• 대한의용생체공학회:의공학회지
• /
• 제31권6호
• /
• pp.481-486
• /
• 2010

In this paper, we suggest an advanced algorithm, to recognize pedestrian/non-pedestrian using differential haar-like feature, which applies Adaboost algorithm to make a strong classification from weak classifications. First, we extract two feature vectors: horizontal haar-like feature and vertical haar-like feature. For the next, we calculate the proposed feature vector using differential haar-like method. And then, a strong classification needs to be obtained from weak classifications for composite recognition method using the differential area of horizontal and vertical haar-like. In the proposed method, we use one feature vector and one strong classification for the first stage of recognition. Based on our experiment, the proposed algorithm shows higher recognition rate compared to the traditional method for the pedestrian and non-pedestrian.

• 대한산업공학회지
• /
• 제31권4호
• /
• pp.349-357
• /
• 2005

The Adaptive Multi-Rate (AMR) speech codec is the mandatory for voice service in WCDMA systems. The AMR codec can be used efficiently to provide a balanced trade-off between the capacity and quality of voice by adjusting various service rates. In this paper, three ways of AMR mode allocation schemes on the downlink in WCDMA system are evaluated. To evaluate users satisfaction efficiently, new system performance measure and analytic models are proposed. The proposed analytic models can be applied to obtain optimal mode allocation ways while considering the system capacity and quality of voice. In numerical examples, the ways of finding optimal parameters are illustrated for the given traffic loads and the performances of three mode allocation schemes are compared.

• 한국항해항만학회:학술대회논문집
• /
• 한국항해항만학회 2015년도 춘계학술대회
• /
• pp.272-274
• /
• 2015

Success of communication at sea is directly linked with clear and complete delivery and receipt of the target message between interlocutors. It can be said that speakers' effective delivery of their intended message and listeners' precise decoding and accurate understanding are the keys to successful maritime communication. From this perspective, the scope of maritime English education and training needs to be reconceptualized and expanded into the area of communication itself, beyond the simple acquisition of, and familiarization with, IMO Standard Maritime Communication Phrases (SMCP). Therefore, in order to make learners' acquisition of marine communication knowledge more feasible, and the knowledge learned more practically applicable, training on effective and clear oral delivery should be also considered within the frame of maritime English education. Thus, critical training elements to realize this goal need to be suggested as guidelines. In this presentation, the theoretical background on this will be introduced in terms of English as a Lingua Franca (ELF) and Lingua Franca Core (LFC), which are the current mainstream forms of English communication in the international business context. Based on this, six key training elements will be discussed; that is, speech rate, word groups, pauses, nuclear stresses, consonants (including consonant clusters), and vowels (specifically long and short vowels). Finally, the practical pedagogical methods of each element, and its actual application into a real ESP classroom, will be suggested.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 1998년도 학술발표대회 논문집 제17권 2호
• /
• pp.381-384
• /
• 1998

본 논문은 주행중인 자동차 환경에서 운전자의 안전성 및 편의성의 동시 확보를 위하여, 보조적인 스위치 조작없이 상시 음성의 입, 출력이 가능하도록 한다. 이때 잡음에 강인한 threshold 값을 구하기 위하여, 일정한 시간마다 기준 에너지와 영교차율(Zero Crossing Rate)을 변경하며, 밴드패스 필터(bandpass filter)를 이용하여 1차, 2차로 나누어 실시간 상태에서 자동으로, 정확하게 끝점검출(End Point Detection)을 처리한다. 기준패턴(reference pattern)은 DMS(Dynamic Multi-Section)을 사용하며, 화자의 변별력을 높이기 위하여 2개의 모델사용을 제안한다. 또한 주행중인 차량의 잡음환경에 강인하기 위하여 일반주행(80km/h 이내), 고속주행(80km/h 이상)등으로 나누며 차량의 가변잡음 크기에 따라 자동으로 선택하도록 한다. 음성의 특징 벡터와 인식 알고리즘은 PLP 13차와 One-Stage Dynamic Programming (OSDP)를 이용한다. 실험결과, 자주 사용되는 차량 편의장치 제어명령 33개에 대하여 중부, 영동 고속도로(시속 80Km/h 이상)에서 화자독립 89.75%, 화자종속 90.08%의 인식율을 구하였으며, 경부 고속도로에서는 화자독립 92.29%, 화자종속 92.42%의 인식율을 구하였다. 그리고 저속 주행중인 자동차 환경(80km/h 이내, 시멘트, 아스팔트 등의 서울시내 및 시외독립)에서는 화자독립 92.89%, 화자종속 94.44% 인식율을 구하였다.

