Search

The number of young native speakers of Taiwanese, the variant of Southern Min spoken in Taiwan, has decreased. Technological advancements such as text-to-speech (TTS) systems could help arrest this decline. The aim of this study was to design a robust tonal phoneme corpus and a speech synthesis system for Modern Literal Taiwanese (MLT). MLT subsyllables were analyzed using phonetics and phonology to establish tonal phoneme models. These robust tonal phoneme models and hidden Markov models were used to construct an MLT TTS synthesis system. Algorithm-based training resulted in 869 balanced sentences containing 12,544 syllables, with each sentence containing an average of 14.4 syllables. In total, 218 sentences, which included rare phonemes, were manually drafted to supplement the corpus. The synthesized phonemes were deemed to have high intelligibility and could be included in the developed TTS system. According to the HTK speech recognition tool, the overall phoneme recognition rate was 96.47%. Testers, who were native Taiwanese speakers, assigned the synthesized sentences a mean opinion score of 4, indicating that they sounded natural. This developed system and the results described herein can inspire future developments in speech technology and computational linguistics. [ABSTRACT FROM AUTHOR]

As the penetration of photovoltaic (PV) solar generation has increased, a significant number of residential and commercial solar PV systems have been installed without meters. The majority of these systems, however, are also not monitored by power system operators. Therefore, the uncertainty of the net load owing to this “invisible” solar power generation has raised additional challenges for power system operation. To reduce the impact of this, we propose a novel method to estimate the total solar power generation in a large region from a small representative subset. The proposed method is capable of capturing all relevant information that could assist in the identification of the representative subsets. Moreover, different optimization algorithms are utilized and evaluated to select the optimal number of clusters and representative subsets. As a case study, the power generation of 166 PV sites in Taiwan was collected and analyzed. The proposed method demonstrates a significant improvement in estimating aggregated power generation compared to other existing methods. [ABSTRACT FROM AUTHOR]

Geoelectric time series (TS) has long been studied for its potential for probabilistic earthquake forecasting, and a recent model (GEMSTIP) directly used the skewness and kurtosis of geoelectric TS to provide Time of Increased Probabilities (TIPs) for earthquakes in several months in future. We followed up on this work by applying the Hidden Markov Model (HMM) on the correlation, variance, skewness, and kurtosis TSs to identify two Hidden States (HSs) with different distributions of these statistical indexes. More importantly, we tested whether these HSs could separate time periods into times of higher/lower earthquake probabilities. Using 0.5-Hz geoelectric TS data from 20 stations across Taiwan over 7 years, we first computed the statistical index TSs, and then applied the Baum-Welch Algorithm with multiple random initializations to obtain a well-converged HMM and its HS TS for each station. We then divided the map of Taiwan into a 16-by-16 grid map and quantified the forecasting skill, i.e., how well the HS TS could separate times of higher/lower earthquake probabilities in each cell in terms of a discrimination power measure that we defined. Next, we compare the discrimination power of empirical HS TSs against those of 400 simulated HS TSs, then organized the statistical significance values from these cellular- level hypothesis testing of the forecasting skill obtained into grid maps of discrimination reliability. Having found such significance values to be high for many grid cells for all stations, we proceeded with a statistical hypothesis test of the forecasting skill at the global level, to find high statistical significance across large parts of the hyperparameter spaces of most stations. We therefore concluded that geoelectric TSs indeed contain earthquake-related information, and the HMM approach to be capable at extracting this information for earthquake forecasting. [ABSTRACT FROM AUTHOR]

Due to worldwide concerns on the utilization of fossil fuels and the resulting increase in global carbon emissions, many governments have begun promoting renewable and green energies as alternatives to fossil energy. Wind power is currently one of the most mature renewable energy technologies. Although Taiwan is located in a region with abundant wind power, this energy industry still faces major challenges in growing the market. This study employs hidden Markov models (HMM) to analyze the development of the Taiwanese wind power industry. Because renewable energy development is determined by multiple time-series-related factors, we use a variation of the HMM, called the hidden Markov model with multiple feature streams (HMM/MFS), to simultaneously consider distinct factors influencing the development trends of renewable energy. The results show that our forecast model can make accurate projections of Taiwan's installed wind power capacity, which can be used to assist industry in making correct investment decisions and to provide recommendations to the government for setting suitable green energy policies. [ABSTRACT FROM AUTHOR]

