Hidden Markov models (HMMs) have been successful for modelling the dynamics of carefully dictated speech, but their performance degrades severely when used to model conversational speech. Since speech is produced by a system of loosely coupled articulators, stochastic models explicitly representing this parallelism may have advantages for automatic speech recognition (ASR), particularly when trying to model the phonological effects inherent in casual spontaneous speech. This paper presents a preliminary feasibility study of one such (...) model class: loosely coupled HMMs. Exact model estimation and decoding is potentially expensive, so possible approximate algorithms are also discussed. Comparison of one particular loosely coupled model on an isolated word task suggests loosely coupled HMMs merit further investigation. An approximate algorithm giving performance which is almost always statistically indistinguishable from the exact algorithm is also identified, making more extensive research computationally feasible. (shrink)

Speech recognition is an interdisciplinary subfield of natural language processing (NLP) that facilitates the recognition and translation of spoken language into text by machine. Speech recognition plays an important role in digital transformation. It is widely used in different areas such as education, industry, and healthcare and has recently been used in many Internet of Things and Machine Learning applications. The process of speech recognition is one of the most difficult processes in computer (...) science. Despite numerous searches in this domain, an optimal method for speech recognition has not yet been found. This is due to the fact that there are many attributes that characterize natural languages and every language has its particular highlights. The aim of this research is to provide a comprehensive understanding of the various techniques within the domain of Speech Recognition through a systematic literature review of the existing work. We will introduce the most significant and relevant techniques that may provide some directions in the future research. (shrink)

Deep neural networks (DNNs) have been playing a significant role in acoustic modeling. Convolutional neural networks (CNNs) are the advanced version of DNNs that achieve 4–12% relative gain in the word error rate (WER) over DNNs. Existence of spectral variations and local correlations in speech signal makes CNNs more capable of speech recognition. Recently, it has been demonstrated that bidirectional long short-term memory (BLSTM) produces higher recognition rate in acoustic modeling because they are adequate to reinforce (...) higher-level representations of acoustic data. Spatial and temporal properties of the speech signal are essential for high recognition rate, so the concept of combining two different networks came into mind. In this paper, a hybrid architecture of CNN-BLSTM is proposed to appropriately use these properties and to improve the continuous speech recognition task. Further, we explore different methods like weight sharing, the appropriate number of hidden units, and ideal pooling strategy for CNN to achieve a high recognition rate. Specifically, the focus is also on how many BLSTM layers are effective. This paper also attempts to overcome another shortcoming of CNN, i.e. speaker-adapted features, which are not possible to be directly modeled in CNN. Next, various non-linearities with or without dropout are analyzed for speech tasks. Experiments indicate that proposed hybrid architecture with speaker-adapted features and maxout non-linearity with dropout idea shows 5.8% and 10% relative decrease in WER over the CNN and DNN systems, respectively. (shrink)

This paper implements the continuous Hindi Automatic Speech Recognition (ASR) system using the proposed integrated features vector with Recurrent Neural Network (RNN) based Language Modeling (LM). The proposed system also implements the speaker adaptation using Maximum-Likelihood Linear Regression (MLLR) and Constrained Maximum likelihood Linear Regression (C-MLLR). This system is discriminatively trained by Maximum Mutual Information (MMI) and Minimum Phone Error (MPE) techniques with 256 Gaussian mixture per Hidden Markov Model(HMM) state. The training of the baseline system has (...) been done using a phonetically rich Hindi dataset. The results show that discriminative training enhances the baseline system performance by up to 3%. Further improvement of ~7% has been recorded by applying RNN LM. The proposed Hindi ASR system shows significant performance improvement over other current state-of-the-art techniques. (shrink)

