• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.4
• /
• pp.240-248
• /
• 2014

We present a robust speaker identification algorithm that uses novel features based on soft bag-of-word representation and a simple Naive Bayes classifier. The bag-of-words (BoW) based histogram feature descriptor is typically constructed by summarizing and identifying representative prototypes from low-level spectral features extracted from training data. In this paper, we define a generalization of the standard BoW. In particular, we define three types of BoW that are based on crisp voting, fuzzy memberships, and possibilistic memberships. We analyze our mapping with three common classifiers: Naive Bayes classifier (NB); K-nearest neighbor classifier (KNN); and support vector machines (SVM). The proposed algorithms are evaluated using large datasets that simulate medical crises. We show that the proposed soft bag-of-words feature representation approach achieves a significant improvement when compared to the state-of-art methods.

• English Language & Literature Teaching
• /
• v.13 no.2
• /
• pp.129-153
• /
• 2007

When English language teachers are deciding on their priorities for teaching pronunciation, it is imperative to know what kind of differences and errors are most likely to interfere with communication, and what special problems particular first-language speakers will have with English pronunciation. In other words, phoneme discrimination skill is an integral part of speech processing for the EFL learners' learning to converse in English. Training using sound visualization technique can be effective in improving second language learners' perceptions and productions of segmental and suprasegmental speech contrasts. This study assessed the efficacy of a pronunciation training that provided visual feedback for EFL learners acquiring pitch and durational contrasts to produce and perceive English phonemic distinctions. The subjects' ability to produce and to perceive novel English words was tested in two contexts before and after training; words in isolation and words in sentences. In comparison with an untrained control group, trainees showed improved perceptual and productive performance, transferred their knowledge to new contexts, and maintained their improvement three months after training. These findings support the feasibility of learner-centered programs using sound visualization technique for English language pronunciation instruction.

• The Journal of the Acoustical Society of Korea
• /
• v.30 no.8
• /
• pp.428-435
• /
• 2011

Many isolated word recognition systems may generate irrelevant words for recognition results because they use only acoustic information or small amount of language information. In this paper, I propose word similarity that is used for selecting (or removing) less common words from candidates by applying Levenshtein distance. Word similarity is obtained by using positional accuracy that reflects the frequency information along to character's alignment information. This paper also discusses various improving techniques of selection of disparate words. The methods include different loss values, phone accuracy based on confusion information, weights of candidates by ranking order and partial comparisons. Through experiments, I found that the proposed methods are effective for removing heterogeneous words without loss of performance.

• Phonetics and Speech Sciences
• /
• v.7 no.3
• /
• pp.73-78
• /
• 2015

This study explores the degree of velarization of English word-final /l/ (i.e., /l/-darkness) according to the lexical status. Lexical status is defined as whether a speech stimulus is considered as a word or a non-word. We examined the temporal and spectral properties of word-final /l/ in terms of the duration and the frequency difference of F2-F1 values by varying the immediate pre-liquid vowels. The result showed that both temporal and spectral properties were contrastive across all vowel contexts in the way of real words having shorter [l] duration and low F2-F1 values, compared to non-words. That is, /l/ is more heavily velarized in words than in non-words, which suggests that lexical status whether language users encode the speech signal as a word or not is deeply involved in their speech production.

• Phonetics and Speech Sciences
• /
• v.2 no.2
• /
• pp.17-25
• /
• 2010

This study explores whether Koreans exploit their native coda-neutralization process when recognizing words in Korean continuous speech. According to the phonological rules in Korean, coda-neutralization process must come before the liaison process, as long as the latter(i.e. liaison process) occurs between 'words', which results in liaison-consonants being coda-neutralized ones such as /b/, /d/, or /g/, rather than non-neutralized ones like /p/, /t/, /k/, /ʧ/, /ʤ/, or /s/. Consequently, if Korean listeners use their native coda-neutralization rules when processing speech input, word recognition will be hampered when non-neutralized consonants precede vowel-initial targets. Word-spotting and word-monitoring tasks were conducted in Experiment 1 and 2, respectively. In both experiments, listeners recognized words faster and more accurately when vowel-initial target words were preceded by coda-neutralized consonants than when preceded by coda non-neutralized ones. The results show that Korean listeners exploit the coda-neutralization process when processing their native spoken language.

