Abstract:Nowadays, the large amount of audio-visual content available has fostered the need to develop new robust automatic speaker diarization systems to analyse and characterise it. This kind of system helps to reduce the cost of doing this process manually and allows the use of the speaker information for different applications, as a huge quantity of information is present, for example, images of faces, or audio recordings. Therefore, this paper aims to address a critical area in the field of speaker diarization systems, the integration of audio-visual content of different domains. This paper seeks to push beyond current state-of-the-art practices by developing a robust audio-visual speaker diarization framework adaptable to various data domains, including TV scenarios, meetings, and daily activities. Unlike most of the existing audio-visual speaker diarization systems, this framework will also include the proposal of an approach to lead the precise assignment of specific identities in TV scenarios where celebrities appear. In addition, in this work, we have conducted an extensive compilation of the current state-of-the-art approaches and the existing databases for developing audio-visual speaker diarization.
Abstract:Representation learning is an approach that allows to discover and extract the factors of variation from the data. Intuitively, a representation is said to be disentangled if it separates the different factors of variation in a way that is understandable to humans. Definitions of disentanglement and metrics to measure it usually assume that the factors of variation are independent of each other. However, this is generally false in the real world, which limits the use of these definitions and metrics to very specific and unrealistic scenarios. In this paper we give a definition of disentanglement based on information theory that is also valid when the factors of variation are not independent. Furthermore, we relate this definition to the Information Bottleneck Method. Finally, we propose a method to measure the degree of disentanglement from the given definition that works when the factors of variation are not independent. We show through different experiments that the method proposed in this paper correctly measures disentanglement with non-independent factors of variation, while other methods fail in this scenario.
Abstract:Prototypical Learning is based on the idea that there is a point (which we call prototype) around which the embeddings of a class are clustered. It has shown promising results in scenarios with little labeled data or to design explainable models. Typically, prototypes are either defined as the average of the embeddings of a class or are designed to be trainable. In this work, we propose to predefine prototypes following human-specified criteria, which simplify the training pipeline and brings different advantages. Specifically, in this work we explore two of these advantages: increasing the inter-class separability of embeddings and disentangling embeddings with respect to different variance factors, which can translate into the possibility of having explainable predictions. Finally, we propose different experiments that help to understand our proposal and demonstrate empirically the mentioned advantages.
Abstract:Audio segmentation is a key task for many speech technologies, most of which are based on neural networks, usually considered as black boxes, with high-level performances. However, in many domains, among which health or forensics, there is not only a need for good performance but also for explanations about the output decision. Explanations derived directly from latent representations need to satisfy "good" properties, such as informativeness, compactness, or modularity, to be interpretable. In this article, we propose an explainable-by-design audio segmentation model based on non-negative matrix factorization (NMF) which is a good candidate for the design of interpretable representations. This paper shows that our model reaches good segmentation performances, and presents deep analyses of the latent representation extracted from the non-negative matrix. The proposed approach opens new perspectives toward the evaluation of interpretable representations according to "good" properties.
Abstract:Unsupervised Multiple Domain Translation is the task of transforming data from one domain to other domains without having paired data to train the systems. Typically, methods based on Generative Adversarial Networks (GANs) are used to address this task. However, our proposal exclusively relies on a modified version of a Variational Autoencoder. This modification consists of the use of two latent variables disentangled in a controlled way by design. One of this latent variables is imposed to depend exclusively on the domain, while the other one must depend on the rest of the variability factors of the data. Additionally, the conditions imposed over the domain latent variable allow for better control and understanding of the latent space. We empirically demonstrate that our approach works on different vision datasets improving the performance of other well known methods. Finally, we prove that, indeed, one of the latent variables stores all the information related to the domain and the other one hardly contains any domain information.
Abstract:Audio signal segmentation is a key task for automatic audio indexing. It consists of detecting the boundaries of class-homogeneous segments in the signal. In many applications, explainable AI is a vital process for transparency of decision-making with machine learning. In this paper, we propose an explainable multilabel segmentation model that solves speech activity (SAD), music (MD), noise (ND), and overlapped speech detection (OSD) simultaneously. This proxy uses the non-negative matrix factorization (NMF) to map the embedding used for the segmentation to the frequency domain. Experiments conducted on two datasets show similar performances as the pre-trained black box model while showing strong explainability features. Specifically, the frequency bins used for the decision can be easily identified at both the segment level (local explanations) and global level (class prototypes).