Geoelectric time series (TS) have long been studied for their potential for probabilistic earthquake forecasting, and a recent model (GEMSTIP) directly used the skewness and kurtosis of geoelectric TS to provide times of increased probability (TIPs) for earthquakes for several months in the future. We followed up on this work by applying the hidden Markov model (HMM) to the correlation, variance, skewness, and kurtosis TSs to identify two hidden states (HSs) with different distributions of these statistical indexes. More importantly, we tested whether these HSs could separate time periods into times of higher/lower earthquake probabilities. Using 0.5 Hz geoelectric TS data from 20 stations across Taiwan over 7 years, we first computed the statistical index TSs and then applied the Baum–Welch algorithm with multiple random initializations to obtain a well-converged HMM and its HS TS for each station. We then divided the map of Taiwan into a 16-by-16 grid map and quantified the forecasting skill, i.e., how well the HS TS could separate times of higher/lower earthquake probabilities in each cell in terms of a discrimination power measure that we defined. Next, we compare the discrimination power of empirical HS TSs against those of 400 simulated HS TSs and then organized the statistical significance values from this cellular-level hypothesis testing of the forecasting skill obtained into grid maps of discrimination reliability. Having found such significance values to be high for many grid cells for all stations, we proceeded with a statistical hypothesis test of the forecasting skill at the global level to find high statistical significance across large parts of the hyperparameter spaces of most stations. We therefore concluded that geoelectric TSs indeed contain earthquake-related information and the HMM approach is capable of extracting this information for earthquake forecasting. [ABSTRACT FROM AUTHOR]

This paper proposes a multimedia English learning (MEL) system, based on Hidden Markov Models (HMMs) and mastery theory strategy, for teaching students with the aim of enhancing their English phonetic awareness and pronunciation. It can analyze phonetic structures, identify and capture pronunciation errors to provide students with targeted advice in pronunciation, intonation, rhythm and volume. In addition, the paper also applies the mastery learning to effectively help students practice pronunciation of English words and sentences. Finally, this paper adopts a quasi-experimental design and lasting for 12 weeks and 120 third-graders, aged 9-10 years, from an elementary school in Yunlin County in Taiwan. These students were recruited and randomly assigned as the experimental group and the control group, respectively. The former used the MEL system, while the latter received the conventional English teaching. Research data were collected through the Phonemic Awareness Test and the English Achievement Test. The results showed that the experimental group with low phonemic awareness performed significantly better than the control group in the English Achievement Test. [ABSTRACT FROM AUTHOR]

Context: Promoting patient value-concordant end-of-life care is a priority in health care systems but has rarely been examined in randomized clinical trials.Objectives: To examine the effectiveness of an advance care planning intervention in facilitating concordance between cancer patients' preferred and received life-sustaining treatment (LST) states and to explore modifiable factors facilitating or impeding such concordance.Methods: Terminal cancer patients (N = 460) were randomly assigned 1:1 to the experimental and control arms of a randomized clinical trial, with 430 deceased participants comprising the final sample. States of preferred LSTs (cardiopulmonary resuscitation, intensive care unit care, chest compression, intubation with mechanical ventilation, intravenous nutrition, and nasogastric tube feeding) and LSTs received in the last month were examined by hidden Markov modeling. Concordance and its modifiable predictors were evaluated by kappa and multivariate logistic regression, respectively.Results: We identified three LST-preference states (uniformly preferring LSTs, rejecting LSTs except intravenous nutrition support, and mixed LST preferences) and three received LST states (uniformly receiving LSTs, received intravenous nutrition only, and selectively receiving LSTs). Concordance was not significantly higher in the experimental than the control arm (kappa [95% CI]: 0.126 [0.032, 0.221] vs. 0.050 [-0.028, 0.128]; arm difference: odds ratio [95% CI]: 1.008 [0.675, 1.5001]). Preferred-received LST-state concordance was facilitated by accurate prognostic awareness, better quality of life, and more depressive symptoms, whereas concordance was impeded by more anxiety symptoms.Conclusions: Our advance care planning intervention did not facilitate concordance between terminally ill cancer patients' preferred and received LST states, but patient value-concordant end-of-life care may be facilitated by interventions to cultivate accurate prognostic awareness, improve quality of life, support depressive patients, and clarify anxious patients' overexpectations of LST efficacy. [ABSTRACT FROM AUTHOR]