Despite the lack of invariance problem (the many‐to‐many mapping between acoustics and percepts), human listeners experience phonetic constancy and typically perceive what a speaker intends. Most models of human speech recognition (HSR) have side‐stepped this problem, working with abstract, idealized inputs and deferring the challenge of working with real speech. In contrast, carefully engineered deep learning networks allow robust, real‐world automatic speech recognition (ASR). However, the complexities of deep learning architectures and training regimens make (...) it difficult to use them to provide direct insights into mechanisms that may support HSR. In this brief article, we report preliminary results from a two‐layer network that borrows one element from ASR, long short‐term memory nodes, which provide dynamic memory for a range of temporal spans. This allows the model to learn to map real speech from multiple talkers to semantic targets with high accuracy, with human‐like timecourse of lexical access and phonological competition. Internal representations emerge that resemble phonetically organized responses in human superior temporal gyrus, suggesting that the model develops a distributed phonological code despite no explicit training on phonetic or phonemic targets. The ability to work with real speech is a major advance for cognitive models of HSR. (shrink)

Although researchers studying human speech recognition (HSR) and automatic speech recognition (ASR) share a common interest in how information processing systems (human or machine) recognize spoken language, there is little communication between the two disciplines. We suggest that this lack of communication follows largely from the fact that research in these related fields has focused on the mechanics of how speech can be recognized. In Marr's (1982) terms, emphasis has been on the algorithmic and (...) implementational levels rather than on the computational level. In this article, we provide a computational‐level analysis of the task of speech recognition, which reveals the close parallels between research concerned with HSR and ASR. We illustrate this relation by presenting a new computational model of human spoken‐word recognition, built using techniques from the field of ASR that, in contrast to current existing models of HSR, recognizes words from real speech input. (shrink)

Although researchers studying human speech recognition (HSR) and automatic speech recognition (ASR) share a common interest in how information processing systems (human or machine) recognize spoken language, there is little communication between the two disciplines. We suggest that this lack of communication follows largely from the fact that research in these related fields has focused on the mechanics of how speech can be recognized. In Marr's (1982) terms, emphasis has been on the algorithmic and (...) implementational levels rather than on the computational level. In this article, we provide a computational-level analysis of the task of speech recognition, which reveals the close parallels between research concerned with HSR and ASR. We illustrate this relation by presenting a new computational model of human spoken-word recognition, built using techniques from the field of ASR that, in contrast to current existing models of HSR, recognizes words from real speech input. (shrink)

In this paper, we propose to incorporate the local attention in WaveNet-CTC to improve the performance of Tibetan speech recognition in multitask learning. With an increase in task number, such as simultaneous Tibetan speech content recognition, dialect identification, and speaker recognition, the accuracy rate of a single WaveNet-CTC decreases on speech recognition. Inspired by the attention mechanism, we introduce the local attention to automatically tune the weights of feature frames in a window and (...) pay different attention on context information for multitask learning. The experimental results show that our method improves the accuracies of speech recognition for all Tibetan dialects in three-task learning, compared with the baseline model. Furthermore, our method significantly improves the accuracy for low-resource dialect by 5.11% against the specific-dialect model. (shrink)

In this paper, we study the performance of baseline hidden Markov model (HMM) for segmentation of speech signals. It is applied on single-speaker segmentation task, using Hindi speech database. The automatic phoneme segmentation framework evolved imitates the human phoneme segmentation process. A set of 44 Hindi phonemes were chosen for the segmentation experiment, wherein we used continuous density hidden Markov model (CDHMM) with a mixture of Gaussian distribution. The left-to-right topology with no skip states has been selected (...) as it is effective in speech recognition due to its consistency with the natural way of articulating the spoken words. This system accepts speech utterances along with their orthographic “transcriptions” and generates segmentation information of the speech. This corpus was used to develop context-independent hidden Markov models (HMMs) for each of the Hindi phonemes. The system was trained using numerous sentences that are relevant to provide information to the passengers of the Metro Rail. The system was validated against a few manually segmented speech utterances. The evaluation of the experiments shows that the best performance is obtained by using a combination of two Gaussians mixtures and five HMM states. A category-wise phoneme error analysis has been performed, and the performance of the phonetic segmentation has been reported. The modeling of HMMs has been implemented using Microsoft Visual Studio 2005 (C++), and the system is designed to work on Windows operating system. The goal of this study is automatic segmentation of speech at phonetic level. (shrink)

