• The Journal of the Convergence on Culture Technology
• /
• v.7 no.4
• /
• pp.675-681
• /
• 2021

With the recent growth of the AI speaker market, the demand for speech synthesis technology that enables natural conversation with users is increasing. Therefore, there is a need for a multi-speaker speech synthesis system that can generate voices of various tones. In order to synthesize natural speech, it is required to train with a large-capacity. high-quality speech DB. However, it is very difficult in terms of recording time and cost to collect a high-quality, large-capacity speech database uttered by many speakers. Therefore, it is necessary to train the speech synthesis system using the speech DB of a very large number of speakers with a small amount of training data for each speaker, and a technique for naturally expressing the tone and rhyme of multiple speakers is required. In this paper, we propose a technology for constructing a speaker encoder by applying the deep learning-based x-vector technique used in speaker recognition technology, and synthesizing a new speaker's tone with a small amount of data through the speaker encoder. In the multi-speaker speech synthesis system, the module for synthesizing mel-spectrogram from input text is composed of Tacotron2, and the vocoder generating synthesized speech consists of WaveNet with mixture of logistic distributions applied. The x-vector extracted from the trained speaker embedding neural networks is added to Tacotron2 as an input to express the desired speaker's tone.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.15 no.1
• /
• pp.97-105
• /
• 1997

In order to conserve architecture accurately as original feature, we need to get the data which is measured in detail about the information of size, position, type, shape, and other terms for architectures. Recently, we need to get data fast and correctly, to compose and to update the database of geographic information system, but using existing method. it is difficult to update the data effectively in the rapid development of information industry. In this study, therefore, It is measured precisely architecture using accurate close-range photogrammetry The aims of this study are real-measured plan by processing acquired image through multi-image matching method if photogrammetry, suggesting the method which is able to draw in detail front plane and plane of well-established architecture, and giving data to obtain accurate size and 3-D coordinate using analytical method.

• The Journal of the Korea Contents Association
• /
• v.12 no.4
• /
• pp.1-9
• /
• 2012

The purpose of this study is to present a methodology to effectively produce the e-learning contents in two platform types of PC based and mobile based platforms. Currently, the compatibility problem of flash technology widely used in the e-learning content production of existing PC based platform not supporting flash in some mobile platforms is being created while the condition of accompanying inconvenience in learning due to small screen and difference of input interface is being created in case of using the contents for PC even in case of the mobile platforms in which flash is supported. Considering this situation, this study is presenting a content development methodology that can provide services in both platform types based on standard technologies such as HTML 5. In conclusion, the methodology proposed in this study separates the content and the view of learning contents while supporting to enable content implementation using various forms of views based on the content database that has been constructed to show very efficiently characteristics in the content reuse and the multi platform support categories.

• Journal of the Korean Society of Systems Engineering
• /
• v.7 no.2
• /
• pp.7-11
• /
• 2011

A high-speed railway system represents a typical example of large-scale multi-disciplinary system, consisting of subsystems such as rolling-stock, electrical hardware, electronics, control, information, communication, civil technology etc. The system design and acquisition data of the large-scale system must be the subject under strict configuration control and management. Systems engineering technology development project for Korea next generation High-speed Electric Multiple Unit (HEMU) system in progress is a national large system development project that is not only a large-size and complex but also multi-disciplinary in nature. Therefore, all stakeholders must understand and share the functional and performance requirements of HEMU throughout its life-cycle phases. Also in the test and evaluation phase, all systems requirements must be verified. In 2011, the prototype train manufacturing will be completed. It will do test run on the commercial line and all systems requirements are verified until 2012. For the system verification, the test and evaluation process have to be established before the test trial run. Using a systems engineering tool, the system design database(SDD) with requirements traceability and development process management in the course of the development have to be established. This paper represents the test and evaluation process development based on the SEMP(Systems Engineering Management Plan) developed in the design stage. The test and evaluation process is refined and updated in comparison to the design stage one. The test and evaluation process consists of procedure, test and evaluation method and schedule. So through this process, it is defined that each systems requirements is verified on which test and about what time.

• Journal of Korea Multimedia Society
• /
• v.13 no.2
• /
• pp.216-224
• /
• 2010

A RAM-based Neural Network(RNN) has the advantages of processing speed and hardware implementation. In spite of these advantages, it has a saturation problem, weakness of repeated learning and extract of a generalized pattern. To resolve these problems of RNN, the 3DNS model using cumulative multi discriminator was proposed. But that model does not solve the saturation problem yet. In this paper, we proposed a adaptive weight cumulative neural net(AWCNN) using the adaptive weight neuron (AWN) for solving the saturation problem. The proposed nets improved a recognition rate and the saturation problem of 3DNS. We experimented with the MNIST database of NIST without preprocessing. As a result of experimentations, the AWCNN was 1.5% higher than 3DNS in a recognition rate when all input patterns were used. The recognition rate using generalized patterns was similar to that using all input patterns.