Abstract:Despite the maturity of modern speaker verification technology, its performance still significantly degrades when facing non-neutrally-phonated (e.g., shouted and whispered) speech. To address this issue, in this paper, we propose a new speaker embedding compensation method based on a minimum mean square error (MMSE) estimator. This method models the joint distribution of the vocal effort transfer vector and non-neutrally-phonated embedding spaces and operates in a principal component analysis domain to cope with non-neutrally-phonated speech data scarcity. Experiments are carried out using a cutting-edge speaker verification system integrating a powerful self-supervised pre-trained model for speech representation. In comparison with a state-of-the-art embedding compensation method, the proposed MMSE estimator yields superior and competitive equal error rate results when tackling shouted and whispered speech, respectively.
Abstract:The presence of decision-making algorithms in society is rapidly increasing nowadays, while concerns about their transparency and the possibility of these algorithms becoming new sources of discrimination are arising. There is a certain consensus about the need to develop AI applications with a Human-Centric approach. Human-Centric Machine Learning needs to be developed based on four main requirements: (i) utility and social good; (ii) privacy and data ownership; (iii) transparency and accountability; and (iv) fairness in AI-driven decision-making processes. All these four Human-Centric requirements are closely related to each other. With the aim of studying how current multimodal algorithms based on heterogeneous sources of information are affected by sensitive elements and inner biases in the data, we propose a fictitious case study focused on automated recruitment: FairCVtest. We train automatic recruitment algorithms using a set of multimodal synthetic profiles including image, text, and structured data, which are consciously scored with gender and racial biases. FairCVtest shows the capacity of the Artificial Intelligence (AI) behind automatic recruitment tools built this way (a common practice in many other application scenarios beyond recruitment) to extract sensitive information from unstructured data and exploit it in combination to data biases in undesirable (unfair) ways. We present an overview of recent works developing techniques capable of removing sensitive information and biases from the decision-making process of deep learning architectures, as well as commonly used databases for fairness research in AI. We demonstrate how learning approaches developed to guarantee privacy in latent spaces can lead to unbiased and fair automatic decision-making process.
Abstract:This paper explores three novel approaches to improve the performance of speaker verification (SV) systems based on deep neural networks (DNN) using Multi-head Self-Attention (MSA) mechanisms and memory layers. Firstly, we propose the use of a learnable vector called Class token to replace the average global pooling mechanism to extract the embeddings. Unlike global average pooling, our proposal takes into account the temporal structure of the input what is relevant for the text-dependent SV task. The class token is concatenated to the input before the first MSA layer, and its state at the output is used to predict the classes. To gain additional robustness, we introduce two approaches. First, we have developed a Bayesian estimation of the class token. Second, we have added a distilled representation token for training a teacher-student pair of networks using the Knowledge Distillation (KD) philosophy, which is combined with the class token. This distillation token is trained to mimic the predictions from the teacher network, while the class token replicates the true label. All the strategies have been tested on the RSR2015-Part II and DeepMine-Part 1 databases for text-dependent SV, providing competitive results compared to the same architecture using the average pooling mechanism to extract average embeddings.
Abstract:Area under the ROC curve (AUC) optimisation techniques developed for neural networks have recently demonstrated their capabilities in different audio and speech related tasks. However, due to its intrinsic nature, AUC optimisation has focused only on binary tasks so far. In this paper, we introduce an extension to the AUC optimisation framework so that it can be easily applied to an arbitrary number of classes, aiming to overcome the issues derived from training data limitations in deep learning solutions. Building upon the multiclass definitions of the AUC metric found in the literature, we define two new training objectives using a one-versus-one and a one-versus-rest approach. In order to demonstrate its potential, we apply them in an audio segmentation task with limited training data that aims to differentiate 3 classes: foreground music, background music and no music. Experimental results show that our proposal can improve the performance of audio segmentation systems significantly compared to traditional training criteria such as cross entropy.