Downloads & Free Reading Options - Results

The report contains a discussion of artificial intelligence, computer processing of natural-language data programming languages for problem-solving systems, and structural (contextual, linguistic) pattern recognition. It includes an in-depth survey of recent research results that includes examples of deduction, text recognition, and dialogue by and with computers followed by discussion of speech data, applications of understand-software, and suggestions for further research.

Speech processing requires very efficient methods and algorithms. Finite-state transducers have been shown recently both to constitute a very useful abstract model and to lead to highly efficient time and space algorithms in this field. We present these methods and algorithms and illustrate them in the case of speech recognition. In addition to classical techniques, we describe many new algorithms such as minimization, global and local on-the-fly determinization of weighted automata, and efficient composition of transducers. These methods are currently used in large vocabulary speech recognition systems. We then show how the same formalism and algorithms can be used in text-to-speech applications and related areas of language processing such as morphology, syntax, and local grammars, in a very efficient way. The tutorial is self-contained and requires no specific computational or linguistic knowledge other than classical results.

المعالجة الآلية للغة

This presentation discusses two efforts of the NARI AI/ML Intern team during the Fall 2021 OSTEM Internship term. For Letters of Agreement (LoA), we have studied how LoAs are structured and explored the question ‘What is an LoA constraint?’ To do this, our approach is data-driven, iterative, and assisted by machine learning when available. In this presentation, we will walk through our tasks of manually scanning through documents, performing a preliminary entity labelling task, and our unsupervised analysis on LoA procedures sections. After this research phase, we define the smallest constraint unit in an LoA, and start to perform entity extraction. Looking towards constraint extraction, we are also exploring the use of a one-class support vector machine (OneClassSVM) model to identify patterns within the data.\n\nThe second effort of our team this term is focused on Air Traffic Control System Command Center (ATCSCC) advisory meetings, and the subsequent advisory documents that get published from their content. These advisory documents are important to give readily accessible summaries of daily operations, so that data centers, airline officials, and other stakeholders can easily understand the context of these meetings in real time. In applying machine learning to this scenario, two natural language processing tasks are used. First is developing machine learning models to convert the meeting speech data into text. With this text, use of extractive and abstractive text summarization models are used to automatically generate preliminary versions of the advisory documents.

This presentation discusses two efforts of the NARI AI/ML Intern team during the Fall 2021 OSTEM Internship term. For Letters of Agreement (LoA), we have studied how LoAs are structured and explored the question ‘What is an LoA constraint?’ To do this, our approach is data-driven, iterative, and assisted by machine learning when available. In this presentation, we will walk through our tasks of manually scanning through documents, performing a preliminary entity labelling task, and our unsupervised analysis on LoA procedures sections. After this research phase, we define the smallest constraint unit in an LoA, and start to perform entity extraction. Looking towards constraint extraction, we are also exploring the use of a one-class support vector machine (OneClassSVM) model to identify patterns within the data.\n\nThe second effort of our team this term is focused on Air Traffic Control System Command Center (ATCSCC) advisory meetings, and the subsequent advisory documents that get published from their content. These advisory documents are important to give readily accessible summaries of daily operations, so that data centers, airline officials, and other stakeholders can easily understand the context of these meetings in real time. In applying machine learning to this scenario, two natural language processing tasks are used. First is developing machine learning models to convert the meeting speech data into text. With this text, use of extractive and abstractive text summarization models are used to automatically generate preliminary versions of the advisory documents.

This volume contains the 12 papers, presented in 4 sessions at the Information Systems Technology Panel Workshop held in Aalborg, Denmark on 8th September 2001. The papers presented covered the following headings: * N4 Corpus and Speaker Identification * Non-Native Speech * Speech Recognition Language Identification and Multilingual Applications.

This volume contains the 12 papers, presented in 4 sessions at the Information Systems Technology Panel Workshop held in Aalborg, Denmark on 8th September 2001. The papers presented covered the following headings: * N4 Corpus and Speaker Identification * Non-Native Speech * Speech Recognition Language Identification and Multilingual Applications.

This volume contains the 12 papers, presented in 4 sessions at the Information Systems Technology Panel Workshop held in Aalborg, Denmark on 8th September 2001. The papers presented covered the following headings: * N4 Corpus and Speaker Identification * Non-Native Speech * Speech Recognition Language Identification and Multilingual Applications.

