ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술대회 End-to-end Korean Digits Speech Recognition
Cited 1 time in scopus Download 0 time Share share facebook twitter linkedin kakaostory
저자
노종혁, 조관태, 김영삼, 조상래
발행일
201910
출처
International Conference on Information and Communication Technology Convergence (ICTC) 2019, pp.1137-1139
DOI
https://dx.doi.org/10.1109/ICTC46691.2019.8939697
협약과제
19HH3900, 고신뢰 지능정보 서비스에서 휴먼(H)-인프라(I)-서비스(S)를 연결하는 Portal Device 보안 기술 개발, 조상래
초록
The traditional speech recognition model consisting of an acoustic model and a language model is mainly used. Recently, an end-to-end speech recognition model consisting of a single integrated neural network model is being studied. This model has the advantage that it does not require a lot of training and it is easy to understand the structure of the model. In this paper, we designed the end-to-end model for Korean digit speech recognition and showed the performance results. We tried the digit speech recognition model in two forms: Word model and character model.
KSP 제안 키워드
End to End(E2E), End-to-End Speech Recognition, Integrated neural network, Language model, Recognition model, acoustic model, neural network model