Action perception and recognition are core abilities fundamental for human social interaction. A parieto-frontal network (the mirror neuron system) matches visually presented biological motion information onto observers' motor representations. This process of matching the actions of others onto our own sensorimotor repertoire is thought to be important for action recognition, providing a non-mediated “motor perception” based on a bidirectional flow of information along the mirror parieto-frontal circuits. State-of-the-art machine learning strategies for hand action identification have shown better performances (...) when sensorimotor data, as opposed to visual information only, are available during learning. As speech is a particular type of action (with acoustic targets), it is expected to activate a mirror neuron mechanism. Indeed, in speech perception, motor centers have been shown to be causally involved in the discrimination of speech sounds. In this paper, we review recent neurophysiological and machine learning-based studies showing (a) the specific contribution of the motor system to speech perception and (b) that automatic phone recognition is significantly improved when motor data are used during training of classifiers (as opposed to learning from purely auditory data). (shrink)

_Introduction:_ Clinically relevant information can go uncaptured in the conventional scoring of a verbal fluency test. We hypothesize that characterizing the temporal aspects of the response through a set of time related measures will be useful in distinguishing those with MCI from cognitively intact controls. _Methods:_ Audio recordings of an animal fluency test administered to 70 demographically matched older adults (mean age 90.4 years), 28 with mild cognitive impairment (MCI) and 42 cognitively intact (CI) were professionally transcribed and fed into (...) an automatic speech recognition (ASR) system to estimate the start time of each recalled word in the response. Next, we semantically cluster participant generated animal names and through a novel set of time-based measures, we characterize the semantic search strategy of subjects in retrieving words from animal name clusters. This set of time-based features along with standard count-based features (e.g., number of correctly retrieved animal names) were then used in a machine learning algorithm trained for distinguishing those with MCI from CI controls. _Results:_ The combination of both count-based and time-based features, automatically derived from the test response, achieved 77% on AUC-ROC of the support vector machine (SVM) classifier, outperforming the model trained only on the raw test score (AUC, 65%), and well above the chance model (AUC, 50%). _Conclusion:_ This approach supports the value of introducing time-based measures to the assessment of verbal fluency in the context of this generative task differentiating subjects with MCI from those with intact cognition. (shrink)

This paper outlines the background development of “intelligent“ technologies such as speech recognition. Despite significant progress in the development of these technologies, they still fall short in many areas, and rapid advances in areas such as dictation are actually stalled. In this paper we have proposed semi-automatic solutions — smart integration of human and intelligent efforts. One such technique involves improvement to the speech recognition editing interface, thereby reducing the perception of errors to the viewer. (...) Other techniques that are described in the paper are batch enrollment, which allows the user to reduce the amount of time required for enrollment, and content spotting, which can be used for applications that have repeated content flow, such as movies or museum tours. The paper also suggests a general concept of distributive training of speech recognition systems that is based on data collection across a network. (shrink)

In a series of 5 experiments, we investigated whether the processing of phonologically assimilated utterances is influenced by language learning. Previous experiments had shown that phonological assimilations, such as /lean#bacon/→ [leam bacon], are compensated for in perception. In this article, we investigated whether compensation for assimilation can occur without experience with an assimilation rule using automatic event-related potentials. Our first experiment indicated that Dutch listeners compensate for a Hungarian assimilation rule. Two subsequent experiments, however, failed to show compensation for (...) assimilation by both Dutch and Hungarian listeners. Two additional experiments showed that this was due to the acoustic properties of the assimilated utterance, confirming earlier reports that phonetic detail is important in compensation for assimilation. Our data indicate that compensation for assimilation can occur without experience with an assimilation rule, in line with phonetic-phonological theories that assume that speech production is influenced by speech-perception abilities. (shrink)

Miscommunication phenomena such as repair in dialogue are important indicators of the quality of communication. Automatic detection is therefore a key step toward tools that can characterize communication quality and thus help in applications from call center management to mental health monitoring. However, most existing computational linguistic approaches to these phenomena are unsuitable for general use in this way, and particularly for analyzing human–human dialogue: Although models of other-repair are common in human-computer dialogue systems, they tend to focus on (...) specific phenomena, missing the range of repair and repair initiation forms used by humans; and while self-repair models for speech recognition and understanding are advanced, they tend to focus on removal of “disfluent” material important for full understanding of the discourse contribution, and/or rely on domain-specific knowledge. We explain the requirements for more satisfactory models, including incrementality of processing and robustness to sparsity. We then describe models for self- and other-repair detection that meet these requirements and investigate how they perform on datasets from a range of dialogue genres and domains, with promising results. (shrink)

