Speech Transcription Research Articles

France, Implementation of the Israeli Nuclear Program and the Discussion on this Issue

.

An automated assessment framework for atypical prosody and stereotyped idiosyncratic phrases related to autism spectrum disorder

Autism Spectrum Disorder (ASD), a neurodevelopmental disability, has become one of the high incidence diseases among children. Studies indicate that early diagnosis and intervention treatments help to achieve positive longitudinal outcomes. In this paper, we focus on the speech and language abnormalities of young children with ASD and present an automated assessment framework in quantifying atypical prosody and stereotyped idiosyncratic phrases related to ASD. For detecting atypical prosody from speech, we propose both the hand-crafted feature based method as well as the end-to-end deep learning framework. First, we use the OpenSMILE toolkit to extract utterance level high dimensional acoustic features followed by a support vector machine (SVM) backend as the conventional baseline. Second, we propose several end-to-end deep neural network setups and configurations to model the atypical prosody label directly from the constant Q transform spectrogram of speech. Third, we apply cross-validation on the training data to perform segments selection and enhance the subject level classification performance. Fourth, we fuse the deep learning based methods with the conventional baseline at the score level to further enhance the overall system performance. For detecting the stereotyped idiosyncratic usage of words or phrases from speech transcripts, we adopt language model, dependency treebank and Term Frequency–Inverse Document Frequency (TF–IDF) in addition to Linguistic Inquiry and Word Count software (LIWC) methods to extract a set of text features followed by a standard SVM backend. We collect a database of spontaneous Mandarin speech recorded during the Autism Diagnostic Observation Schedule (ADOS) Module 2 and Module 3 sessions. The Module 2 part consists of 118 children while the Module 3 part includes 71 children. Experimental results on this database show that our proposed methods can effectively predict the atypical prosody and stereotyped idiosyncratic phrases codes for young children with the risk of ASD. On the two categories classification task, the unweighted accuracy of the aforementioned two tasks are 88.1% and 77.8%, respectively.

Gender and affiliation differences in topic selection in U.S. congressional speeches

Abstract The aim of this paper was to study gender differences in topic choice selection using the corpus of speeches given in the 113th United States Congress. We also looked at whether there are topic choice selection differences with respect to party affiliation and chamber, and finally, whether conversational topics chosen by male and female politicians correlate with any other category we measured in our corpus. The corpus was composed of 672 speeches by the female and 2,983 speeches by the male politicians. The speech transcripts were downloaded from the official repository Thomas and analyzed using the text analysis software Linguistic Inquiry and Word Count (LIWC) to identify the use of vocabulary related to seven conversational topics recorded by LIWC. The data was analyzed both quantitatively, using statistical analysis, and qualitatively, to determine if there are significant gender differences in speech topic selection. The analyses showed that there are overall gender and affiliation differences in topic selection by the male and female politicians in the 113th Congress, some confirming the trend of long-standing prevalence of home-related references in women’s speeches, and death and religion references in men’s speeches, others marking a social shift for some of the categories compared to previous studies on the topic, such as the increasing share of references to work, money achievement in women’s speeches, as well as women’s preference for security, and men’s preference for competitiveness, as signaled by their lexical choices. Further correlation test results recorded subtler differences which pointed to linguistic changes in stereotypization, such as women signaling less emotion and choosing more formal ways of expression.

Representative and Commissive Illocutionary Acts in Donald Trump’s Inauguration Speech

This paper is entitled Representative and Commissive Illocutionary Acts in Donald Trump’s Inauguration Speech. The aims of this study are to find out the classification and the intended meaning of the representative and commissive illocutionary act found in the speech. The research method used in the study was qualitative analysis method. The data were collected from the Donald Trump’s inauguration speech transcript and then identified the function of the illocutionary act and explained the intended meaning. Based on the analysis, the functions of the representative illocutionary act that were found are stating, predicting. affirming, describing and denying. Meanwhile, the function types of commissive illocutionary act are assuring, promising and pledging. The intended meanings that were found in the context of situation analysis are presenting, unifying, revealing, clarifying, convincing and committing.

