Wiener filtering을 이용한 잡음환경에서의 음성인식 ( Speech Recognition in Noisy Environments using Wiener Filtering )

엄기완,김진영 대한전자공학회 1996 대한전자공학회 학술대회 Vol.1996 No.1

품사셋에 의한 운율경계강도의 예측

엄기완,김진영,김선미,이현복,Eom Ki-Wan,Kim Jin-Yeong,Kim Seon-Mi,Lee Hyeon-Bok 대한음성학회 1998 말소리 Vol.35 No.-

This study intended to determine the most appropriate POS(Part of Speech) sets for predicting prosodic boundary strength efficiently. We used 3-level POB bets which Kim(1997), one of the authors, has devised. Three POS sets differ from each other according to how much grammatical information they have: the first set has maximal syntactic and morphological information which possibly affects prosodic phrasing, and the third set has minimal one. We hand-labelled 150 sentences using each of three POS sets and conducted perception test. Based on the results of the test, stochastic language modeling method was used to predict prosodic boundary strength. The results showed that the use of each POS set led to not too much different efficiency in the prediction, but the second set was a little more efficient than the other two. As far as the complexity in stochastic language modeling is concerned, however, the third set may be also preferable.

어메니티를 위한 도시환경 디자인 연구 : 부산시 보행자 안내 사인디자인 중심으로

엄기완 新羅大學校 1998 論文集 Vol.46 No.-

Recent days most of advanced countries recognized the importance of amenity as a significant goal in the urban development. In order to create the pleasant urban environment, many different methods have been employed. In this study attempts were made to investigate the problems and solutions in the design of Sign Map for the Pusan Pedestrian. Theoretical backgrounds made by previous studies were researched and analyzed comparatively in present study. The data including 170 pictures from Pusan and 90 pictures from foreign countries were analyzed in the forms of function, design, and material and suggestions were made for the solution of problems. The results from this study will be used in the development of design of Sign Map for the Pusan Pedestrian. Improved design in Sign Map for the Pusan Pedestrian from this study will provide the pleasant urban sight as well as accurate information to the residents in Pusan ans visitors.

Some properties of the index of coset in group

엄기완 원광대학교 사범대학 수학교육과 1988 시그마 Vol.5 No.-

DM(Direct Mail)廣告 List에 關한 硏究

嚴基完 新羅大學校 1985 論文集 Vol.19 No.-

In present marketing circumstance which is characterized by the individualization of consumers` desire and variety of marketing, mass media Advertisement which treat men as mass and thus uniform contents of information can`t offer the informations which consumers have asked for. Now we need the Advertisement of sales lebel, which lead consumers to point of purchase and carry on real buying. As on approach to this problem the importance of DM(Direct Mail) as direct response meditation has been more appealed. We have felt deeply the necessity for professional research about DM when we considered the fact that DM has been already occupied the third position in sequence after TV. Newspaper, in Europe and Japan, etc, and the reality which has culminated recognition about DM in our country every year. In this back ground, I studied the elements of list which are cores of DM. They are based on consumer`s giving out questions and DM sample questions in side of positioning. * Collecting DM list and viewpoints of management In the most important parts of DM, we have many difficulties in collecting list and effective dispatching and management. The lack of professional list lending enterprise and the difficulties of small scale management, social and human co-operative system have been obstacle elements. We must make a great efforts to collect and effectively manage the list in order to maintain the communication channel with consumers through DM. The frequent test about continuously changing list and the investigations about consumers and effective dispatching method that delivers the list to consumers and leads them to read it and carry on real buying action must be searched for. And the method must be invented that helps marketing activities and induces to make response rate high.

Wiener Filtering을 이용한 잡음환경에서의 음성인식

김진영,엄기완,최홍섭 한국음성과학회 1997 음성과학 Vol.1 No.-

In this paper, we present a robust recognition algorithm based on the Wiener filtering method as a research tool develop the Korean Speech recognition system. We especially used Wiener filtering method in cepstrum-domain, because the method in frequency-domain is computationally expensive and complex. Evaluation of the effectiveness of this method has been conducted in speaker-independent isolated Korean digit recognition tasks using discrete HMM speech recognition systems. In these tasks, we used 12th order weighted cepstral as a feature vector and added computer simulated white gaussian noise of different levels to clean speech signals for recognition experiments under noisy conditions. Experimental results show that the presented algorithm can provide an improvement in recognition of as much as from 5% to 20% in comparison to spectral subtraction method.

음성합성 시스템을 위한 음색제어규칙

김진영,엄기완 한국음성과학회 1997 음성과학 Vol.2 No.-

When listening the various speech synthesis systems developed and being used in our country, we find that though the quality of these systems has improved, they lack naturalness. Moreover, since the voice color of these systems are limited to only one recorded speech DB, it is necessary to record another speech DB to create different voice colors. 'Voice Color' is an abstract concept that characterizes voice personality. So speech synthesis systems need a voice color control function to create various voices. The aim of this study is to examine several factors of voice color control rules for the text-to-speech system which makes natural and various voice types for the sounding of synthetic speech. In order to find such rules from natural speech, glottal source parameters and frequency characteristics of the vocal tract for several voice colors have been studied. In this paper voice colors were catalogued as: deep, sonorous, thick, soft, harsh, high, tone, shrill, and weak. For the voice source model, the LF-model was used and for the frequency characteristics of vocal tract, the formant frequencies, bandwidths, and amplitudes were used. These acoustic parameters were tested through multiple regression analysis to achieve the general relation between these parameters and voice colors. Keywords: voice color, speech synthesis, vocal tract, voice personality,

