• Title/Summary/Keyword: 사용자 정의 기동어

Search Result 2, Processing Time 0.015 seconds

A study on user defined spoken wake-up word recognition system using deep neural network-hidden Markov model hybrid model (Deep neural network-hidden Markov model 하이브리드 구조의 모델을 사용한 사용자 정의 기동어 인식 시스템에 관한 연구)

  • Yoon, Ki-mu;Kim, Wooil
    • The Journal of the Acoustical Society of Korea
    • /
    • v.39 no.2
    • /
    • pp.131-136
    • /
    • 2020
  • Wake Up Word (WUW) is a short utterance used to convert speech recognizer to recognition mode. The WUW defined by the user who actually use the speech recognizer is called user-defined WUW. In this paper, to recognize user-defined WUW, we construct traditional Gaussian Mixture Model-Hidden Markov Model (GMM-HMM), Linear Discriminant Analysis (LDA)-GMM-HMM and LDA-Deep Neural Network (DNN)-HMM based system and compare their performances. Also, to improve recognition accuracy of the WUW system, a threshold method is applied to each model, which significantly reduces the error rate of the WUW recognition and the rejection failure rate of non-WUW simultaneously. For LDA-DNN-HMM system, when the WUW error rate is 9.84 %, the rejection failure rate of non-WUW is 0.0058 %, which is about 4.82 times lower than the LDA-GMM-HMM system. These results demonstrate that LDA-DNN-HMM model developed in this paper proves to be highly effective for constructing user-defined WUW recognition system.

Design for Database Retrieval System using Virtual Database in Intranet (인트라넷에서 가상데이터베이스를이용한 데이터베이스 검색 시스템의 설계)

  • Lee, Dong-Wook;Park, Young-Bae
    • The Transactions of the Korea Information Processing Society
    • /
    • v.5 no.6
    • /
    • pp.1404-1417
    • /
    • 1998
  • Currently, there exists two different methods for database retrieval in the internet. First is to use the search engine and the second is to use the plug-in or ActiveX technology, If a search engine, which makes use of indices built from keywords of simple text data in order to do a search, is used when accessing a database, first it is not possible to access more than one database at a time, second it is also not possible to support various conditional retrievals as in using query language, and third the set of data received might include many unwanted data, in other words, precision rate might be relatively low. Plug in or Active technology make use of Web browset to execute chents' query in order to do a database retrieval. Problems associated with this is that it is not possible to activate more than one DBMS simultaneously even if they are of the same data model. sefond it is not possible to execute a user query other than the ones thai arc previou sly defined by the client program In this paper, to resolve those aforementioned problems we design and implement database retrieval system using a virtual database, which makes it possible to provide direct query jntertacc through the conventional Web browser. We assume that the virtual database is designed and aggregated from more than one relational database using the same data model.

  • PDF