Religiosity, emotional states, and strategy in the family firm: Edm. Schluter & Co Ltd., 1953-1980

Centré sur le cas de l’entreprise de négoce international de matières premières Edm. Schluter & Co. Ltd. dans la période 1953-1980, cet article contribue à un tournant affectif croissant dans l’histoire des entreprises et les études d’entreprises familiales en examinant comment la religiosité familiale partagée a généré des émotions positives qui ont créé un avantage concurrentiel pour cette firme pendant une période de turbulences externes accrues. À partir d’un éventail de sources qualitatives, y compris des lettres d’information commerciales et des transcriptions de discours prononcés par des dirigeants appartenant à la famille, nous identifions quatre émotions prédominantes : la nostalgie, l’anxiété, l’espoir et la confiance. Face à un environnement en rapide mutation caractérisé par la décolonisation de l’Afrique et la sécularisation de la Grande-Bretagne, l’entreprise s’est appuyée sur ces émotions pour construire sa résilience. Le commerce mondial des produits de base est dominé par les grandes sociétés. Le fait que cette petite entreprise familiale a survécu en gardant ses valeurs intactes jusqu’à ce jour est, à notre avis, la preuve que la famille, l’émotion et la religion ont fourni des ressources importantes à l’entreprise. L’article innove en appliquant l’histoire des émotions comme cadre d’interprétation de l’intersection à trois voies de la famille, des émotions et de la foi religieuse dans une entreprise familiale.

Bimodal classification of English allophones employing acoustic speech signal and facial motion capture

A method for automatic transcription of English speech into International Phonetic Alphabet (IPA) system is developed and studied. The principal objective of the study is to evaluate to what extent the visual data related to lip reading can enhance recognition accuracy of the transcription of English consonantal and vocalic allophones. To this end, motion capture markers were placed on the faces of seven speakers to obtain lip tracking data synchronized with the audio signal. 32 markers were used, 20 of which were placed on the speaker's inner lips and 4 on a special cap, which served as the point of reference and stabilized the FMC image while post-processing. Speech samples were simultaneously recorded as a list of approximately 300 words in which all English consonantal and vocalic allophones were represented. Different parameterization strategies were tested and the accuracy of vocalic segments recognition in different experiments was analyzed. The process of multimodal feature extraction is explained, the applied feature selection methods are presented and the obtained results are discussed. Further challenges related to the bi-modal feature extraction process and neural network-based decision systems employment are discussed. [Research sponsored by the Polish National Science Centre, Dec. No. 2015/17/B/ST6/01874.]

Context-aware result inference in crowdsourcing

Many result inference methods have been proposed to address the quality-control problem in crowdsourcing. However, existing methods are ineffective for context-sensitive tasks (CSTs), e.g., handwriting recognition, translation, speech transcription, where context correlation within a task cannot be ignored for two reasons. Firstly, it is ineffective to crowdsource a whole CST (e.g., recognizing handwritten texts) and use task-level inference methods to infer the answer, because it is rather hard to correctly complete a whole complicated task. Secondly, although a CST is composed of a set of atomic subtasks (e.g., recognizing a handwritten word), it is unsuitable to split it into multiple subtasks and adopt a subtask-level inference algorithm to infer the result, because this will lose the context correlation (e.g., phrases) among subtasks and increase the difficulty to complete a task. Thus it calls for a new approach to handling CSTs. In this work, we study the result inference problem for CSTs and propose a context-aware inference algorithm. We design an inference algorithm by incorporating the context information. Furthermore, we introduce an iterative method to improve the quality. The results of experiments on real-world CSTs demonstrated the superiority of our approach compared with the state-of-the-art methods.

An Analysis of Speech as a Modality for Activity Recognition during Complex Medical Teamwork.

