• Title/Summary/Keyword: Language Adaptation

Search Result 181, Processing Time 0.02 seconds

Integration of WFST Language Model in Pre-trained Korean E2E ASR Model

  • Junseok Oh;Eunsoo Cho;Ji-Hwan Kim
    • KSII Transactions on Internet and Information Systems (TIIS)
    • /
    • v.18 no.6
    • /
    • pp.1692-1705
    • /
    • 2024
  • In this paper, we present a method that integrates a Grammar Transducer as an external language model to enhance the accuracy of the pre-trained Korean End-to-end (E2E) Automatic Speech Recognition (ASR) model. The E2E ASR model utilizes the Connectionist Temporal Classification (CTC) loss function to derive hypothesis sentences from input audio. However, this method reveals a limitation inherent in the CTC approach, as it fails to capture language information from transcript data directly. To overcome this limitation, we propose a fusion approach that combines a clause-level n-gram language model, transformed into a Weighted Finite-State Transducer (WFST), with the E2E ASR model. This approach enhances the model's accuracy and allows for domain adaptation using just additional text data, avoiding the need for further intensive training of the extensive pre-trained ASR model. This is particularly advantageous for Korean, characterized as a low-resource language, which confronts a significant challenge due to limited resources of speech data and available ASR models. Initially, we validate the efficacy of training the n-gram model at the clause-level by contrasting its inference accuracy with that of the E2E ASR model when merged with language models trained on smaller lexical units. We then demonstrate that our approach achieves enhanced domain adaptation accuracy compared to Shallow Fusion, a previously devised method for merging an external language model with an E2E ASR model without necessitating additional training.

Difficulties of immigrant women through international marriage - in Raising Children and Needs for Adaptation Program - (여성결혼이민자의 자녀양육 어려움과 적응 프로그램 요구도)

  • Hyun, Eun-Ae;Rha, Jong-Hay
    • Korean Journal of Human Ecology
    • /
    • v.18 no.3
    • /
    • pp.675-687
    • /
    • 2009
  • The purpose of this study is to investigate the difficulties and needs for adaptation program of married immigrant women raising children in Korea. Data were collected from 27 Day care centers in Guemsan County, Chungnam Province. 20 senior teachers of day care centers and 58 mothers who had immigrated into Korea by international marriage were asked to complete a packet of questionnaire designed by researchers for this study, and in-depth interviews were conducted to 20 senior teachers and 6 mothers who could communicate in Korean. Data were analyzed by using frequency analysis, one way ANOVA, and cross-section analysis. The outcome of this study showed as follows: 1) The number of entire married immigrant women's children attending day care centers at Guemsan area was 102 persons. 2) Immigrant mothers felt that the lack of their understanding Korean culture makes them difficult in raising children in Korea, whereas the day care teachers felt that children's language development was delayed. 3) According to the survey of adaptation program a for married immigrant women's children, there was few program at the time of survey, only two daycare centers conducted multi-cultural education and Korean language education once or twice a year. Teachers required Korean language education, grievance counseling (?) for the mothers. Immigrant mothers required multi-cultural education and Korean language education.

A Research on the Control Effect of Teacher's Support in the Course of School Adaptation for Children from Multi-cultural Families (다문화가정 아동의 학교생활적응 과정에서 교사지지의 조절효과에 관한 연구)

  • Rho, Seon-Deok;Kang, Gil-Hyun
    • Journal of the Korea Society of Computer and Information
    • /
    • v.20 no.3
    • /
    • pp.157-167
    • /
    • 2015
  • In this study, we studied the influence that the stress placed on children (originating from presence of two cultures at multi-cultural families) can have on adapting to school, and further assessed the control effect of teachers' support as a protection factor. In order to achieve the goal of the research, we identified samples of children between Grades 3-6 who came from multi-cultural families to carry out surveys. Based on the results of our analysis, the experience of discrimination and language conflicts experienced by children from multi-cultural families affected their academic activities (among adaptation factors to school experiences), while teachers' support appeared to have a control effect. However, for the influence of language conflicts on friend relationships (among adaptation factors to school experiences), teachers' support did not have a control effect. As such, based on this research result, we've proposed plans to improve school adaptation of children from multicultural families.

A Comparative Study on the Occurrence Loci of Disfluency between Neurogenic and Developmental Stuttering (신경인성과 발달성 말더듬의 비유창성 발생 자리에 대한 연구)

  • Shin, Myung-Sun;Kwon, Do-Ha;Yoon, Chi-Yeon
    • Speech Sciences
    • /
    • v.13 no.3
    • /
    • pp.185-195
    • /
    • 2006
  • This study aims to clarify disfluency loci in a neurogenic stuttering group and to examine how the characteristics are different from a developmental stuttering group. For the study, spoken language samples were collected from 11 adults with developmental stuttering and 11 adults with neurogenic stuttering in the course of speaking tasks including reading, monologue and conversation. Using the collected samples, disfluency characteristics of the two groups were to be investigated by analyzing adaptation effect, consistency effect and frequency of disfluency occurrence according to word position, which are related to the occurrence loci of disfluency. Results of this study were as follows: First, while the neurogenic stuttering group did not show any adaptation effect, the developmental stuttering group showed the adaptation effect that the percent of disfluency word reducing as they read the same materials repeatedly. Second, there was no meaningful difference of consistency effect between the two stuttering groups. Third, the neurogenic stuttering group showed more disfluency frequency in final sounds among the word position compared to the developmental stuttering group.