This volume contains the 12 papers, presented in 4 sessions at the Information Systems Technology Panel Workshop held in Aalborg, Denmark on 8th September 2001. The papers presented covered the following headings: * N4 Corpus and Speaker Identification * Non-Native Speech * Speech Recognition Language Identification and Multilingual Applications.

Transfer learning is a vital technique that generalizes models trained for one setting or task to other settings or tasks. For example in speech recognition, an acoustic model trained for one language can be used to recognize speech in another language, with little or no re-training data. Transfer learning is closely related to multi-task learning (cross-lingual vs. multilingual), and is traditionally studied in the name of `model adaptation'. Recent advance in deep learning shows that transfer learning becomes much easier and more effective with high-level abstract features learned by deep models, and the `transfer' can be conducted not only between data distributions and data types, but also between model structures (e.g., shallow nets and deep nets) or even model types (e.g., Bayesian models and neural models). This review paper summarizes some recent prominent research towards this direction, particularly for speech and language processing. We also report some results from our group and highlight the potential of this very interesting research field.

An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition

This OSF repository contains the data and code for the the study "Effects of speech rate on anticipatory eye movements in the Visual World Paradigm: Evidence from aging, native, and non-native language processing" (accepted at the Quarterly Journal of Experimental Psychology). The repository contains two data files (.csv) containing the accuracy data (Fernandez_et_al_accuracy_data.csv) and the eye movement data (Fernandez_et_al_speech_rate_data.csv) and one R script providing the code used for statistical analysis in the accepted article (Fernandez_et_al_code_speech_rate.R). See the .R file for additional information and descriptions of the variables in the .csv files.

The infant brain is not fully matured at birth. The newborn EEG is initially dominated by slow electrophysiological activity and accelerates across infancy and early childhood. This acceleration is observed as increased (relative) power in high frequencies as well as an increase of the individual alpha frequency (IAF) across infancy (e.g., Schaworonkov & Voytek, 2021). It has previously been hypothesized, that the trajectory of neural maturation—from slow to fast—shapes early language acquisition by determining which part of the speech signal infants can process and learn from (Menn et al., 2023a; Snijders & Menn, in press). However, direct empirical evidence for this proposal is currently lacking. The current study will assess the developmental trajectory of maturation of the EEG across early childhood and link individual differences in neural maturation to infants’ speech processing as well as the emergence of acoustic-phonological and higher linguistic knowledge. We expect to see that developmental changes in EEG markers are driven by biological maturation and that individual differences in neural maturation relate to speech processing as well as language abilities. Electrophysiological correlates of acoustic speech processing will be assessed via neural tracking of prosody, syllables and phonemes as well as via neural encoding of the spectrogram. Language abilities will be assessed via neural measures of acoustic-phonological knowledge through EEG encoding models, the EEG measures of word onset responses, as an index for word segmentation, and behavioral assessments of vocabulary as a measure for lexical-semantic knowledge.

Good-enough approach to language comprehension assumes that people do not always engage in full detailed processing of linguistic input. Rather, the parser forms shallow representations when confronted with some difficulty such as complex syntactic structure or noisy input. Although the good-enough approach has been studied for some time, we still do not know what factors trigger this type of processing. In this study, we investigate two factors that might influence the reliance on the good-enough language processing strategy in oral speech perception — age and noise.

A range of deficits can cause children difficulty understanding speech in challenging listening environments, like noisy classrooms. Children with listening difficulties are at risk of having poor long-term academic outcomes and social skills (Barry et al., 2015), especially when clinicians cannot detect or remediate their specific deficits. Deficits in auditory, speech, language, or cognition abilities may present in similar manners on the individual child. Currently, it is difficult to determine the cause of these difficulties. A systematic approach to differentiate between these causes has been devised (Dillon & Cameron, 2021). The proposed approach is a tri-level test battery, a combination of top-level speech perception ability, mid-level phoneme identification ability, and low-level acoustic resolution ability (Dillon & Cameron, 2021). In conjunction with cognitive test scores, the combined approach will allow for differentiation of the cause of the observed listening deficit. A richer understanding of the underlying reasons for listening difficulties is central to developing customized efficacious interventions to avoid and alleviate long-term consequences.