We analyzed the nature of verbal communication among team members in a dynamic medical setting of trauma resuscitation to inform the design of a speech-based automatic activity recognition system. Using speech transcripts from 20 resuscitations, we identified common keywords and speech patterns for different resuscitation activities. Based on these patterns, we developed narrative schemas (speech "workflow" models) for five most frequently performed activities and applied linguistic models to represent relationships between sentences. We evaluated the narrative schemas with 17 new cases, finding that all five schemas adequately represented speech during activities and could serve as a basis for speech-based activity recognition. We also identified similarities between narrative schemas of different activities. We conclude with design implications and challenges associated with speech-based activity recognition in complex medical processes.

Analysing sport policy and politics: the promises and challenges of synthesising methodological approaches

ABSTRACTThis article reflects upon the potential analytical utility of synthesising contrasting methodological frameworks when conducting research on sport policy and politics, with specific reference to methods derived from the fields of critical discourse analysis, political discourse analysis and narrative analysis. With its central emphasis on the content and production of political discourse relating to the 2014 Glasgow Commonwealth Game sand the Scottish independence referendum, the methodological processes used in this thesis focused upon the identification of recurrent discursive forms from primary and secondary data sources including interviews with political actors, speech transcripts, manifestos, policy documents and press releases. Given the methodological synthesis evident in this approach, this article firstly provides a discussion of the paradigmatic assumptions of this analytical synthesis. Discussion then focuses on the provision of a more detailed outline of the frameworks of 1) Fairclough and Fairclough and 2) Somers, providing an explanation of the synthesis of these specific analytical frameworks within this research. Finally, this article provides a reflexive critique of the relative merits of the chosen methodology, with specific consideration given to the promises and challenges of synthesising analytical frameworks in a similar fashion in future research within the field of sport policy and politics.

Referential Cohesion in Donald Trump`s Speech Transcript

Cohesion establishes the text or sentences and gives their meaning by the grammatical and lexical linking. Halliday and Hasan (1976) In Cohesion in English, identify five general categories of cohesive devices that are possibly implemented in texts: reference, ellipsis, substitution, lexical cohesion and conjunction. One of the cohesion meanings is reference. Reference is one of cohesive devices, and it contains the specific nature of the information needed to be retrieved for the comprehensive information. The problems discussed in this study are the type of reference applied in the speech transcripts and referential meaning presented in the speech transcripts of Donald Trump`s Speech. The aims of the study were to find out the reference applied in the speech transcripts and referential meaning of the reference presented in the speech transcripts. The main theory used was Theory of Cohesion proposed by Halliday and Hasan (1976). The data in this study were taken from the website of Washington Post in form of transcripts. The data were collected using observation method and analyzed using qualitative method. There were two kinds of reference found in the speech; which were personal reference and demonstrative reference. The personal reference that occurred in the speech transcript included I, you, we, he, she, they, and it. Meanwhile, the demonstrative references that appeared in the speech transcript were this, these, there, here, now and the. These references were presented as endophoric, whether anaphoric or cataphoric. The comparative reference was absent. From this study, it can be concluded that referential cohesion takes an important role in a text or in this case speech transcripts, since the function of reference is to hold a text together and give its meaning.

Automatic meeting summarization and topic detection system

PurposeProducing meeting documents requires an instantaneous recorder during meetings, which costs extra human resources and takes time to amend the file. However, a high-quality meeting document can enable users to recall the meeting content efficiently. The paper aims to discuss these issues.Design/methodology/approachAn application based on this framework is developed to help the users find topics and obtain summarizations of meeting contents without extra effort. This app uses the Bluemix speech recognizer to obtain speech transcripts. It then combines latent Dirichlet allocation and a TextTiling algorithm with the speech script of meetings to detect boundaries between different topics and evaluate the topics in each segment. TextTeaser, an open API based on a feature-based approach, is then used to summarize the speech transcripts.FindingsThe results indicate that the summaries generated by the machine are 85 percent similar to the records written by humankind.Originality/valueTo reduce the human effort in generating meeting reports, this paper presents a framework to record and analyze meeting contents automatically by voice recognition, topic detection, and extractive summarization.

