선형예측계수에 기초한 퍼지추론 단어 인식

최승호,김형근,Choi, Seung-Ho,Kim, Hyeong-Geun 한국음향학회 1994 韓國音響學會誌 Vol.13 No.1

LPC열로 구성된 음성패턴의 주파수변동을 해결하기위해 LPC와 스펙트럼, LPC차수와 스펙트럼의 관계를 고찰하여 새로운 형태의 멤버쉽함수를 제안하였다. 또한, 시간변동을 해결하기위해서는 음성구간을 여러구간으로 등간격분할하는 다구간 등분할법을 사용하였으며, 이때 오인식은 주로 동일음절이 같은 발성위치에 있을때 발생되었다. 이러한 오인식을 줄이기위해 제안된 멤버쉽함수로 퍼지추론한뒤 구간별 확신도에 가중치를 부여하고, 세번째후보까지를 인식대상으로 하는 판정알고리즘을 제안하였다. 본 방법의 타당성을 검증하기위해, DDD지역명 28개를 대상으로 인식실험한결과, 삼각형멤버쉽함수에 의한 퍼지추론은 $92.0\%$, 삼각형멤버쉽함수에의한 퍼지추론과 판정알고리즘은 $92.9\%$, 제안된 멤버쉽함수에의한 퍼지추론과 판정알고리즘은 $93.8\%$의 인식률을 보였다. To solve the frequency variation of speech patterns which consist of LPC sequences, new membership function view from LPC, spectrum and the relations between the order of LPC and spectrum is proposed. To solve the time variation, multi-secation equi-segmentation method which equally divide the speech section into several section are applied. False recognition mainly occur at time when the same syllable is placed at the same utterance. To reduce the error, fuzzy inference is executed using the proposed membership function and weights are assigned into sectional certainty and then the decision method for recognized the section up to the third candidate. To testify the validation of this method, we experimented the recognition test of 28 DDD area names. The recognition rate of the fuzzy inference by the triangle membership function is $92\%$. That of the combined method of the fuzzy inference and the dicision method is $92.9\%$ and that of fuzzy inference by the proposed membership funtion is $93.8\%$.

후두골격수술

최승호,권민수,Choi, Seung-Ho,Kwon, Minsu 대한후두음성언어의학회 2013 대한후두음성언어의학회지 Vol.24 No.2

Laryngeal framework surgery (LFS) is a unique phonosurgical concept that enables us to influence the laryngeal biomechanics by changing the shape/position of the laryngeal cartilages. LFS procedures can be favorably combined with one another but also with other phonosurgical methods, and they are usually reversible and correctable. Type I thyroplasty and arytenoid adduction are still useful in spite of the recent popularity of injection laryngoplasty. Basic surgical principles have seldom been changed since Isshiki's development, but a number of modifications have been tried and are still going on. These delicate surgeries require exhaustive training, but the reward is great to both the surgeon and the patient.

위험관리정보 2 - 중국 상해의 성능위주 화재안전설계 적용

최승호,Choi, Seung-Ho 한국화재보험협회 2009 防災와 保險 Vol.132 No.-

피부의 특성 추출을 위한 가시광선 반사 스펙트럼의 역 추적 최적화 알고리즘 개선

최승호,임창환,정병조,Choi, Seung-Ho,Im, Chang-Hwan,Jung, Byung-Jo 한국광학회 2007 한국광학회지 Vol.18 No.3

In order to extract more accurate skin parameters, this study was focused on the improvement of the efficiency of a previous inverse fitting algorithm based on genetic algorithms. The algorithm provides the best fitting result of the diffusion approximation model to a VRS (visual reflectance spectroscopy) curve of skin. Simplex and wavelength selection methods were applied to the previous algorithm. Nine skin parameters were inversely extracted from the modeling studies. The revised inverse fitting algorithm was determined to produce an 83% reduction of computation time and a 0.64% reduction of sum of square error, compared to the previous algorithm. In conclusion, we confirmed that the new algorithm provides faster and more accurate solutions for the diffusion approximation model. 본 논문은 가시광선 반사 스펙트럼으로부터 피부조직의 광학적, 구조적 특성을 역 추출하는 알고리즘에 대한 연구를 다룬다. 본 연구에서는 유전알고리즘을 사용한 기존의 역 추적 최적화 알고리즘에 파장 선택법과 심플렉스법을 추가하여 알고리즘의 최적화 능력을 개선하였다. 파장 선택법은 기존 알고리즘에 사용된 135개의 파장 중 10개의 유효 파장만을 사용한 결과 계산시간이 83% 감소하였다. 심플렉스법을 사용해 유전알고리즘으로 얻은 근사적인 탐색 점으로부터 재 탐색을 수행하였고, 이를 통해 얻은 탐색 점의 SSE(sum of square error)는 기존 유전알고리즘 만으로 얻은 탐색 점이 가지는 SSE의 0.64%였다. 본 연구의 모델링 결과를 통해 개선된 알고리즘이 기존 알고리즘에 비해 빠르고 정확한 해를 찾는다는 것을 확인할 수 있었다.