LDA와 tri-tone 모델을 이용한 운율경계강도 예측

강평수,엄기완,김진영 한국음향학회 1999 韓國音響學會誌 Vol.18 No.7

본 논문에서는 발화된 문장으로부터 운율 경계 강도를 효과적으로 예측하기 위해 LDA와 tri-tone 모델을 혼합한 방법을 제안하였다. 이 방법은 기존의 LDA 방법을 사용하여 음절과 휴지기의 길이 정보를 운율경계강도 예측에 적용하고 피치정보를 벡터양자화에 적용하여 tri-tone이란 개념을 도입한 혼합형 모형이다. 제안된 방법은 주어진 200문장의 운율경계 강도를 예측하는 실험에서 72%의 정확성을 나타내었다. In this paper we propose a new mixed method of LDA and tri-tone model to predict Korean prosodic break indices(PBI) for a given utterance. PBI can be used as an important cue of syntactic discontinuity in continuous speech recognition(CSR). The model consists of three steps. At the first step, PBI was predicted with the information of syllable and pause duration through the linear discriminant analysis (LDA) method. At the second step, syllable tone information was used to estimate PBI. In this step we used vector quantization (VQ) for coding the syllable tones and PBI is estimated by tri-tone model. In the last step, two PBI predictors were integrated by a weight factor. The proposed method was tested on 200 literal style spoken sentences. The experimental results showed 72% accuracy.

운율구 단위의 연속음 인식

강지영,엄기완,김진영,최승호 한국음향학회 1999 韓國音響學會誌 Vol.18 No.8

Generally, a speaker structures utterances very clearly by grouping words into phrases. This facilitates the listener's recovery of the meaning of the utterance and the speaker's intention. To this purpose, a speaker uses, among other things, prosodic information such as intonation pause, duration, intensity, etc. The research described here is concerned with the relationship between the strength of prosodic boundaries in spoken utterances as perceived by untrained listeners(Perceptual boundary strength, PBS)-In this paper, the preceptual boundary strength is used as the same meaning of the prosodic boundary strength-and prosodic information. We made a rule determinating the prosodic boundaries and verified the usefulness of the prosodic phrase as a recognition unit. Experiments results showed that the performance of speech recognition(SR) is improved in aspect of recognition rate and time compared with that using sentences as recognition unit. In the future we will suggest the methods that estimate more appropriate boundaries and study more various methods of prosody assisted SR. 일반적으로 사람은 말을 할 때 어절들은 몇몇의 구로 그룹핑하여 발음함으로써 발화한다. 이것은 듣는 사람으로 하여금 발화의 의미와 의도를 잘 파악하도록 도와준다. 특히, 이러한 목적으로 발화자는 무의식적으로 운율정보(억양, 장단, 리듬 등)를 적절히 사용하게 된다. 본 논문에서는 발화된 문장에서 운율경계를 인식의 단위로 하는 음성인식방법에 대하여 제안한다. 즉, 발화된 문장을 운율구단위로 나누는 방법을 제안하고 나누어진 단위에 따라 연속음 인식실험을 수행하였다. 인식실험결과 연속음인식 시간의 감소를 관찰할 수 있었으며, 물론 음성인식률도 20-10%정도 증가하였다.

코퍼스 기반 음성합성기의 데이터베이스 축소 방법

최승호,엄기완,강상기,김진영 한국음향학회 2003 韓國音響學會誌 Vol.22 No.8

코퍼스 기반 음성합성방식은 그 합성음의 자연성이 매우 우수하여 널리 사용되고 있으나 대용량의 데이터베이스 (DB)를 사용하기 때문에 그 적용분야가 매우 제한적이다. 본 연구에서는 이러한 코퍼스 기반 음성합성기의 대용량 DB 문제를 해결하기 위한 방안으로서 DB 축소 방법 대한 알고리듬을 제안하고 평가하였다. 본 논문에서는 DB 축소 알고리듬으로서 세 가지 방법을 제안하였는데, 첫 번째는 Modified K-means 군집화를 이용한 DB 축소 알고리듬이고 다음은 적절한 문장 셋을 정의하고 이 문장 셋을 합성할 때 사용된 단위들을 이용하는 방법이다. 마지막으로는 대용량 문장 셋을 정의하고 해당 문장을 음성합성하고, 음편들의 사용 빈도수를 고려하여 군집화를 하는 것이다. 세 가지 방법을 이용하여 합성 DB를 유사한 크기로 축소하였을 때, 대용량 문장 셋과 빈도를 고려한 세 번째 방법이 가장 우수한 음질을 보였다. 또한 마지막 방법은 합성음의 음질은 저하시키지 않으면서 합성 DB만을 감소시키는 성능을 보여, 제안된 방법의 타당함을 입증할 수 있었다. Because of their human-like synthesized speech quality, recently Corpus-Based Text-To-Speech(CB-TTS) have been actively studied worldwide. However, due to their large size speech database (DB), their application is very restricted. In this paper we propose and evaluate three DB reduction algorithms to which are designed to solve the above drawback. The first method is based on a K-means clustering approach, which selects k-representatives among multiple instances. The second method is keeping only those unit instances that are selected during synthesis, using a domain-restricted text as input to the synthesizer. The third method is a kind of hybrid approach of the above two methods and is using a large text as input in the system. After synthesizing the given sentences, the used unit instances and their occurrence information is extracted. As next step a modified K-means clustering is applied, which takes into account also the occurrence information of the selected unit instances, Finally we compare three pruning methods by evaluating the synthesized speech quality for the similar DB reduction rate, Based on perceptual listening tests, we concluded that the last method shows the best performance among three algorithms. More than this, the results show that the last method is able to reduce DB size without speech quality looses.