  • PDF

Transcending Cultural Boundaries: A Study of the Adaptation of Shakespeare's Othello by Vishal Bharadwaj

  • Roychowdhury, Iti
    • Cross-Cultural Studies
    • /
    • v.28
    • /
    • pp.55-66
    • /
    • 2012
  • Ever since they were first produced, more than four hundred years ago, Shakespeare's plays have been reproduced and adapted into countless film and TV productions, into ballets and operas and theatre performances across the globe. The present paper, within a broad conceptual framework, aims to investigate the cross cultural dimension of adaptation of a stage play, written for the Renaissance England, into a $21^{st}$centuryIndianfeaturefilm. The paper uses Omkara, an adaptation of Othello by Vishal Bharadwaj, as a case study to: (i.) Explore the use of the idiomatic language of cinema in such an adaptation. (ii.) Posit a re-reading of Shakespeare with the help of local/native signifiers.

Language Model Adaptation for Broadcast News Recognition (방송 뉴스 인식을 위한 언어 모델 적응)

  • Kim Hyun Suk;Jeon Hyung Bae;Kim Sanghun;Choi Joon Ki;Yun Seung
    • MALSORI
    • /
    • no.51
    • /
    • pp.99-115
    • /
    • 2004
  • In this parer, we propose LM adaptation for broadcast news recognition. We collect information of recent articles from the internet on real time, make a recent small size LM, and then interpolate recent LM with a existing LM composed of existing large broadcast news corpus. We performed interpolation experiments to get the best type of articles from recent corpus because collected recent corpus is composed of articles which are related with test set, and which are unrelated. When we made an adapted LM using recent LM with similar articles to test set through Tf-Idf method and existing LM, we got the best result that ERR of pseudo-morpheme based recognition performance has 17.2 % improvement and the number of OOV has reduction from 70 to 27.

  • PDF

A comparative Study of English Loans in Russian and Swahili

  • Dzahene-Quarshie, Josephine;Csajbok-Twerefou, Ildiko
    • Cross-Cultural Studies
    • /
    • v.24
    • /
    • pp.99-111
    • /
    • 2011
  • This paper is a comparative study of English loans in Russian and Swahili. In the twenty first century, due to the advantage of English as a global language, a language of technology and business, it has had contact with many languages of the world and has become a major source of loans to many languages. Though very different from each other, both Russian and Swahili currently have English as their main source of loanwords. This study reports the extensive adaptation of English loans by Russian and Swahili and examines how these loan items are assimilated into the two languages. It concludes that besides the adaption of pure English loans they have both employed other strategies such as loan translations, semantic extensions and loanblends for vocabulary expansion.

Poly-encoder based COVID-19 Question and Answering with Task Adaptation (Poly-encoder기반의 COVID-19 질의 응답 태스크)

  • Lee, Seolhwa;Lim, Heuiseok
    • Annual Conference on Human and Language Technology
    • /
    • 2020.10a
    • /
    • pp.188-191
    • /
    • 2020
  • 본 연구는 COVID-19 질의 응답 태스크를 위한 Poly-encoder 기반의 태스크를 제안하였다. COVID-19 질의 응답 시스템은 사람들에게 최신 정보에 대해 빠르고 신뢰성이 높은 정보를 전달하는 특성을 가져야한다. 검색 기반 질의 응답 시스템은 pairwise 연산을 기반으로 수행되는데, Poly-encoder는 사전 학습된 트랜스포머(transformer)기반의 pairwise 연산 방법론 중 기존 Cross-encoder와 Bi-encoder보다 실사용 및 성능이 뛰어남을 보였다 [1]. 특히, Poly-encoder는 정확도가 높으면서도 빠른 응답속도를 가지며 검색기반의 각종 태스크에서 좋은 성능을 보였다. 따라서 본 연구는 COVID-19를 위한 Poly-encoder기반의 질의 응답 태스크를 위하여 기존 질의 응답 태스크와 페르소나 기반의 질의 응답 태스크로 두 가지 유형의 태스크를 생성하여 모델을 학습하였다. 또한 신뢰성 있는 리소스정보로부터 모델에 최신 정보 반영을 위하여 자동 크롤러를 구축하여 데이터를 수집하였다. 마지막으로 전문가를 통한 데이터셋을 구축하여 질문-응답과 질의어-질문에 대한 모델 검증을 수행하였다.

  • PDF

Effects of Forest Healing Program on Cultural Adaptation Stress and Psychological Wellbeing of Foreign Workers

  • Kim, Han-Won;Shin, Won-Sop;Jang, Cheoul-Soon
    • Journal of People, Plants, and Environment
    • /
    • v.22 no.5
    • /
    • pp.505-513
    • /
    • 2019
  • As the number of foreign workers increases in Korea, people's interest in their adaptation to this society has also increased. The purpose of this study was to investigate the effect of a forest-healing program on foreign workers' cultural adaptation stress and psychological wellbeing. The experiment was conducted in a recreation forest located in Hongcheon-gun, Gangwon-do. The subjects who participated in this study included 40 foreign workers (16 males, 24 females). The experiment was conducted for three days and two nights from June 26 to 28, 2019. The forest healing program was composed of activities that can enhance immunity and restore physical and mental health by utilizing various healing environmental factors such as landscapes, sunlight, anion and oxygen in forest. The independent sample t-test was conducted to examine changes in cultural adaptation stress and psychological well-being before and after the program. The collected data were analyzed using SPSS 18.0 program. As a result, cultural adaptation stress was statistically significantly reduced in sub-factors including discrimination experience, language conflict and legal status. Psychological wellbeing improved in all the surveyed sub-factors such as self-confidence, ego, immersion, and joy. The results of this study are expected to be used as basic data for reducing the stress foreign workers experience from cultural adaptation and restoring their psychological wellbeing in the future.