• 한국음향학회지
• /
• 제23권3호
• /
• pp.228-234
• /
• 2004

본 논문에서는 공통 코드북의 평균값과 개인 코드북의 평균값 간의 바이어스 제거에 의한 채널 불일치 보상 알고리즘을 제안하였다. 제안한 방식은 학습시 공통 코드북의 센터값과 학습 데이터의 센터값과의 차수별 차를 미리 보상하여 학습하고, 확인시에도 공통 코드북의 센터값과 학습 데이터의 센터값과의 차수별 차를 보상하여 확인함으로써 채널의 불일치에 의한 급격한 본인 인식율 하락을 해결한다. 그러나 무조건적인 평균값 보상은 사칭자의 인증오류를 가져오게 되므로 채널의 변이에 비례하는 적절한 가중치를 통한 평균값 보상이 필요하다. 따라서, 제안하는 방식은 음성구간을 제외한 묵음구간의 분포를 고려하여 학습시 채널과의 변이차이를 비선형함수에 의한 가중치로 보상해준다. 모의 실험 결과 기존의 켑스트럼 평균 차감법을 사용할 때보다 제안한 알고리즘을 적용했을 때의 본인 거부 오류율이 평균 14.95% 감소함을 알 수 있었다.

• 한국음향학회지
• /
• 제18권2호
• /
• pp.77-82
• /
• 1999

정현파 모델은 음성과 오디오 신호의 분석과 합성에 많이 활용되어 왔으며 최근 고음질 저비트율 오디오 부호화에 효율적인 방법의 하나로 대두되고 있다. 정현파 모델을 이용한 오디오 신호의 분석과 합성에서 중요한 단계 중의 하나는 순음의 검출이다. 본 논문은 정현파를 이용한 오디오 신호의 분석과 합성에 매스킹 효과와 매스킹 인덱스 그리고 JNDf(Just Noticeable Difference in Frequency) 등의 심리음향적 기준들을 활용하는 효율적인 방안을 제안하였다. 모의실험 결과, 심리음향적 기준을 사용하면 합성된 음질에 거의 영향을 주지 않으면서 합성에 사용되는 정현파의 개수를 현저하게 줄일 수 있었음을 알 수 있었다.

• 한국전자통신학회논문지
• /
• 제6권2호
• /
• pp.237-242
• /
• 2011

본 논문에서는 다층 퍼셉트론 신경회로망을 사용하여 각 프레임에서의 유성음, 무성음, 그리고 묵음 구간을 검출하는 구간검출 알고리즘을 제안한다. 다층 퍼셉트론 신경회로망의 입력으로는 고속 푸리에변환에 의한 전력스펙트럼 및 고속 푸리에변환 계수가 사용되어 네트워크가 학습된다. 본 실험에서는 원 음성에 백색잡음이 중첩된 음성을 신경회로망에 입력함으로서 각 프레임에서의 유성음, 무성음, 묵음 구간의 검출성능 결과를 나타낸다. 본 실험에서는 신경회로망의 학습 데이터 및 평가 데이터가 다를 경우에도 이러한 음성 및 백색잡음에 대하여 92% 이상의 검출율을 구할 수 있었다.

