环境（声音）语料库中，我们修改以提供多渠道试验。我们发布从公共来源创建数据集的完整食谱作为Multisv语料库，我们提供了两种多通道扬声器验证系统，其中两个多通道扬声器验证系统，基于神经网络的波束成形，基于预测理想二进制掩码或更新的CONV-TASNet更新。

最先进的说话者验证系统本质上取决于某种人类监督，因为它们接受了大量标记数据的培训。但是，手动注释的话语缓慢，昂贵，无法扩展到当今可用的数据量。在这项研究中，我们通过直接从原始音频中学习表征来探索说话者验证的自我监督学习。目的是生成具有较小的言论扬声器和较大言论扬声器差异的稳健扬声器嵌入。我们的方法基于最新信息最大化学习框架和密集的数据增强预处理步骤。我们在表明它们与对比度损失相结合之前表明它们实现更好的性能之前，评估了这些方法在没有对比样本的情况下工作的能力。此外，我们进行实验表明，与现有技术相比，我们的方法达到了竞争成果，并且在用一小部分标记数据进行微调时，与监督基线相比，可以获得更好的性能。

While modern Text-to-Speech (TTS) systems can produce speech rated highly in terms of subjective evaluation, the distance between real and synthetic speech distributions remains understudied, where we use the term \textit{distribution} to mean the sample space of all possible real speech recordings from a given set of speakers; or of the synthetic samples that could be generated for the same set of speakers. We evaluate the distance of real and synthetic speech distributions along the dimensions of the acoustic environment, speaker characteristics and prosody using a range of speech processing measures and the respective Wasserstein distances of their distributions. We reduce these distribution distances along said dimensions by providing utterance-level information derived from the measures to the model and show they can be generated at inference time. The improvements to the dimensions translate to overall distribution distance reduction approximated using Automatic Speech Recognition (ASR) by evaluating the fitness of the synthetic data as training data.

State-of-the-art speaker verification frameworks have typically focused on speech enhancement techniques with increasingly deeper (more layers) and wider (number of channels) models to improve their verification performance. Instead, this paper proposes an approach to increase the model resolution capability using attention-based dynamic kernels in a convolutional neural network to adapt the model parameters to be feature-conditioned. The attention weights on the kernels are further distilled by channel attention and multi-layer feature aggregation to learn global features from speech. This approach provides an efficient solution to improving representation capacity with lower data resources. This is due to the self-adaptation to inputs of the structures of the model parameters. The proposed dynamic convolutional model achieved 1.62\% EER and 0.18 miniDCF on the VoxCeleb1 test set and has a 17\% relative improvement compared to the ECAPA-TDNN.

Automatic Speech Recognition (ASR) for air traffic control is generally trained by pooling Air Traffic Controller (ATCO) and pilot data into one set. This is motivated by the fact that pilot's voice communications are more scarce than ATCOs. Due to this data imbalance and other reasons (e.g., varying acoustic conditions), the speech from ATCOs is usually recognized more accurately than from pilots. Automatically identifying the speaker roles is a challenging task, especially in the case of the noisy voice recordings collected using Very High Frequency (VHF) receivers or due to the unavailability of the push-to-talk (PTT) signal, i.e., both audio channels are mixed. In this work, we propose to (1) automatically segment the ATCO and pilot data based on an intuitive approach exploiting ASR transcripts and (2) subsequently consider an automatic recognition of ATCOs' and pilots' voice as two separate tasks. Our work is performed on VHF audio data with high noise levels, i.e., signal-to-noise (SNR) ratios below 15 dB, as this data is recognized to be helpful for various speech-based machine-learning tasks. Specifically, for the speaker role identification task, the module is represented by a simple yet efficient knowledge-based system exploiting a grammar defined by the International Civil Aviation Organization (ICAO). The system accepts text as the input, either manually verified annotations or automatically generated transcripts. The developed approach provides an average accuracy in speaker role identification of about 83%. Finally, we show that training an acoustic model for ASR tasks separately (i.e., separate models for ATCOs and pilots) or using a multitask approach is well suited for the noisy data and outperforms the traditional ASR system where all data is pooled together.