Automatic quantitative analysis of spontaneous aphasic speech

Spontaneous speech analysis plays an important role in the study and treatment of aphasia, but can be difficult to perform manually due to the time consuming nature of speech transcription and coding. Techniques in automatic speech recognition and assessment can potentially alleviate this problem by allowing clinicians to quickly process large amount of speech data. However, automatic analysis of spontaneous aphasic speech has been relatively under-explored in the engineering literature, partly due to the limited amount of available data and difficulties associated with aphasic speech processing. In this work, we perform one of the first large-scale quantitative analysis of spontaneous aphasic speech based on automatic speech recognition (ASR) output. We describe our acoustic modeling method that sets a new recognition benchmark on AphasiaBank, a large-scale aphasic speech corpus. We propose a set of clinically-relevant quantitative measures that are shown to be highly robust to automatic transcription errors. Finally, we demonstrate that these measures can be used to accurately predict the revised Western Aphasia Battery (WAB-R) Aphasia Quotient (AQ) without the need for manual transcripts. The results and techniques presented in our work will help advance the state-of-the-art in aphasic speech processing and make ASR-based technology for aphasia treatment more feasible in real-world clinical applications.

26.4 LANGUAGE DISTURBANCE AS A PREDICTOR OF PSYCHOSIS ONSET IN YOUTH AT ENHANCED CLINICAL RISK

BackgroundLanguage offers a privileged view into the mind; it is the basis by which we infer others’ thoughts. Subtle language disturbance is evident in schizophrenia prior to psychosis onset, including decreases in coherence and complexity, as measured using clinical ratings in familial and clinical high-risk (CHR) cohorts. Bearden et al previously used manual linguistic analysis of baseline speech transcripts in CHR to show that illogical and referential thinking, and poverty of content, predict later psychosis onset. Then, Bedi et al used automated natural language processing (NLP) of CHR transcripts to show that decreased semantic coherence and reduction in syntactic complexity predicted psychosis onset. To determine validity and reproducibility, we have applied automated NLP methods, with machine learning, to Bearden’s original CHR transcripts to identify a language profile predictive of psychosis.MethodsParticipants in the Bearden UCLA cohort include 59 CHR, of whom 19 developed psychosis (CHR+) within 2 years, whereas 40 did not (CHR-), as well as 16 recent-onset psychosis and 21 healthy individuals, similar in demographics; speech was elicited using Caplan’s “Story Game. Participants in the Bedi NYC cohort include 34 CHR (29 CHR+), with speech elicited using open-ended interview. Speech was audiotaped, transcribed, de-identified and then subjected to latent semantic analysis to determine coherence and part-of-speech tagging to characterize syntactic structure and complexity. A machine-learning speech classifier of psychosis onset was derived from the UCLA CHR cohort, and then applied both to the NYC CHR cohort and to the UCLA psychosis/control comparison, with convex hull (three-dimension depiction of model) and receiver operating characteristics analyses. Correlational analyses with demographics, symptoms and manual linguistic features were also done.ResultsA four-factor model language classifier derived from the UCLA CHR cohort that comprised three semantic coherence variables and one syntax (usage of possessive pronouns) predicted psychosis t with accuracy of 83% (intra-protocol) for UCLA CHR, 79% (cross-protocol) for NYC CHR, and 72% for discriminating psychosis from normal speech (UCLA psychosis/control). Convex hulls were defined as the smallest space containing all datapoints within a set for CHR- or healthy controls: these convex hulls showed substantial overlap, with CHR+ and psychosis speech datapoints largely outside these convex hulls. Coherence was associated with age, but speech variables did not vary by gender, race, or socioeconomic status in this study. While automated text features were unrelated to prodromal symptom severity, they were highly correlated with manual text features (r = 0.7, p < .000001).DiscussionIn this small preliminary study, we identified and cross-validated a robust language classifier of psychosis risk that comprised measures of semantic coherence (flow of meaning in language) and syntactic usage (usage of possessive pronouns). This classifier had utility in discriminating speech in individuals with recent-onset psychosis from the norm. It demonstrated concurrent validity in that it was highly correlated with manual linguistic features previously identified by Bearden et al, important as automated methods are fast and inexpensive. Automated language features were unrelated to sex, ethnicity or social class in these small samples, and semantic coherence increased with age, consistent with prior studies of normal language development. Of interest, overlapping convex hulls could be defined for groups of individuals without psychosis (UCLA CHR-, NYC CHR- and UCLA healthy), suggesting a constrained hull of normal language in respect to syntax and semantics, from which pre-psychosis and psychosis speech deviates. The RDoC linguistic corpus-based variables of semantic coherence and syntactic structure hold promise as biomarkers of psychosis risk and expression, with initial validation and reproducibility. Next steps in biomarker development include larger multisite studies with standardization of protocols for speech elicitation, test-retest, and attention to traction/feasibility, acceptability, cost, and utility. Mechanistic studies can also yield neural and physiological correlates of abnormal semantic coherence and syntax.