• 대한치과보철학회지
• /
• 제39권6호
• /
• pp.698-708
• /
• 2001

This study was to evaluate the clinical evidence for the success and the predictability of the osseointegrated dental implants in the partially edentulous Korean patients. 201 patients have received total of 502 Branemark implants, which were restored with either single or multiunit fixed superstructures at the Implants Clinic, Yonsei University Dental Hospital. The clinical and radiographic evaluations carried out on the patients for maximum 8 years were assessed annually for peri-implant inflammation, implant mobility (PTV), Plaque Index (PI), Gingival Index (GI), Bleeding on Probing (BOP), Keratinized Mucosa width and any changes in the surrounding bone level. The radiographs were taken at completion of the restoration, and annually thereafter. On the last recall appointments the patients filled a questionnaire consist of 29 questions in four categories. The cumulative non-failure rate of success was 93.9%. The first year mean bone loss was 0.3mm and less than 0.2mm annually thereafter. The periodontal parameters, keratinized tissue width and periotest values stabilized after initial changes in the first few years. The questionnaire has shown general satisfactory responses in all four aspect of dental implants treatment, including chewing efficacy, comfort, aesthetics and speech. The results support the predictability and success of the long-term rehabilitation of implant supported prostheses in partially edentulous Korean patients.

• 대한후두음성언어의학회지
• /
• 제9권2호
• /
• pp.134-141
• /
• 1998

Mutational falsetto is a kind of voice disorders due to the failure to acquire proper low-pitched voice during the puberty. The patients with mutational falsetto can produce the normal low-pitched voice by the surgical treatment, like the type III-thyroplasty, or the voice therapy. The present study is, focusing on the latter treatment, to consider the efficiency of voice therapy for the mutational falsetto. The 7 patients who were diagnosed as mutational falsetto by the laryngologists, and treated by the voice therapist were selected as subjects. Their voices of pretherapy and posttherapy were analyzed on the aspects of acoustics and aerodynamics. Acoustic analysis was done by the MDVP(Multidimensional Voice Program) of CSL(Computerized Speech Lab, Kay Elemetrics, Co.), and aerodynamic analysis, by the Maximum Sustained Phonation of Aerophone II(Kay Elemetrics, Co.). By these measurements, we could find that fundamental frequency(F0) was significantly lowered, on the average, 65Hz. Maximum phonation time(MPT) was increased 4.57 second, and shimmer was decreased 1.644%, respectively, and each changes was statistically significant, too. On the average, jitter was decreased 0.499%, mean flow rate(MFR) was decreased 27.71ml/sec, and NHR was increased 0.023 which was the only parameter not showing improvement. But the changes of jitter, MFR and NHR were not statistically significant.

• 대한후두음성언어의학회지
• /
• 제11권1호
• /
• pp.98-103
• /
• 2000

Background and Objectives : The etiology and pathophysiology of spasmodic dysphonia is yet unknown. This study was performed to determine if any laryngeal aerodynamic parameter distinguish the voice of patient diagnosed as having adductor spasmodic dysphonia from individuals with normal voice production and to investigate the pathophysiology of spasmodic dysphonia. Materials and Methods : fifteen women diagnosed as having adductor spasmodic dysphonia and fifteen normal control women participitated in this study Maximum phonation time, mean air flow rate, subglottic pressure, vocal efficiency, Vfo, NHR, VTI, FTRI, ATRI, Jitter percent, Shimmer percent were obtained from the participants using 'MDVP(multi-dimensional voice program)' of CSL(Computerized Speech lab, Kay Elemetrics, Co., Model No. 4300), and 'maximum sustained phonation' and 'IPIPI test' of AP II(Aerophone II, Kay Elemetrics, Co., Model 6800). Results : T-test statistical analysis revealed statistically different values for vocal efficiency, Vfo, NHR, MPT, litter percent, Shimmer percent between the spasmodic dysphonia group and the control group. Conclusions : Spasmodic dysphonia affects the ability of the laryngeal mechanism to function effectively. Results from our study demonstrate that certain aerodynamic and acoustic parameters distinguish adductor spasmodic dysphonia from normal voice.