• Journal of the Korean Society for information Management
• /
• v.24 no.4
• /
• pp.73-96
• /
• 2007

In this study, the characteristics of scholarly communication through the Web and scientific journals are explored, and scholarly communication patterns in two scientific disciplines are compared to reveal the difference. Economics and Computer Science-Information Systems are selected as two disciplines to be analyzed. In the data collection process, 10 keywords are extracted from a database for each subject field, and scholarly Web pages and journal articles related to these keywords are collected and analyzed. Our investigation includes the characteristics of scholarly Web pages, Multi-Dimensional Scaling (MDS) analysis of co-linked Web pages as well as co-cited journal articles, and changes in the scholarly communication activities occurring on the Web and in scientific journals respectively over time. We found certain differences as well as common features in scholarly communication patterns between the Web and scientific journals for both fields of Economics and Computer Science. We also found that scholarly communication occurring on the Web displays unique features for each subtopic within the same field of study.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.13 no.6
• /
• pp.243-248
• /
• 2013

With rapid growth and fast diffusion of smartphone technologies, many users are deeply concerned about the smart applications and many mobile applications converged with various related technologies are rapidly disseminated. Especially, the convergence technologies like mobile apps that can establish the wireless ad hoc network between smartphone and other peripherals and exchange data are appear and progressed continuously. In this paper, we design and implement the smart app using bluetooth based wireless ad hoc sensor network that can connect smartphone with sensors and exchange data for various smart applications. The proposed smart application in this paper collects data obtained from more than 2 multi-sensors in real time and fulfills the decision making function by storing data at the database and analysing it. The smart application designed and implemented in this paper is the healthcare application that can analyze and evaluate the patient's health condition with sensing data from multi-sensors in real time through bluetooth module.

• Journal of the Ergonomics Society of Korea
• /
• v.34 no.5
• /
• pp.501-517
• /
• 2015

Objective: The aim of this study was to investigate the technology trend on developing helmets based on the analysis of Korean, US, European and PCT patents. Background: The usage of the personal protective equipment such as the safety helmet would be essential for preventing injuries and reducing the severity of injuries depending on the type of activity and the exposed environment. Method: Database of WIPSON was used for searching the patents. The objects were the Korean, US, European and PCT patents registered or opened from January 2011 to December 2014. The keyword to investigate was helmet, and International Patent Classification (IPC) was used for investigating the technology trend of the patents. Results: The cases of US patents annually increased, but the cases of European patents decreased. US patents belonged to all sections but Korean and European patents and PCT patents did not. The largest number of IPC, to which four areas' patents belonged, was 'Headwear' and the second largest one was 'Hygiene'. In the subgroup level, the most common largest subgroups were 'Parts, details or accessories of helmets' and 'Masks, shields, or hoods for welders'. Korean patents had the lowest ratio of single_IPC at 11%, but the ratios of single_IPC for US and European patents were 52%, respectively. 'Human necessities', 'Mechanical engineering', 'Physics', and 'Electricity' were used as a single_IPC, but lower levels of 'Performing operations', 'Chemistry', 'Textiles' and 'Fixed constructions' were used as multi_IPC. Most main applicants had patents which belonged to 'Human necessities'. Conclusion: The cases of US patents annually increased unlike others. The common technologies for creating the helmet were Parts of helmets under Headwear and Shields for welders under Hygiene. Many Korean patents tended to be categorized as multi_IPC, different from others. Application: To know the technological characteristic of patents would be helpful to initial stage developers.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.1
• /
• pp.35-41
• /
• 2003

Feature vectors for conventional speech recognition are usually extracted in full frequency band. Therefore, each sub-band contributes equally to final speech recognition results. In this paper, feature Teeters are extracted indepedently in each sub-band. The cepstral resolution of each sub-band feature is controlled for the optimal speech recognition. For this purpose, different dimension of each sub-band ceptral vectors are extracted based on the multi-band approach, which extracts feature vector independently for each sub-band. Speech recognition rates and clustering quality are suggested as the criteria for finding the optimal combination of sub-band Teeter dimension. In the connected digit recognition experiments using TIDIGITS database, the proposed method gave string accuracy of 99.125%, 99.775% percent correct, and 99.705% percent accuracy, which is 38%, 32% and 37% error rate reduction relative to baseline full-band feature vector, respectively.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2007.10a
• /
• pp.574-577
• /
• 2007

While disk can be overwritten on updating data, because flash memory can not be overwritten on updating data, new data are updated in new area. If data are frequently updated, garbage collection, which is achieved by erasing blocks, should be performed to reclaim new area. Hence, because the number of erase operations is limited due to characteristics of flash memory, every block should be evenly written and erased. However, if data with access locality are processed by cost benefit algorithm with separation of hot block and cold block, though the performance of processing is high, wear-leveling is not even. In this paper, we propose CB-MB (Cost Benefit between Multi Bank) algorithm in which hot data are allocated in one bank and cold data in another bank, and in which role of hot bank and cold bank is exchanged every period. CB-MB showed that its performance was similar to that of others for uniform workload, however, the method provides much better performance than that of others for workload of access locality.