Literacy assessment is an important activity for education administrators across the globe. Typically achieved in a school setting by testing a child's oral reading, it is intensive in human resources. While automatic speech recognition (ASR) is a potential solution to the problem, it tends to be computationally expensive for hand-held devices apart from needing language and accent-specific speech for training. In this work, we propose a system to predict the word-decoding skills of a student based on simple acoustic features derived from the recording. We first identify a meaningful categorization of word-decoding skills by analyzing a manually transcribed data set of children's oral reading recordings. Next the automatic prediction of the category is attempted with the proposed acoustic features. Pause statistics, syllable rate and spectral and intensity dynamics are found to be reliable indicators of specific types of oral reading deficits, providing useful feedback by discriminating the different characteristics of beginning readers. This computationally simple and language-agnostic approach is found to provide a performance close to that obtained using a language dependent ASR that required considerable tuning of its parameters.
相關內容
The scattering of electromagnetic waves from obstacles with wave-material interaction in thin layers on the surface is described by generalized impedance boundary conditions, which provide effective approximate models. In particular, this includes a thin coating around a perfect conductor and the skin effect of a highly conducting material. The approach taken in this work is to derive, analyse and discretize a system of time-dependent boundary integral equations that determines the tangential traces of the scattered electric and magnetic fields. In a familiar second step, the fields are evaluated in the exterior domain by a representation formula, which uses the time-dependent potential operators of Maxwell's equations. The time-dependent boundary integral equationis discretized with Runge--Kutta based convolution quadrature in time and Raviart--Thomas boundary elements in space. Using the frequency-explicit bounds from the well-posedness analysis given here together with known approximation properties of the numerical methods, the full discretization is proved to be stable and convergent, with explicitly given rates in the case of sufficient regularity. Taking the same Runge--Kutta based convolution quadrature for discretizing the time-dependent representation formulas, the optimal order of convergence is obtained away from the scattering boundary, whereas an order reduction occurs close to the boundary. The theoretical results are illustrated by numerical experiments.
Patterns of brain activity are associated with different brain processes and can be used to identify different brain states and make behavioral predictions. However, the relevant features are not readily apparent and accessible. To mine informative latent representations from multichannel recordings of ongoing EEG activity, we propose a novel differentiable decoding pipeline consisting of learnable filters and a pre-determined feature extraction module. Specifically, we introduce filters parameterized by generalized Gaussian functions that offer a smooth derivative for stable end-to-end model training and allow for learning interpretable features. For the feature module, we use signal magnitude and functional connectivity estimates. We demonstrate the utility of our model towards emotion recognition from EEG signals on the SEED dataset, as well as on a new EEG dataset of unprecedented size (i.e., 761 subjects), where we identify consistent trends of music perception and related individual differences. The discovered features align with previous neuroscience studies and offer new insights, such as marked differences in the functional connectivity profile between left and right temporal areas during music listening. This agrees with the respective specialisation of the temporal lobes regarding music perception proposed in the literature.
The spread in the use of tracking systems in sport has made fine-grained spatiotemporal analysis a primary focus of an emerging sports analytics industry. Recently publicized tracking data for men's professional tennis allows for the first detailed spatial analysis of return impact. Mixture models are an appealing model-based framework for spatial analysis in sport, where latent variable discovery is often of primary interest. Although finite mixture models have the advantages of interpretability and scalability, most implementations assume standard parametric distributions for outcomes conditioned on latent variables. In this paper, we present a more flexible alternative that allows the latent conditional distribution to be a mixed member of finite Gaussian mixtures. Our model was motivated by our efforts to describe common styles of return impact location of professional tennis players and is the reason we name the approach a 'latent style allocation' model. In a fully Bayesian implementation, we apply the model to 142,803 return points played by 141 top players at Association of Tennis Professional events between 2018 and 2020 and show that the latent style allocation improves predictive performance over a finite Gaussian mixture model and identifies six unique impact styles on the first and second serve return.
Automatic speech recognition (ASR) systems are ubiquitously present in our daily devices. They are vulnerable to adversarial attacks, where manipulated input samples fool the ASR system's recognition. While adversarial examples for various English ASR systems have already been analyzed, there exists no inter-language comparative vulnerability analysis. We compare the attackability of a German and an English ASR system, taking Deepspeech as an example. We investigate if one of the language models is more susceptible to manipulations than the other. The results of our experiments suggest statistically significant differences between English and German in terms of computational effort necessary for the successful generation of adversarial examples. This result encourages further research in language-dependent characteristics in the robustness analysis of ASR.