Top-down feedback does not benefit speech recognition; on the contrary, it can hinder it. No experimental data imply that feedback loops are required for speech recognition. Feedback is accordingly unnecessary and spoken word recognition is modular. To defend this thesis, we analyse lexical involvement in phonemic decision making. TRACE (McClelland & Elman 1986), a model with feedback from the lexicon to prelexical processes, is unable to account for all the available data on phonemic decision making. (...) The modular Race model (Cutler & Norris 1979) is likewise challenged by some recent results, however. We therefore present a new modular model of phonemic decision making, the Merge model. In Merge, information flows from prelexical processes to the lexicon without feedback. Because phonemic decisions are based on the merging of prelexical and lexical information, Merge correctly predicts lexical involvement in phonemic decisions in both words and nonwords. Computer simulations show how Merge is able to account for the data through a process of competition between lexical hypotheses. We discuss the issue of feedback in other areas of language processing and conclude that modular models are particularly well suited to the problems and constraints of speech recognition. Key Words: computational modeling; feedback; lexical processing; modularity; phonemic decisions; reading; speech recognition; word recognition. (shrink)

It is well known that the unpredictable speech production brought on by stress from the task at hand has a significant negative impact on the performance of speech processing algorithms. Speech therapy benefits from being able to detect stress in speech. Speech processing performance suffers noticeably when perceptually produced stress causes variations in speech production. Using the acoustic speech signal to objectively characterize speaker stress is one method for assessing production variances brought on (...) by stress. Real-world complexity and ambiguity make it difficult for decision-makers to express their conclusions with clarity in their speech. In particular, the Neutrosophic speech algorithm is used to encode the language variables because they cannot be computed directly. Neutrosophic sets are used to manage indeterminacy in a practical situation. Existing algorithms are used except for stress on Neutrosophic speech recognition. The creation of algorithms that calculate, categorize, or differentiate between different stress circumstances. Understanding stress and developing strategies to combat its effects on speech recognition and human-computer interaction system are the goals of this recognition. (shrink)

Cochlear implants (CI) support the development of oral language in hearing-impaired children. However, even with CI, speech recognition in noise (SRiN) is limited. This raised the question, whether these restrictions are related to the quality of life (QoL) of children and adolescents with CI and how SRiN and QoL are related to each other. As a result of a systematic literature research only three studies were found, indicating positive moderating effects between SRiN and QoL of young CI users. (...) Thirty studies addressed the quality of life of children and adolescents with CI. Following the criteria of the World Health Organization (WHO) for pediatric HRQoL (1994) only a minority used validated child centered and age appropriate QoL instruments. Moreover, despite the consensus that usually children and adolescents are the most prominent informants of their own QoL (parent- reports complement the information of the children) only a minority of investigators used self-reports. Restricted SRiN may be a burden for the QoL of children and adolescents with CI. Up to now the CI community does not seem to be aware of a possible impairment of QoL in young CI users. Further studies addressing this topic are urgently needed, which is also relevant for parents, clinicians, therapists, teachers, and politicians. Additionally investigators should use valid pediatric QoL instruments. They should be aware, that most of the young CI users are able to inform about their quality of life also themselves (additionally to parents). (shrink)

Speech recognition technology has played an indispensable role in realizing human-computer intelligent interaction. However, most of the current Chinese speech recognition systems are provided online or offline models with low accuracy and poor performance. To improve the performance of offline Chinese speech recognition, we propose a hybrid acoustic model of deep convolutional neural network, long short-term memory, and deep neural network. This model utilizes DCNN to reduce frequency variation and adds a batch normalization layer (...) after its convolutional layer to ensure the stability of data distribution, and then use LSTM to effectively solve the gradient vanishing problem. Finally, the fully connected structure of DNN is utilized to efficiently map the input features into a separable space, which is helpful for data classification. Therefore, leveraging the strengths of DCNN, LSTM, and DNN by combining them into a unified architecture can effectively improve speech recognition performance. Our model was tested on the open Chinese speech database THCHS-30 released by the Center for Speech and Language Technology of Tsinghua University, and it was concluded that the DLD model with 3 layers of LSTM and 3 layers of DNN had the best performance, reaching 13.49% of words error rate. (shrink)

