• Phonetics and Speech Sciences
• /
• v.16 no.1
• /
• pp.67-76
• /
• 2024

Recent advances in text-to-speech (TTS) technology have significantly improved the quality of synthesized speech, reaching a level where it can closely imitate natural human speech. Especially, TTS models offering various voice characteristics and personalized speech, are widely utilized in fields such as artificial intelligence (AI) tutors, advertising, and video dubbing. Accordingly, in this paper, we propose a one-shot multi-speaker TTS system that can ensure acoustic diversity and synthesize personalized voice by generating speech using unseen target speakers' utterances. The proposed model integrates a speaker encoder into a TTS model consisting of the FastSpeech2 acoustic model and the HiFi-GAN vocoder. The speaker encoder, based on the pre-trained RawNet3, extracts speaker-specific voice features. Furthermore, the proposed approach not only includes an English one-shot multi-speaker TTS but also introduces a Korean one-shot multi-speaker TTS. We evaluate naturalness and speaker similarity of the generated speech using objective and subjective metrics. In the subjective evaluation, the proposed Korean one-shot multi-speaker TTS obtained naturalness mean opinion score (NMOS) of 3.36 and similarity MOS (SMOS) of 3.16. The objective evaluation of the proposed English and Korean one-shot multi-speaker TTS showed a prediction MOS (P-MOS) of 2.54 and 3.74, respectively. These results indicate that the performance of our proposed model is improved over the baseline models in terms of both naturalness and speaker similarity.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.9
• /
• pp.2280-2288
• /
• 1997

An agent is a software element that, by making use of knowledge and inference, performs tasks on behalf of the user. In general, an agent has the properties of autonomy, social ability, reactivity, and durability. Many researches on agents are more and more aiming at the multiagent systems since it is not sufficient to let a single agent do the whole things, especially in a real world where tasks require many diverse activities. However, the multiagent frameworks still have some limitations in the processing of user queries that are often ambiguous and goal-oriented. Also, a series of procedures or plans could not be generated from a single query directly. In order to give more intelligence to the multiagent framework, we propose a method of extending the framework with a plan generation module. The open agent architecture (OAA), which is a multiagent framework that we developed, is integrated with UCPOP, which is a AI planner. A travel schedule management agent (TSMA) system is implemented to explore the effects of the method. The extended system enables the user to only specify goal-oriented queries, and the plans and procedures to satisfy these goals are generated automatically. Also, this system provides a cooperative and knowledge-sharing environment that integrates several knowledge-based systems and planning systems that are distributed and used independently.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.9 no.6 s.46
• /
• pp.13-19
• /
• 2005

Floor response spectra for dynamic response of subsystem such as equipment, or piping in nuclear power plants are usually generated without considering dynamic interaction between main structure and subsystem. This study describes the analytic method in which equipment response spectra can be obtained through dynamic analysis considering equipment-structure Interaction(ESI). In this method, dynamic response of the equipment by this method is based on a dynamic substructure method in which the equipment-structure system is partitioned into the single-degree-ol-freedom system(SDOF) representing the equipment and the equipment support impedance representing the dynamic charactenstics of the structure ai the equipment support. A family of equipment response spectra is developed by applying this method to calculate the maximum responses of a family of SDOF equipment systems with wide banded equipment frequency, damping ratio, and mass. The method is validated by comparing the floor response spectrum from this method with the floor response spectrum generated from the rigorous analysis including equipments on the containment building of a prototypical nuclear power plant. in order to Investigate ESI effect in the response of equipment, response values from the method and the conventional approach without considering ESI are compared for the equipment having the mass less than 1% of the total structural mass. Response spectra from the method showed lower spectral amplitudes than those of the conventional floor response spectra around controlling frequencies.

• The Journal of Bigdata
• /
• v.5 no.2
• /
• pp.77-84
• /
• 2020

In this study, we conducted a study on the win-loss predicton analysis of korean professional baseball by artificial intelligence models. Based on the model, we predicted the winner as well as each team's final rank in the league. Additionally, we developed a website for viewers' understanding. In each game's first, third, and fifth inning, we analyze to select the best model that performs the highest accuracy and minimizes errors. Based on the result, we generate the rankings. We used the predicted data started from May 5, the season's opening day, to August 30, 2020 to generate the rankings. In the games which Kia Tigers did not play, however, we used actual games' results in the data. KNN and AdaBoost selected the most optimized machine learning model. As a result, we observe a decreasing trend of the predicted results' ranking error as the season progresses. The deep learning model recorded 89% of the model accuracy. It provides the same result of decreasing ranking error trends of the predicted results that we observe in the machine learning model. We estimate that this study's result applies to future KBO predictions as well as other fields. We expect broadcasting enhancements by posting the predicted winning percentage per inning which is generated by AI algorism. We expect this will bring new interest to the KBO fans. Furthermore, the prediction generated at each inning would provide insights to teams so that they can analyze data and come up with successful strategies.

• Journal of the Society of Disaster Information
• /
• v.19 no.2
• /
• pp.344-351
• /
• 2023

Purpose: This study aimed to evaluate the effectiveness of a dielectric barrier discharge (DBD) plasma module for sterilizing airborne bacteria in indoor spaces and measure the concentration of ozone generated during plasma discharge. Method: The DBD plasma module was installed in a 76m³ space, and air samples were collected under various discharge times to compare the reduction of airborne bacteria. Result: The results showed a significant decrease in airborne bacteria, ranging from 92.057% to 99.999%, with an average ozone concentration of 0.04 ppm, below the reference value. Conclusion: The study suggests that plasma discharge can be used as a means of preventing the spread of airborne bacteria and viruses, while ensuring safety for human exposure.

