• Proceedings of the KSME Conference
• /
• 2005.11a
• /
• pp.225.2-225.2
• /
• 2005

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2013.06a
• /
• pp.56-57
• /
• 2013

본 논문에서는 연극공연을 관람하는 관객의 반응정보를 수집하기 위하여, 청각센서를 통해 관객의 음성을 획득하고 획득된 음성에 대한 감정을 예측하여 관객 반응정보 관리시스템에 전송하는 음성신호 기반 감정인식 시스템을 구현한다. 이를 위해, 관객용 헤드셋 마이크와 다채널 녹음장치를 이용하여 관객음성을 획득하는 인터페이스와 음성신호의 특징벡터를 추출하여 SVM (support vector machine) 분류기에 의해 감정을 예측하는 시스템을 구현하고, 이를 관객 반응정보 수집 시스템에 적용한다. 실험결과, 구현된 시스템은 6가지 감정음성 데이터를 활용한 성능평가에서 62.5%의 인식률을 보였고, 실제 연극공연 환경에서 획득된 관객음성과 감정인식 결과를 관객 반응정보 수집 시스템에 전송함을 확인하였다.

• Journal of KIISE:Computing Practices and Letters
• /
• v.11 no.2
• /
• pp.101-111
• /
• 2005

The present web informations are mainly described in terms of HTML, which users obtain through input devices such as mouse, keyboard, etc. Thus the existing GUI environment have not supported human's most natural information acquisition means, that is, voice. To solve the problem, several vendors are developing voice user interface. However these products are deficient in man -machine interactivity and their accommodation of existing web environment. This paper presents a VUI(Voice User Interface) supporting web browser by utilizing more and more maturing speech recognition technology and VoiceXML, a markup language derived from XML. It provides users with both interfaces, VUI as well as GUI. In addition, XML Island technology is applied to the bowser in a way that VoiceXML fragments are nested in HTML documents to accommodate the existing web environment. Also for better interactivity, dialogue scenarios for menu, bulletin, and search engine are suggested.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.45 no.5
• /
• pp.162-172
• /
• 2008

In this paper, we propose a multimodal speaker verification method using teeth image and voice as biometric trait for personal verification in mobile terminal equipment. The proposed method obtains the biometric traits using image and sound input devices of smart-phone that is one of mobile terminal equipments, and performs verification with biometric traits. In addition, the proposed method consists the multimodal-fashion of combining two biometric authentication scores for totally performance enhancement, the fusion method is accompanied a weighted-summation method which has comparative simple structure and superior performance for considering limited resources of system. The performance evaluation of proposed multimodal speaker authentication system conducts using a database acquired in smart-phone for 40 subjects. The experimental result shows 8.59% of EER in case of teeth verification 11.73% in case of voice verification and the multimodal speaker authentication result presented the 4.05% of EER. In the experimental result, we obtain the enhanced performance more than each using teeth and voice by using the simple weight-summation method in the multimodal speaker verification system.

• Journal of Platform Technology
• /
• v.11 no.6
• /
• pp.79-88
• /
• 2023

Recently, industrial accidents have continued to increase due to the industrialization, and worker safety management is recognized as essential to reduce losses due to hazardous factors at work places. To manage the safety of workers, it is required to apply customized safety management artificial intelligence technology that takes into account the characteristics of industrial sites, and a service for real-time risk detection and response to workers depending on the situation based on safety accident types and risk analysis for each task and process. The proposed safety management service consists of worker devices to acquire sensor data, edge devices to collect from IoT-based sensors, and a voice chatbot to support workers' disaster response. The voice chatbot plays a major role in interacting with workers at disaster sites to respond to risks. This paper focuses on real-time risk response using an IoT-based system and voice chatbot on a server for work safety according to the worker's situation. A Scenario-based voice chatbot is used to process responses at the edge level to provide safety management services.