Advances in technology have transformed and expanded the ways in which policing is run. One new manifestation is the mass acquisition and processing of private facial images via automatic facial recognition by the police: what we conceptualise as AFR-based policing. However, there is still a lack of clarity on the manner and extent to which this largely-unregulated technology is used by law enforcement agencies and on its impact on fundamental rights. Social understanding and involvement are still insufficient in (...) the context of AFR technologies, which in turn affects social trust in and legitimacy and effectiveness of intelligent governance. This article delineates the function creep of this new concept, identifying the individual and collective harms it engenders. A technological, contextual perspective of the function creep of AFR in policing will evidence the comprehensive creep of training datasets and learning algorithms, which have by-passed an ignorant public. We thus argue individual harms to dignity, privacy and autonomy, combine to constitute a form of cultural harm, impacting directly on individuals and society as a whole. While recognising the limitations of what the law can achieve, we conclude by considering options for redress and the creation of an enhanced regulatory and oversight framework model, or Code of Conduct, as a means of encouraging cultural change from prevailing police indifference to enforcing respect for the human rights violations potentially engaged. The imperative will be to strengthen the top-level design and technical support of AFR policing, imbuing it with the values implicit in the rule of law, democratisation and scientisation-to enhance public confidence and trust in AFR social governance, and to promote civilised social governance in AFR policing. (shrink)

Speech recognition by second language (L2) learners in optimal and suboptimal conditions has been examined extensively with English as the target language in most previous studies. This study extended existing experimental protocols ( Wang et al., 2013 ) to investigate Mandarin speech recognition by Japanese learners of Mandarin at two different levels (elementary vs. intermediate) of proficiency. The overall results showed that in addition to L2 proficiency, semantic context, F0 contours, and listening condition all affected the (...) recognition performance on the Mandarin sentences. However, the effects of semantic context and F0 contours on L2 speech recognition diverged to some extent. Specifically, there was significant modulation effect of listening condition on semantic context, indicating that L2 learners made use of semantic context less efficiently in the interfering background than in quiet. In contrast, no significant modulation effect of listening condition on F0 contours was found. Furthermore, there was significant interaction between semantic context and F0 contours, indicating that semantic context becomes more important for L2 speech recognition when F0 information is degraded. None of these effects were found to be modulated by L2 proficiency. The discrepancy in the effects of semantic context and F0 contours on L2 speech recognition in the interfering background might be related to differences in processing capacities required by the two types of information in adverse listening conditions. (shrink)

Norris, McQueen & Cutler claim that all known speech recognition data can be accounted for with their autonomous model, “Merge.” But this claim is doubly misleading. (1) Although speech recognition is autonomous in their view, the Merge model is not. (2) The body of data which the Merge model accounts for, is not, in their view, speech recognition data. Footnotes1 Author is also affiliated with the Center for the Study of Language and Information, Stanford (...) University, Stanford, CA 94305, [email protected]. (shrink)

The primary focus of this project is the silent and subvocal speech-recognition interface unveiled in 2018 as an ambulatory device wearable on the neck that detects a myoelectrical signature by electrodes worn on the surface of the face, throat, and neck. These emerge from an alleged “intending to speak” by the wearer silently-saying-something-to-oneself. This inner voice is believed to occur while one reads in silence or mentally talks to oneself. The artifice does not require spoken sounds, opening the (...) mouth, or any explicit or external movement of the lips. The essay then considers such subvocal “speech” as a mode of writing or saying and the interior of the mouth or oral cavity as its writing surface. It briefly revisits discussions of telepathy to recontextualize Heidegger’s warning against enframing language exclusively within calculative technics and physiology, which he suggests is detrimental to Mundarten (mouth-modes of regional dialects). It closes in reconsideration of Husserl’s phenomenology of language and meaning in Ideas as it might apply to subvocal speech-recognition interfaces. It suggests ways by which the electrophysiology that the device detects and deciphers (as an alleged intention of a presumed natural language unspoken vocally or aloud) might supplement Husserl’s insinuation of the Leiblichkeit of language through a self-stamping extraction of an extension of meaning. (shrink)