• Korean Journal of Materials Research
• /
• v.14 no.1
• /
• pp.13-18
• /
• 2004

Titanium alloys are the one of promising candidate materials for medium high temperature parts in the aircraft, automobile, petrochemistry and electrochemistry because of their high strength with low density in medium high temperature. In this study, the effects of aging and heat treatments on the mechanical properties of Ti-15-3 alloy in medium high temperature, which was $400^{\circ}C$, were studied. Solid solution treatment was performed at $8000^{\circ}C$ of $\beta$ phase region for 1 h and the alloy was quenched in water. The alloy was aged at $5000^{\circ}C$ of $\alpha$ and $\beta$ two-phase region for 1, 2, 4, 8, ... and 100 h to increase the mechanical property. The $\beta$ single phase was observed at all parts of specimens in Ti-15-3 alloy after ST. As the aging at $500^{\circ}C$, fine precipitates of a phase was generated from matrix of $\beta$ phase and the microstructure was consisted of weaving structure such as Widmanstiitten a phase. The most suitable aging time is 24h in$ 400^{\circ}C$. At this time, strength is 1164 MPa and elongation is about 12%. In room temperature, elongation of Ti-15-3 alloy aged at $500^{\circ}C$ for 16 h is poor (=3%) in spite of high tensile strength (1458 MPa).

• Korean Journal of Cognitive Science
• /
• v.23 no.2
• /
• pp.205-224
• /
• 2012

How would people who buy an insurance policy respond to a low probability risk with a high future cost? Presented with a scenario describing a low probability accident of a chemical plant, participants in four experiments were asked to rate their perception of the risk and also their intention to buy an insurance of a given premium, an insurance, or a ratio insurance. Participants differently responded only to ratio insurance when rating their perception of risk, not to either premium or insurance. The pattern of results in four experiments converged to the conclusion that ratio insurance, an ecologically valid cue, makes people sensitive to the level of risk expressed in low probabilities of an accident. Our results were consistent with the prediction generated by the ecological cue hypothesis which empathizes the importance of frequency over probability in risk perception (Gigerenzer, 2000).

• Proceedings of the Korean Society of Developmental Biology Conference
• /
• 2003.10a
• /
• pp.7-8
• /
• 2003

Since it was first reported in 1997, somatic cell cloning has been demonstrated in several other mammalian species. On the mouse, it can be cloned from embryonic stem (ES) cells, fetus-derived cells, and adult-derived cells, both male and female. While cloning efficiencies range from 0 to 20%, rates of just 1-2% are typical (i.e. one or two live offspring per one hundred initial embryos). Recently, abnormalities in mice cloned from somatic cells have been reported, such as abnormal gene expression in embryo (Boiani et al., 2001, Bortvin et al., 2003), abnormal placenta (Wakayama and Yanagimachi 1999), obesity (Tamashiro et ai, 2000, 2002) or early death (Ogonuki et al., 2002). Such abnormalities notwithstanding, success in generating cloned offspring has opened new avenues of investigation and provides a valuable tool that basic research scientists have employed to study complex processes such as genomic reprogramming, imprinting and embryonic development. On the other hand, mouse ES cell lines can also be generated from adult somatic cells via nuclear transfer. These 'ntES cells' are capable of differentiation into an extensive variety of cell types in vitro, as well assperm and oocytes in vivo. Interestingly, the establish rate of ntES cell line from cloned blastocyst is much higher than the success rate of cloned mouse. It is also possible to make cloned mice from ntES cell nuclei as donor, but this serial nuclear transfer method could not improved the cloning efficiency. Might be ntES cell has both character between ES cell and somatic cell. A number of potential agricultural and clinical applications are also are being explored, including the reproductive cloning of farm animals and therapeutic cloning for human cell, tissue, and organ replacement. This talk seeks to describe both the relationship between nucleus donor cell type and cloning success rate, and methods for establishing ntES cell lines. (중략)

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.8
• /
• pp.873-877
• /
• 2010

CAPTCHA is a security tool that prevents the automatic sign-up by a spam or a robot. This CAPTCHA usually depends on the smart readability of humans. However, the common and plain CAPTCHA with text-based system is not difficult to be solved by intelligent web-bot and machine learning tools. In this paper, we propose a new sub-image based CAPTCHA system totally different from the text based system. Our system offers a set of cropped sub-image from a whole digital picture and asks user to identify the correct orientation. Though there are some nice machine learning tools for this job, but they are useless for a cropped sub-images, which was clearly revealed by our experiment. Experiment showed that our sub-image based CAPTCHA is easy to human solver, but very hard to all kinds of machine learning or AI tools. Also our CAPTCHA is easy to be generated automatical without any human intervention.

• Applied Chemistry for Engineering
• /
• v.16 no.1
• /
• pp.74-80
• /
• 2005

The effective removal of turbid-inducing particulates and algae-inducing phosphorous was systematically investigated by the variation of physico-chemical parameters such as pH, alkalinity, and coagulant types. Al(III)-based and Fe(III)-based coagulants exhibited high removal efficiency of turbidity and phosphorous at optimal pH ranges of 7~9, in which zeta potential nearly approached to zero. The removal rate of turbidity rapidly increased with the increase of coagulant dosages, whereas the removal rate of phosphorous gradually increased due to an equivalent reaction of phosphorous with metallic ions. The generation of flocs during coagulation was visualized by high speed camera (Motion Scope 2000, Redlake Co.), and the images of singular flocs were captured by optical microscope. The flocs generated by Fe(III)-based coagulant was more compact than those induced by Al(III)-based coagulant, and the settlabiltiy of Fe(III)-induced flocs was superior to that of Al(III)-induced flocs.