http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
조철우,리타오 國立 昌原大學校 產業技術硏究院 2004 産技硏論文集 Vol.18 No.-
In this paper we tried to classify the pathological voice signal with severe noise component based on two different parameters, the spectral slope and the ratio of energies in the harmonic and noise components (HNR). The spectral slope is obtained by using a curve fitting method and the HNR is computed in cepstrum quefrency domain. Speech data from normal peoples and patients are collected, diagnosed and divided into three different classes (normal, relatively less noisy and severely noisy data). The mean values and the standard deviations of the spectral slope and the HNR are computed and compared with in the three kinds of data to characterize and classify the severely noisy pathological voice signals from others.
조철우 國立 昌原大學校 精報通信硏究所 1999 精報通信論文集 Vol.3 No.-
Recently interests about audio-visual speech activities are growing. The knowledge of audio-visual speech can be used for speech recognition to improve the recognition rate and also can be used for speech synthesis to add the visual images for enhanced message transfer. There are many researchers carried on about audio-visual speech area. But the way of measuring visual speech activity varies according to the research groups. Some of the method use expensive tracking device. Some of the method use simple markers to trace the movement of articulatory organs. The main aim of this experiments is to collect some audio-visual materials which can be used for later experiments to estimate and model the actions of human articulatory organs such as mouth, jaw etc. In this collection process we collect audio-visual data from the seven directions separately. Twelve makers are used to trace the movements.
조철우,김광인,박지현 창원대학교 정보통신연구소 2000 精報通信論文集 Vol.4 No.-
In this paper procedures for collecting, analyzing and implementing audio-visual speech are described. First, we defined a list of Koren vesemes according to the articulatory positions of the phoneme set. By the list we defined pronunciation list. Based on the list audio-cisual speech is collected. Next, part of the collected audio-visual speech is analysed to obtain variation of coordinates which corresponds to the motion of real articulatory organs. Finally the measured result is applied to the graphic facial model to synthesize audio-visual speech.
조철우 國立 昌原大學校 情報通信硏究所 1997 精報通信論文集 Vol.1 No.-
This paper discusses some aspects about representing emotional contents in the form of speech signals in multimedia environments. First the characteristics of general emotional speech is described. Then the methodology of collecting emotional speech and also the merits and drawbacks of each methods are discussed. Meanwhile some characteristics of emotional speech signals, which is measured at the previously performed experiments are mentioned. Finally some elements which is required to implement emotions in speech domain are discussed.
조철우,방호균 國立 昌原大學校 産業技術硏究所 1995 産技硏論文集 Vol.9 No.-
최근 고품질의 음성합성과 코딩을 위한 연구의 일환으로 성문폐쇄시점(GCI:Glottal Closure Instant) 검출을 위한 다양한 기법들의 개발이 진행 중이다. 본 연구에서는 GCI 검출에 가장 효과적인 기법으로 알려진 힐버트 변환법을 이용한 EFLPR(Epoch Filtering of Linear Prediction Residual)법을 제안하였다. 본 연구에서 제안된 SGCID법은 합성음과 남성, 여성 화자의 자연음의 단모음과 숫자음 그리고 연속음 등 다양한 음성에 적용한 결과 안정된 유성음 구간에서 기존 방법들에 비해 정확한 GCI를 검출할 수 있었다.
TIMIT speech data를 이용한 HMM 음성분할기의 성능평가에 관한 연구
조철우 창원대학교 정보통신연구소 2001 精報通信論文集 Vol.5 No.-
This paper describes a procedure to measure the performance of HMM based automatic speech segmenter. To evaluate the performance of the HMM based segmentation, hand labeled TIMIT database is used. Relative position differences between HMM labels and manual labels are measured. Also frequency distributions of differences according to the various phonetic environments. The measured result showed that current HMM label is correct with 95% correctness within 5ms difference range. Also the biggest errors are caused in the case of combination of vowel and silence. The differences are measured in terms of various phonetic environments. As a result some statistics on the segmental errors are obtained.
조철우,정인화 國立 昌原大學校 精報通信硏究所 1998 精報通信論文集 Vol.2 No.-
In this paper a multimedia speech player, which can convert normal speech file or synthesized speech file into multimedia form is suggested. This software is improved in the points that it requires only small sized image database compared to that current similar programs require huge amount of image database. This software tool can effectively display animated facial images and speech sounds together in synchronized form even at PC level. Implemented tool can be used as a plugin or an independent form. New multimedia file structure is suggested too.
포만트 합성법을 이용한 한국어 규칙합성중 음소간의 결합규칙의 구현방법에 관한 연구
조철우 國立 昌原大學校 産業技術硏究所 1990 産技硏論文集 Vol.4 No.-
This paper describes an implementation method of concatenation rules between phonemes in Korean synthesis-by-rule system. the method suggested is about the implementation of generation rule of synthesis parameters continuously, which are required during the synthesis procedures for Korean speech using formant speech synthesizer. At first the description rules for tracks are discussed, then allophonic rules characteristics, which are general and similar to the rule description formular in generative phonology, synthesizer, According to the suggested method, rules are described by the distinctive features of each phoneme. And PASCAL, general purpose programming language, is used for the software implementation of the rules.
멀티미디어 환경에서 정서의 표현을 위한 기술언어 ESDML의 제안
조철우 國立 昌原大學校 産業技術硏究所 1998 産技硏論文集 Vol.12 No.-
This paper proposes an ESDML(Emotional Speech Description Markup Language) which can describe an emotional speech in multimedia form. Also ESDML browser is proposed to output emotional speech and images effectively. The grammar of ESDML is described in detail. And the elements for the browser are described as well.
조철우 國立 昌原大學校 産業技術硏究所 1994 産技硏論文集 Vol.8 No.-
Several research results about successful implementation of speech synthsizers are being reported as improvements of speech synthesis technologies, but there are not so much research reports available for the systematic assessment of synthesized speech. In this paper some examples of EC and US researches, some examples of Korean synthesis systems and their assessments are reported.And some points about assessment using nonsense wordset are discussed.