Vocabulary size has been suggested as a useful measure of “verbal abilities” that correlates with speech recognition scores. Knowing more words is linked to better speech recognition. How vocabulary knowledge translates to general speech recognition mechanisms, how these mechanisms relate to offline speech recognition scores, and how they may be modulated by acoustical distortion or age, is less clear. Age-related differences in linguistic measures may predict age-related differences in speech recognition (...) in noise performance. We hypothesized that speech recognition performance can be predicted by the efficiency of lexical access, which refers to the speed with which a given word can be searched and accessed relative to the size of the mental lexicon. We tested speech recognition in a clinical German sentence-in-noise test at two signal-to-noise ratios (SNRs), in 22 younger (18–35 years) and 22 older (60–78 years) listeners with normal hearing. We also assessed receptive vocabulary, lexical access time, verbal working memory, and hearing thresholds as measures of individual differences. Age group, SNR level, vocabulary size, and lexical access time were significant predictors of individual speech recognition scores, but working memory and hearing threshold were not. Interestingly, longer accessing times were correlated with better speech recognition scores. Hierarchical regression models for each subset of age group and SNR showed very similar patterns: the combination of vocabulary size and lexical access time contributed most to speech recognition performance; only for the younger group at the better SNR (yielding about 85% correct speech recognition) did vocabulary size alone predict performance. Our data suggest that successful speech recognition in noise is mainly modulated by the efficiency of lexical access. This suggests that older adults’ poorer performance in the speech recognition task may have arisen from reduced efficiency in lexical access; with an average vocabulary size similar to that of younger adults, they were still slower in lexical access. (shrink)

This article presents an overview of the single-channel dereverberation methods suitable for distant speech recognition application. The dereverberation methods are mainly classified based on the domain of enhancement of speech signal captured by a distant microphone. Many single-channel speech enhancement methods focus on either denoising or dereverberating the distorted speech signal. There are very few methods that consider both noise and reverberation effects. Such methods are discussed under a multistage approach in this article. The article (...) concludes with a hypothesis that the methods that do not require an a priori reverberation impulse response is desirable in varying the environmental conditions for DSR applications such as intelligent home and office environments, humanoid robots, and automobiles rather than the methods that require an a priori reverberation impulse response. (shrink)

Previous work demonstrated that dynamic pitch (i.e., pitch variation in speech) aids speech recognition in various types of noises. While this finding suggests dynamic pitch enhancement in target speech can benefit speech recognition in noise, it is of importance to know what noise characteristics affect dynamic pitch benefit and who will benefit from enhanced dynamic pitch cues. Following our recent finding that temporal modulation in noise influences dynamic pitch benefit, we examined the effect of (...) speech masker characteristics on dynamic pitch benefit. Specifically, the first goal of the study was to test the hypothesis that dynamic pitch benefit varies depending on the availability of pitch cues in the masker and the intelligibility of masker. The second goal of this study was to investigate whether older listeners as a group can benefit from dynamic pitch for speech recognition in speech maskers. In addition, individual factors of hearing loss and working memory capacity were examined for their impact on older listeners’ dynamic pitch benefit. Twenty-three younger listeners with normal hearing and 37 older listeners with varying levels of hearing sensitivity participated the study, in which speech reception thresholds were measured with sentences in speech maskers. While we did not find an effect of masker characteristics on dynamic pitch benefit, the results showed older listeners can benefit from dynamic pitch for recognizing speech in speech maskers. The data also suggest that among those older listeners with hearing loss, dynamic pitch benefit is stronger for individuals with higher working memory capacity. This can be attributed to their ability to exploit facilitated lexical access in processing of degraded speech signal. (shrink)