• Smart Media Journal
• /
• v.5 no.3
• /
• pp.30-34
• /
• 2016

Nowadays, there are a lot of Korean documents, which often need to be identified in one of printed or handwritten text. Early methods for the identification use structural features, which can be simple and easy to apply to text of a specific font, but its performance depends on the font type and characteristics of the text. Recently, the bag-of-words model has been used for the identification, which can be invariant to changes in font size, distortions or modifications to the text. The method based on bag-of-words model includes three steps: word segmentation using connected component grouping, feature extraction, and finally classification using SVM(Support Vector Machine). In this paper, bag-of-words model based method is proposed using SURF(Speeded Up Robust Feature) for the identification of machine printed and handwritten text in Korean documents. The experiment shows that the proposed method outperforms methods based on structural features.

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.94-99
• /
• 2010

This paper is to present how a network between words is formed. Not only have we looked at the distributivity, frequency and strength in connections between related words, but we have also presented some way to shed lights on what this network means to linguistic and social studies. The target source is morpho-analysis components of Trends 21 corpus which cover all newspaper articles from lour major newspapers, including Chosun, Joongang, Donga, and Hankyoreh, issued between 2000 and 2008. Based on nodes, links, and their connectivity indexes - density, degree, and centralizations, we have been able to retrieve and cluster related words forming the network with 20 event nouns. To reduce noise, we have considered the words whose t-score is above 1.64. By conducting both network and statistical analyses, we have presented the network of each event noun.

• Journal of the Semiconductor & Display Technology
• /
• v.18 no.4
• /
• pp.57-61
• /
• 2019

Nowadays, portable digital cameras such as smart phone cameras are being popularly used for entertainment and visual information recording. Given a database of geo-tagged images, a visual location recognition system can determine the place depicted in a query photo. One of the most common visual location recognition approaches is the bag-of-words method where local image features are clustered into visual words. In this paper, we propose a new bag-of-words-based visual location recognition algorithm using time-series streetview database. The proposed algorithm selects only a small subset of image features which will be used in image retrieval process. By reducing the number of features to be used, the proposed algorithm can reduce the memory requirement of the image database and accelerate the retrieval process.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.7
• /
• pp.135-141
• /
• 2019

The purpose of this study is to extract the key words from the list of research subjects related to 'hotel workers' published in recent 10 years(2009~2018) by using the language network analysis method and to confirm the relation between the key words. In this paper, we propose a semantic network analysis that can overcome limitations of longitudinal study, analyze the recent research trends, and widely use as a research model. The results of this study are as follows ; First, in analyzing major key words in the title of 'Hotel Employer' in recent 10 years, the major keyword of job satisfaction(40), special grade(26), organizational commitment(20), emotional labor(19), service(12), restaurant(10), and turnover intention(9). Second, we analyzed the relation of language network among major key words extracted from the study title of 'hotel workers'. Such a research process is expected to grasp the trends of research related to 'hotel workers' and give implications for the future direction of related research.

• Psychoanalysis
• /
• v.29 no.4
• /
• pp.72-82
• /
• 2018

What should a therapist do if the patient has lost the ability to speak in sessions? How should the therapist understand and approach this 'deadly silence'? Psychosomatic patients have poor ability to mentalize their experience, so they create a 'shunt' from emotional experience to body bypassing symbolic sphere. This makes psychosomatic patients vulnerable to expressing their emotions in words. With the case vignette 'When words are unspeakable: a bridge beyond the silence' written by Wirth in a book "From soma to symbol" we observe how psychosomatic pathology arises, from inappropriate mirroring and poor containment; and how this escalates into a vicious cycle of sin and punishment in the patient's conscience. This conscience scenario imprisons a talented and creative mind in a dark, silent place, wherein symbols and words are forbidden. This seemingly impossible task of providing silent patients with a 'talking cure' finds breakthrough with containment, and transformative thinking using therapists' reveries.