독립 성분 분석과 스펙트럼 향상에 의한 잡음 환경에서의 음성인식

최승호,Choi Seung-Ho 대한음성학회 2003 말소리 Vol.48 No.-

In this paper, we propose a speech recognition method based on independent component analysis (ICA) and spectral enhancement techniques. While ICA tris to separate speech signal from noisy speech using multiple channels, some noise remains by its algorithmic limitations. Spectral enhancement techniques can compensate for lack of ICA's signal separation ability. From the speech recognition experiments with instantaneous and convolved mixing environments, we show that the proposed approach gives much improved recognition accuracies than conventional methods.

He-Ne 레이저를 이용한 표면전단응력 측정에 관한 연구

최승호,이열,Choi, Seung-Ho,Lee, Yeol 대한기계학회 1997 大韓機械學會論文集B Vol.21 No.7

An experimental study of the skin friction measurement in a turbulent boundary-layer has been carried out. The skin friction measurements are made using the laser interferometer skin friction (LISF) meter, which optically detects the rate of thinning of an oil applied to the test surface. This technique produces reliable skin friction data over a wide range of flow situations up to 3-dimensional complicated flows with separation, where traditional skin friction measurement techniques are not applicable. The present measured data in a turbulent boundary-layer on a flat plate using the LISF technique shows a good comparison with the result from the previous velocity profile techniques, which proves the validity of the present technique. An extensive error analysis is carried out for the present technique yielding an uncertainty of about .+-.8%, which makes them suitable for CFD code validation purposes. Finally the measurements of the skin friction in a separated region after a surface-mounted obstacle are also presented.

Binar Convolution을 이용한 고속 디지탈 신경회로망의 VLSI 설계

최승호,김영민,Choi, Seung-Ho,Kim, Young-Min 한국음향학회 1996 韓國音響學會誌 Vol.15 No.5

현재 신경회로망의 구현에 관한 여러 가지 연구가 진행되고 있으며, 이들 중 신경회로망의 VLSI 구현에 대한 연구가 매우 활발하다. 디지털 신경회로망은 느린 처리속도와 넓은 면적을 차지하는 점이 주요 단점으로 지적되는데 본 논문에서는 neural cell을 곱셈과 덧셈을 Binary Convolution 기법과 Counter를 사용하여 설계함으로써 속도를 높이고 단위 뉴런의 소요 Tr수를 줄여 그 소요 면적을 줄이도록 하였다. 본 cell의 구조를 이용하여 layer당 16개씩의 cell을 가지는 3-layer neural network을 구성하였을 경우 0.8${\mu}$ standard cell 설계시 50MHz까지 동작하였으며 26MCPS의 동작을 확보하였다. Recently, for implementation of neural networks extensive studies have been done especially VLSI technology has been regarded as the one of the most attractive means to implement neural networks. The main drawbacks of digital VLSI implementations are their large area and slow processing speed. In this paper to solve the speed and size problems we designed the efficient architecture using the binary convolution method for basic operation of neural cell, multiplication and addition. When it is used for implementing 3-layer network with 16 neural cell per layer that used neural cell based on binary convolution, clock of 50MHz and 26MCPS on 0.8${\mu}$ standard cell library has been achieved.