This paper combines domestic and international research results to analyze and study the difference between the attribute features of English phrase speech and noise to enhance the short-time energy, which is used to improve the threshold judgment sensitivity; noise addition to the discrepancy data set is used to enhance the recognition robustness. The backpropagation algorithm is improved to constrain the range of weight variation, avoid oscillation phenomenon, and shorten the training time. In the real English phrase sound (...) class='Hi'>recognition system, there are problems such as massive training data and low training efficiency caused by the super large-scale model parameters of the convolutional neural network. To address these problems, the NWBP algorithm is based on the oscillation phenomenon that tends to occur when searching for the minimum error value in the late training period of the network parameters, using the K-MEANS algorithm to obtain the seed nodes that approach the minimal error value, and using the boundary value rule to reduce the range of weight change to reduce the oscillation phenomenon so that the network error converges as soon as possible and improve the training efficiency. Through simulation experiments, the NWBP algorithm improves the degree of fitting and convergence speed in the training of complex convolutional neural networks compared with other algorithms, reduces the redundant computation, and shortens the training time to a certain extent, and the algorithm has the advantage of accelerating the convergence of the network compared with simple networks. The word tree constraint and its efficient storage structure are introduced, which improves the storage efficiency of the word tree constraint and the retrieval efficiency in the English phrase recognition search. (shrink)

In lieu of an abstract, here is a brief excerpt of the content:Philosophy, Psychiatry, & Psychology 10.2 (2003) 161-167 [Access article in PDF] Social, Cognitive, and Neural Constraints on Subjectivity and Agency:Implications for Dissociative Identity Disorder Peter Q. Deeley In this commentary, I consider Matthew's argument after making some general observations about dissociative identity disorder (DID). In contrast to Matthew's statement that "cases of DID, although not science fiction, are extraordinary" (p. 148), I believe that there are natural analogs of (...) the disorder that, when considered, make it seem less puzzling and exotic. After discussing these examples, I examine the relations between social, cognitive, and neural processes supporting subjectivity and agency, to identify sources of personal identity in cases of DID that can be used as a basis for assessing Matthew's points against Behnke and Sinnott-Armstrong. I am not proposing a comprehensive theory, but rather identifying what I believe are components of such a theory. I then outline an approach to determining criminal responsibility in DID, drawing on analogies with other psychiatric disorders. Dissociative Identity Disorder DID is a disorder described in DSM-IV, which "is characterised by the presence of two or more distinct identities or personality states that recurrently take control of the individual's behaviour accompanied by an inability to recall important personal information that is too extensive to be explained by ordinary forgetfulness" (American Psychiatric Association, p. 519). Recognizing Personal Identity How is personal identity recognized in normal cognition? If we take personhood to involve recognition of continuity of social, psychological, and physical identity for a given individual, then such judgments are routinely made rapidly and automatically as an aspect of what has come to be termed social cognition (Adolphs 2003). There is likely to be considerable redundancy in [End Page 161] the types of information used to make judgments about whether a person is the same as one who was previously encountered, or perhaps heard about—for example, facial features, movement patterns, speech prosody, dress, speech content, gender, ethnicity, social role, among other features, all provide information about individual identity, and hence personhood. When uncertainty exists, it is likely to reflect a loss or alteration of cues that would ordinarily be used in a rapid, automatic, unexamined way (Adolphs 2003). In such extreme cases (for example, severe mental illness or disfigurement), the problem of identification becomes the subject of explicit reasoning (and hence of potential disagreement, because we then engage in post hoc rationalizations to establish criteria for determining personhood, whereas normally we rely on essentially automatic, implicit abilities that work in most circumstances, some of which make little use of linguistic representations and formal reasoning—for example, individual facial identity recognition, which may occur as quickly as 170 ms after stimulus onset [Adolphs 2003]). Nevertheless, examining our judgments and intuitions in cases that force us to explicitly consider whether personal identity is preserved may provide analogies and clues for how we should proceed with DID. Let us consider, then, some natural analogs of DID. Natural Analogs of DID Method Acting I suggest that method acting is an analog of DID (Manderino 1985). Method acting refers to an approach to acting training and performance in which emphasis is placed on intensely imagining and identifying with the mental life of the character being portrayed. Training courses and manuals list many techniques and practices that can be used to facilitate absorption in a role, drawing on different kinds of memory—for example, experiential memories, memories of others, general knowledge (semantic memory), and memories of sensations and feelings, to creative a vivid sense of the subjective experience and actions of the character. Manderino, a method acting teacher, comments: "it is important to investigate the way in which your own mental makeup can improperly influence a character. Insecurities, self-consciousness, shyness or self-doubts can maintain a strong hold, and permitting any of these feelings to enter a performance is to remain the actor and not enter into the mental life of the character. For some, acting allows personal problems to be left behind, and they are thankful for the complete escape that... (shrink)