Machine reading comprehension (MRC) aims to teach machines to read and comprehend human languages, which is a long-standing goal of natural language processing (NLP). With the burst of deep neural networks and the evolution of contextualized language models (CLMs), the research of MRC has experienced two significant breakthroughs. MRC and CLM, as a phenomenon, have a great impact on the NLP community. In this survey, we provide a comprehensive and comparative review on MRC covering overall research topics about 1) the origin and development of MRC and CLM, with a particular focus on the role of CLMs; 2) the impact of MRC and CLM to the NLP community; 3) the definition, datasets, and evaluation of MRC; 4) general MRC architecture and technical methods in the view of two-stage Encoder-Decoder solving architecture from the insights of the cognitive process of humans; 5) previous highlights, emerging topics, and our empirical analysis, among which we especially focus on what works in different periods of MRC researches. We propose a full-view categorization and new taxonomies on these topics. The primary views we have arrived at are that 1) MRC boosts the progress from language processing to understanding; 2) the rapid improvement of MRC systems greatly benefits from the development of CLMs; 3) the theme of MRC is gradually moving from shallow text matching to cognitive reasoning.
As part of the Human-Computer Interaction field, Expressive speech synthesis is a very rich domain as it requires knowledge in areas such as machine learning, signal processing, sociology, psychology. In this Chapter, we will focus mostly on the technical side. From the recording of expressive speech to its modeling, the reader will have an overview of the main paradigms used in this field, through some of the most prominent systems and methods. We explain how speech can be represented and encoded with audio features. We present a history of the main methods of Text-to-Speech synthesis: concatenative, parametric and statistical parametric speech synthesis. Finally, we focus on the last one, with the last techniques modeling Text-to-Speech synthesis as a sequence-to-sequence problem. This enables the use of Deep Learning blocks such as Convolutional and Recurrent Neural Networks as well as Attention Mechanism. The last part of the Chapter intends to assemble the different aspects of the theory and summarize the concepts.
In many applications, it is important to characterize the way in which two concepts are semantically related. Knowledge graphs such as ConceptNet provide a rich source of information for such characterizations by encoding relations between concepts as edges in a graph. When two concepts are not directly connected by an edge, their relationship can still be described in terms of the paths that connect them. Unfortunately, many of these paths are uninformative and noisy, which means that the success of applications that use such path features crucially relies on their ability to select high-quality paths. In existing applications, this path selection process is based on relatively simple heuristics. In this paper we instead propose to learn to predict path quality from crowdsourced human assessments. Since we are interested in a generic task-independent notion of quality, we simply ask human participants to rank paths according to their subjective assessment of the paths' naturalness, without attempting to define naturalness or steering the participants towards particular indicators of quality. We show that a neural network model trained on these assessments is able to predict human judgments on unseen paths with near optimal performance. Most notably, we find that the resulting path selection method is substantially better than the current heuristic approaches at identifying meaningful paths.
We study the use of the Wave-U-Net architecture for speech enhancement, a model introduced by Stoller et al for the separation of music vocals and accompaniment. This end-to-end learning method for audio source separation operates directly in the time domain, permitting the integrated modelling of phase information and being able to take large temporal contexts into account. Our experiments show that the proposed method improves several metrics, namely PESQ, CSIG, CBAK, COVL and SSNR, over the state-of-the-art with respect to the speech enhancement task on the Voice Bank corpus (VCTK) dataset. We find that a reduced number of hidden layers is sufficient for speech enhancement in comparison to the original system designed for singing voice separation in music. We see this initial result as an encouraging signal to further explore speech enhancement in the time-domain, both as an end in itself and as a pre-processing step to speech recognition systems.
State-of-the-art speech recognition systems rely on fixed, hand-crafted features such as mel-filterbanks to preprocess the waveform before the training pipeline. In this paper, we study end-to-end systems trained directly from the raw waveform, building on two alternatives for trainable replacements of mel-filterbanks that use a convolutional architecture. The first one is inspired by gammatone filterbanks (Hoshen et al., 2015; Sainath et al, 2015), and the second one by the scattering transform (Zeghidour et al., 2017). We propose two modifications to these architectures and systematically compare them to mel-filterbanks, on the Wall Street Journal dataset. The first modification is the addition of an instance normalization layer, which greatly improves on the gammatone-based trainable filterbanks and speeds up the training of the scattering-based filterbanks. The second one relates to the low-pass filter used in these approaches. These modifications consistently improve performances for both approaches, and remove the need for a careful initialization in scattering-based trainable filterbanks. In particular, we show a consistent improvement in word error rate of the trainable filterbanks relatively to comparable mel-filterbanks. It is the first time end-to-end models trained from the raw signal significantly outperform mel-filterbanks on a large vocabulary task under clean recording conditions.
We present a novel deep learning architecture to address the cloze-style question answering task. Existing approaches employ reading mechanisms that do not fully exploit the interdependency between the document and the query. In this paper, we propose a novel \emph{dependent gated reading} bidirectional GRU network (DGR) to efficiently model the relationship between the document and the query during encoding and decision making. Our evaluation shows that DGR obtains highly competitive performance on well-known machine comprehension benchmarks such as the Children's Book Test (CBT-NE and CBT-CN) and Who DiD What (WDW, Strict and Relaxed). Finally, we extensively analyze and validate our model by ablation and attention studies.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191