Textual Data Selection for Language Modelling in the Scope of Automatic Speech Recognition

The language model is an important module in many applications that produce natural language text, in particular speech recognition. Training of language models requires large amounts of textual data that matches with the target domain. Selection of target domain (or in-domain) data has been investigated in the past. For example [1] has proposed a criterion based on the difference of cross-entropy between models representing in-domain and non-domain-specific data. However evaluations were conducted using only two sources of data, one corresponding to the in-domain, and another one to generic data from which sentences are selected. In the scope of broadcast news and TV shows transcription systems, language models are built by interpolating several language models estimated from various data sources. This paper investigates the data selection process in this context of building interpolated language models for speech transcription. Results show that, in the selection process, the choice of the language models for representing in-domain and non-domain-specific data is critical. Moreover, it is better to apply the data selection only on some selected data sources. This way, the selection process leads to an improvement of 8.3 in terms of perplexity and 0.2% in terms of word-error rate on the French broadcast transcription task.

Automatic Online Lecture Highlighting Based on Multimedia Analysis

Textbook highlighting is widely considered to be beneficial for students. In this paper, we propose a comprehensive solution to highlight the online lecture videos in both sentence- and segment-level, just as is done with paper books. The solution is based on automatic analysis of multimedia lecture materials, such as speeches, transcripts, and slides, in order to facilitate the online learners in this era of e-learning - especially with MOOCs. Sentence-level lecture highlighting basically uses acoustic features from the audio and the output is implemented in subtitle files of corresponding MOOC videos. In comparison with ground truth created by experts, the precision is over 60 percent, which is better than baseline works and also welcomed by user feedbacks. On the other hand, segment-level lecture highlighting works with statistical analysis, mainly by exploring the speech transcripts, the lecture slides and their connections. With the ground truth created by massive users, an evaluation process shows that general accuracy can reach 70 percent, which is fairly promising. Finally, we also attempt to find potential correlation between these two types of lecture highlights.

EUPHEMISM IN DAVID CAMERON’S POLITICAL SPEECH IN ISIS ATTACKS

This study deals with euphemism in political speech of David Cameron about ISIS attacks. The objectives of the study were to find out the types of euphemsim used in David Cameron‟s speech on ISIS attacks and the reason why the most dominant types was choosen. The study was conducted by applying descriptive qualitative method.The data for this study were David Cameron political speech transcript in 20 July 2015, 14 November 2015, 2 December 2015, 5 february 2016. The source of the data was from United Kingdom goverment website . The data were analyzed using the theory of Allan and Burridge (1991). The findings of the study showed that the types of euphemism used in David Camerons political speech about ISIS attacks were: figurative expression (28,8%) ,circomlocution (15,5%), ommision (13,3%), ommision (13,3%) , hyperbole (11,1%), understatement (8,8%) , flippancy (8,8%), abbreviation (8,8%), borrowing (2,2%) and clipping (2,2%) . Figurative expression as the most dominant type used by David Cameron because it make words in speech more organized as a nice thing to be spoken in public in order to convince people.