The installation error of metallurgical machinery parts is one of the common sources of errors in mechanical equipment. Because the installation error of different parts has different influences on different mechanical equipment, a simple linear formula cannot be used to identify the installation error. In the past, the manual recognition method and the touch recognition method lack of error information analysis, which leads to inaccurate recognition results. To improve the problem, an automatic recognition method based (...) on the neural network for metallurgical machinery parts installation error is proposed. According to the principle of automatic recognition of installation error based on the neural network, the nonlinear relation matrix between layers of the neural network is established. The operating state parameters of mechanical equipment are calculated, and the time series of the parameters are divided into several segments averagely. Based on the recognition algorithm, the inspection steps of depth, perpendicularity and center position of reserved hole, base board construction, short-circuit motor line and terminal installation, center mark board, and reference point installation are designed. The experimental results show that the recall rate of the proposed method is 97.66%, and the maximum absolute deviation is 0.09. The experimental data verify the feasibility of the proposed method. (shrink)

The first of its kind to thoroughly cover language technology at all levels and with all modern technologies this book takes an empirical approach to the ...

One particularly prominent strand of hate speech theory conceptualizes the harm in hate speech by considering the immediate illocutionary force of a hate speech ‘act’. What appears to be missing from such a conception, however, is how recognition relations and normative expectations present in a speech situation influence the harm such speech causes to its victims. Utilizing a particular real-world example, this paper illustrates how these defining background conditions and intersubjective relations influence the harm (...) of hate speech as it is experienced from the first-person perspective. This, more nuanced conception of harm, takes note of the effects such speech inflicts on an individual’s recognitive status, and can provide a clearer understanding of the conditions necessary for an act of hate speech to cause the speaker’s desired effect on their targets. (shrink)

Emotional facial expressions can inform researchers about an individual's emotional state. Recent technological advances open up new avenues to automatic Facial Expression Recognition (FER). Based on machine learning, such technology can tremendously increase the amount of processed data. FER is now easily accessible and has been validated for the classification of standardized prototypical facial expressions. However, applicability to more naturalistic facial expressions still remains uncertain. Hence, we test and compare performance of three different FER systems (Azure Face API, (...) Microsoft; Face++, Megvii Technology; FaceReader, Noldus Information Technology) with human emotion recognition (A) for standardized posed facial expressions (from prototypical inventories) and (B) for non-standardized acted facial expressions (extracted from emotional movie scenes). For the standardized images, all three systems classify basic emotions accurately (FaceReader is most accurate) and they are mostly on par with human raters. For the non-standardized stimuli, performance drops remarkably for all three systems, but Azure still performs similarly to humans. In addition, all systems and humans alike tend to misclassify some of the non-standardized emotional facial expressions as neutral. In sum, emotion recognition by automated facial expression recognition can be an attractive alternative to human emotion recognition for standardized and non-standardized emotional facial expressions. However, we also found limitations in accuracy for specific facial expressions; clearly there is need for thorough empirical evaluation to guide future developments in computer vision of emotional facial expressions. (shrink)

One of the strongest defences of free speech holds that autonomy requires the protection of speech. In this paper I examine five conditions that autonomy must satisfy. I survey recent research in social psychology regarding automatic behaviour, and a challenge to autonomy is articulated. I argue that a plausible strategy for neutralising some of the autonomy-threatening automatic responses consists in avoiding the exposure to the environmental features that trigger them. If this is so, we can good (...) autonomy-based pro tanto reasons for controlling exposure to certain forms of speech. (shrink)

The proposition that feedback is never necessary in speech recognition is examined for utterances consisting of sequences of words. In running speech the features near word boundaries are often modified according to language-dependent rules. Application of these rules during word recognition requires top-down processing. Because isolated words are not usually modified by rules, their recognition could be achieved by bottom-up processing only.