尽管针对正常语音的自动语音识别（ASR）技术取得了迅速的进展，但迄今为止，准确认识违反障碍和老年语音仍然是高度挑战的任务。由于这些用户中经常发现的移动性问题，很难为ASR系统开发收集大量此类数据。为此，数据增强技术起着至关重要的作用。与现有的数据增强技术相反，仅修改光谱轮廓的说话速率或整体形状，使用一组新颖的扬声器依赖（SD）生成对抗网络（Gan ）本文基于数据增强方法。这些既可以灵活地允许：a）在可用的语音数据可用时修改时间或速度的正常语音光谱，并更接近受损说话者的扬声器； b）对于非平行数据，SVD分解了正常语音频谱基础特征，要转换为目标老年人说话者的特征，然后再与时间基础重组以生成最先进的TDNN的增强数据和构象体ASR系统培训。实验是针对四个任务进行的：英语Uapseech和Torgo违反语音语音Corpora；英国痴呆症皮特和广东话JCCOCC MOCA老年语音数据集。所提出的基于GAN的数据增强方法始终优于基线速度扰动方法，最多可在Torgo和Dementiabank数据上降低4.91％和3.0％的绝对速度（相对相对9.61％和6.4％）。应用基于LHUC的扬声器适应后，保留了一致的性能改进。

以前的研究已经证实了利用明晰度信息达到改善的语音增强（SE）性能的有效性。通过使用铰接特征的地点/方式增强原始声学特征，可以引导SE过程考虑执行增强时输入语音的剖视特性。因此，我们认为关节属性的上下文信息应包括有用的信息，并可以进一步利用不同的语言。在这项研究中，我们提出了一个SE系统，通过优化英语和普通话的增强演讲中的上下文清晰度信息来提高其性能。我们通过联合列车与端到端的自动语音识别（E2E ASR）模型进行联合列车，预测广播序列（BPC）而不是单词序列的序列。同时，开发了两种培训策略，以基于基于BPC的ASR：多任务学习和深度特征培训策略来培训SE系统。 Timit和TMhint DataSet上的实验结果证实了上下文化学信息促进了SE系统，以实现比传统声学模型（AM）更好的结果。此外，与用单声道ASR培训的另一SE系统相比，基于BPC的ASR（提供上下文化学信息）可以在不同的信噪比（SNR）下更有效地改善SE性能。

Personal assistants, automatic speech recognizers and dialogue understanding systems are becoming more critical in our interconnected digital world. A clear example is air traffic control (ATC) communications. ATC aims at guiding aircraft and controlling the airspace in a safe and optimal manner. These voice-based dialogues are carried between an air traffic controller (ATCO) and pilots via very-high frequency radio channels. In order to incorporate these novel technologies into ATC (low-resource domain), large-scale annotated datasets are required to develop the data-driven AI systems. Two examples are automatic speech recognition (ASR) and natural language understanding (NLU). In this paper, we introduce the ATCO2 corpus, a dataset that aims at fostering research on the challenging ATC field, which has lagged behind due to lack of annotated data. The ATCO2 corpus covers 1) data collection and pre-processing, 2) pseudo-annotations of speech data, and 3) extraction of ATC-related named entities. The ATCO2 corpus is split into three subsets. 1) ATCO2-test-set corpus contains 4 hours of ATC speech with manual transcripts and a subset with gold annotations for named-entity recognition (callsign, command, value). 2) The ATCO2-PL-set corpus consists of 5281 hours of unlabeled ATC data enriched with automatic transcripts from an in-domain speech recognizer, contextual information, speaker turn information, signal-to-noise ratio estimate and English language detection score per sample. Both available for purchase through ELDA at http://catalog.elra.info/en-us/repository/browse/ELRA-S0484. 3) The ATCO2-test-set-1h corpus is a one-hour subset from the original test set corpus, that we are offering for free at https://www.atco2.org/data. We expect the ATCO2 corpus will foster research on robust ASR and NLU not only in the field of ATC communications but also in the general research community.

The objective of this paper is speaker recognition under noisy and unconstrained conditions.We make two key contributions. First, we introduce a very large-scale audio-visual speaker recognition dataset collected from open-source media. Using a fully automated pipeline, we curate VoxCeleb2 which contains over a million utterances from over 6,000 speakers. This is several times larger than any publicly available speaker recognition dataset.Second, we develop and compare Convolutional Neural Network (CNN) models and training strategies that can effectively recognise identities from voice under various conditions. The models trained on the VoxCeleb2 dataset surpass the performance of previous works on a benchmark dataset by a significant margin.