EUPHEMISM IN DAVID CAMERON’S POLITICAL SPEECH IN ISIS ATTACKS

This study deals with euphemism in political speech of David Cameron about ISIS attacks. The objectives of the study were to find out the types of euphemsim used in David Cameron‟s speech on ISIS attacks and the reason why the most dominant types was choosen. The study was conducted by applying descriptive qualitative method.The data for this study were David Cameron political speech transcript in 20 July 2015, 14 November 2015, 2 December 2015, 5 february 2016. The source of the data was from United Kingdom goverment website . The data were analyzed using the theory of Allan and Burridge (1991). The findings of the study showed that the types of euphemism used in David Camerons political speech about ISIS attacks were: figurative expression (28,8%) ,circomlocution (15,5%), ommision (13,3%), ommision (13,3%) , hyperbole (11,1%), understatement (8,8%) , flippancy (8,8%), abbreviation (8,8%), borrowing (2,2%) and clipping (2,2%) . Figurative expression as the most dominant type used by David Cameron because it make words in speech more organized as a nice thing to be spoken in public in order to convince people.

Detection of Sentence Modality on French Automatic Speech-to-text Transcriptions

This article analyzes the detection of sentence modality in French when it is applied on automatic speech-to-text transcriptions. Two sentence modalities are evaluated (questions and statements) using prosodic and linguistic information. The linguistic features consider the presence of discriminative interrogative patterns and two log-likelihood ratios of the sentence being a question rather than a statement: one based on words and the other one based on part-of-speech tags. The prosodic features are based on duration, energy and pitch features estimated over the last prosodic group of the sentence. The classifiers based on linguistic features outperform the classifiers based on prosodic features. The combination of linguistic and prosodic features gives a slight improvement on automatic speech transcriptions, where the correct classification performance reaches 72%. A detailed analysis shows that small errors in the determination of the segment boundaries are not critical.

Classification-based financial markets prediction using deep neural networks

Deep neural networks (DNNs) are powerful types of artificial neural networks (ANNs) that use several hidden layers. They have recently gained considerable attention in the speech transcription and image recognition community (Krizhevsky et al., 2012 ) for their superior predictive properties including robustness to overfitting. However their application to algorithmic trading has not been previously researched, partly because of their computational complexity. This paper describes the application of DNNs to predicting financial market movement directions. In particular we describe the configuration and training approach and then demonstrate their application to backtesting a simple trading strategy over 43 different Commodity and FX future mid-prices at 5-minute intervals. All results in this paper are generated using a C++ implementation on the Intel Xeon Phi co-processor which is 11.4x faster than the serial version and a Python strategy backtesting environment both of which are available as open source code written by the authors.

Discourses of Thrift and Consumer Reasonability in Czech State-Socialist Society

The article examines how notions of thrift, saving, and frugality were present and active in the state-socialist discourses of economic behaviour and what meaning these notions carried. The research is based on three kinds of data: the official state-socialist public discourse of economic behaviour as presented in transcripts of parliamentary speeches, household guides and manuals, and eyewitness accounts of the state-socialist era recollected in oral history interviews. Such a multi-faceted corpus of discourse data made it possible to examine factual and normative aspects of thrift in state-socialist discourses and compare them with the accounts of everyday practices and tactics that may well contradict the official discourse. The analysis reveals that (a) notions of thrift and saving were strongly present throughout the period in all discourses examined, (b) both terms underwent a semantic shift from a productive to a restrictive meaning over time, and (c) both notions were eventually publicly sidelined by the emphasis on raising revenue. Despite fading from public discourse in the late 1980s, the notion of thrift had by then become instilled in the subjective understanding of the 'reasonable consumer', a concept that can therefore be considered a precursor of the contemporary concept of consumer responsibilisation.