인후두역류 관련 연구의 현황 및 전망

최승호,Choi, Seung-Ho 대한후두음성언어의학회 2007 대한후두음성언어의학회지 Vol.18 No.2

Although the prevalence of laryngopharyngeal reflux (LPR) is presumed to be very high, basic researches on LPR have not been active so far. This is a review of present status and prospect of researches on LPR to help readers' understanding of current trend on LPR research. In vitro and in vivo researches on acid/pepsin, intercellular junctions, bile juice, carbonic anhydrase, cytokines, and Helicobacter pylori were discussed. Considering that quality of life is getting more important in accordance with longer life span and that a fair proportion of LPR patients do not respond to conventional therapy, researches on pathogenesis and pathophysiology of treatment measures are necessary and it is anticipated for more and more researchers to concern themselves in basic research on LPR in the future.

정규화신뢰도 기반 가변어휘 고립단어 인식기의 거절기능 성능 분석

최승호,Choi, Seung-Ho 한국음향학회 2006 韓國音響學會誌 Vol.25 No.2

고립단어 인식기의 오 인식 단어를 거절하기 위한 방법으로 정규화 신뢰도가 제안되어 논문 [1-2]에서 성공적으로 적용된 바 있다. 그러나 정규화 신뢰도의 성능 측정을 위해 고정된 단어 셌을 대상으로 실험을 하였다. 본 논문에서는 정규화 신뢰도를 가변어휘 음성인식 영역에 적용하여 신뢰도의 거절성능을 밝히고 특히, 벡터양자화기를 이용하여 미 출현 트라이 폰의 문제를 극복하는 방법을 제안한다. 이때 정규화 신뢰도는 트라이 폰 신뢰도들의 통계적 특징(평균과 표준편차)을 사용한다. 가변어휘 인식실험 결과음소 단위의 정규화방법이 트라이 폰 기반 정규화방법에 비하여 우수한 성능을 보였으며 이러한 결과는 논문 [1-2]의 결과와는 상이한 것으로 트라이 폰 기반 정규화 방법이 미 출현 트라이 폰에 대하여 강인하지 못하다는 점을 시사하고 있다. 따라서 정규화 신뢰도가 음소 또는 트라이 폰에 상관없이 기준 신뢰도인 RLTC 신뢰도 [3]에 비하여 우수한 성능을 보였으며 가변어휘 인식에서도 동작함을 확인 할 수 있었다. Kim et al. Proposed Normalized Confidence Measure (NCM) [1-2] and it was successfully used for rejecting mis-recognized words in isolated word recognition. However their experiments were performed on the fixed word speech recognition. In this Paper we apply NCM to the domain of vocabulary independent speech recognition (VISP) and shows the rejection Performance of NCM in VISP. Specialty we Propose vector quantization (VQ) based method for overcoming the problem of unseen triphones. It is because NCM uses the statistics of triphone confidence in the case of triphone-based normalization. According to speech recognition experiments Phone-based normalization method shows better results than RLJC[3] and also triphone-based normalization approach. This results are different with those of Kim et al [1-2]. Concludingly the Phone-based normalization shows robust Performance in VISP domain.

역 알루미늄 유도 결정화 공정을 이용한 실리콘 태양전지 다결정 시드층 생성

최승호,박찬수,김신호,김양도,Choi, Seung-Ho,Park, Chan-Su,Kim, Shin-Ho,Kim, Yang-Do 한국재료학회 2012 한국재료학회지 Vol.22 No.4

The formation of high-quality polycrystalline silicon (poly-Si) on relatively low cost substrate has been an important issue in the development of thin film solar cells. Poly-Si seed layers were fabricated by an inverse aluminum-induced crystallization (I-AIC) process and the properties of the resulting layer were characterized. The I-AIC process has an advantage of being able to continue the epitaxial growth without an Al layer removing process. An amorphous Si precursor layer was deposited on Corning glass substrates by RF magnetron sputtering system with Ar plasma. Then, Al thin film was deposited by thermal evaporation. An $SiO_2$ diffusion barrier layer was formed between Si and Al layers to control the surface orientation of seed layer. The crystallinity of the poly-Si seed layer was analyzed by Raman spectroscopy and x-ray diffraction (XRD). The grain size and orientation of the poly-Si seed layer were determined by electron back scattering diffraction (EBSD) method. The prepared poly-Si seed layer showed high volume fraction of crystalline Si and <100> orientation. The diffusion barrier layer and processing temperature significantly affected the grain size and orientation of the poly Si seed layer. The shorter oxidation time and lower processing temperature led to a better orientation of the poly-Si seed layer. This study presents the formation mechanism of a poly seed layer by inverse aluminum-induced crystallization.