• The Journal of the Korea Contents Association
• /
• v.20 no.6
• /
• pp.99-113
• /
• 2020

In the future, robots and drones for the convenience of our lives in everyday life will increase. As a method for controlling this, a remote control or a human voice method is most commonly used. However, the remote control needs to be operated by a person and can not ignore ambient noise in the case of voice. In this paper, we propose an economical attitude information acquisition method to accurately acquire the posture information of the arm in real time under the assumption that the surround drones or robots can be controlled wirelessly with the posture information of the arm. For this purpose, the extended Kalman filter was used to eliminate the noise of the arm position information. in order to detect the arm movement, a low cost MEMS type sensor was applied to secure the economical efficiency of the apparatus. To increase the wear ability of the arm, We developed a compact and lightweight attitude information acquisition system by integrating all functions into one chip as much as possible. As a result, the real-time performance of 1 ms was secured and the extended Kalman filter was applied to acquire the accurate attitude information of the arm with noise removed and display the attitude information of the arm in real time. This provides a basis for generating commands using real-time attitude information of the arm.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.13 no.2 s.29
• /
• pp.161-167
• /
• 2007

The goal of this paper is to develop the speech recognition system which can control the ship's auto pilot. The feature parameters predicting the speaker's intention was extracted from the sample wheel orders written in SMCP(IMO Standard Marine Communication Phrases). And we designed the post-recognition procedure based on the parameters which could make a final decision from the list of candidate words. To evaluate the effectiveness of these parameters and the procedure, the basic experiment was conducted with total 525 wheel orders. From the experimental results, the proposed pattern recognition procedure has enhanced about 42.3% over the pre-recognition procedure.

• Journal of Broadcast Engineering
• /
• v.23 no.2
• /
• pp.253-260
• /
• 2018

Personalized media broadcasting services can produce their own broadcasting contents with a variety of creative themes if they have just a transmission platform and devices that can obtain videos and voices of producers without the existing expensive equipment. In this paper, we develop and implement a new broadcasting system by applying this service framework to events such as seminars or academic conferences. The devices can be installed at each conference rooms and the integrated system transmitted to users. They can watch via their multi-resolution screen, such as smart-phones, laptops, and tablet PCs. It has the advantage of being able to receive real-time streaming and VOD services as well as additional information related to the conference. It is expected to provide convenience by allowing attendees to access the information via their devices, thereby creating an impact on participation and the underlying technology for the future research.

• Journal of the Korea Convergence Society
• /
• v.10 no.11
• /
• pp.111-116
• /
• 2019

With recent rapid development of computer capabilities, various technologies that can facilitate the interaction between humans and computers are being studied. The paradigm tends to change to NUI using the body such as 3D motion, haptics, and multi-touch with GUI using traditional input devices. Various studies have been conducted on transferring human movements to computers using sensors. In addition to the development of optical sensors that can acquire 3D objects, the range of applications in the industrial, medical, and user interface fields has been expanded. In this paper, I provide a model that can execute other programs through gestures instead of the mouse, which is the default input device, and control Windows based on the lip motion. To propose a model which converges with an Android application and can be controlled by various media and voice instruction functions using voice recognition and buttons through connection with a main client. It is expected that Internet media such as video and music can be controlled not only by a client computer but also by an application at a long distance and that convenient media viewing can be performed through the proposal model.

• Journal of Industrial Technology
• /
• v.28 no.B
• /
• pp.239-244
• /
• 2008

We have realized the swept source optical coherence tomography(SS-OCT) by using the self-fabricated wavelength swept laser(wavelength tuning range : 80nm, line-width : 0.12nm, wavelength sweeping rate : 50Hz). In addition, we have used the dual balanced detector that could make a mirror image in OCT display suppressed. We can also fabricate the comb filter of Michelson interferometer type for fast-signal processing in OCT. Using this SS-OCT system for measuring an mirror, a 1mm-depth glass and an onion, we confirmed that the in vivo epidermal cross-sectional images for them can be obtained appropriately.