• Proceedings of the Korean Society for Language and Information Conference
• /
• 2007.11a
• /
• pp.241-248
• /
• 2007

In this paper, we propose a Transformation-Based Learning (TBL) method on generating the Korean standard pronunciation. Previous studies on the phonological processing have been focused on the phonological rule applications and the finite state automata (Johnson 1984; Kaplan and Kay 1994; Koskenniemi 1983; Bird 1995). In case of Korean computational phonology, some former researches have approached the phonological rule based pronunciation generation system (Lee et al. 2005; Lee 1998). This study suggests a corpus-based and data-oriented rule learning method on generating Korean standard pronunciation. In order to substituting rule-based generation with corpus-based one, an aligned corpus between an input and its pronunciation counterpart has been devised. We conducted an experiment on generating the standard pronunciation with the TBL algorithm, based on this aligned corpus.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.2E
• /
• pp.43-50
• /
• 2009

To get more natural synthetic speech generated by a Korean TTS (Text-To-Speech) system, we have to know all the possible prosodic rules in Korean spoken language. We should find out these rules from linguistic, phonetic information or from real speech. In general, all of these rules should be integrated into a prosody-generation algorithm in a TTS system. But this algorithm cannot cover up all the possible prosodic rules in a language and it is not perfect, so the naturalness of synthesized speech cannot be as good as we expect. ANNs (Artificial Neural Networks) can be trained to learn the prosodic rules in Korean spoken language. To train and test ANNs, we need to prepare the prosodic patterns of all the phonemic segments in a prosodic corpus. A prosodic corpus will include meaningful sentences to represent all the possible prosodic rules. Sentences in the corpus were made by picking up a series of words from the list of PB (phonetically Balanced) isolated words. These sentences in the corpus were read by speakers, recorded, and collected as a speech database. By analyzing recorded real speech, we can extract prosodic pattern about each phoneme, and assign them as target and test patterns for ANNs. ANNs can learn the prosody from natural speech and generate prosodic patterns of the central phonemic segment in phoneme strings as output response of ANNs when phoneme strings of a sentence are given to ANNs as input stimuli.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.8
• /
• pp.341-354
• /
• 2023

The title of a document is the brief summarization of the document. Readers can easily understand a document if we provide them with its title in their preferred styles and the languages. In this research, we propose a cross-lingual and style-based title generation model using multiple adapters. To train the model, we need a parallel corpus in several languages with different styles. It is quite difficult to construct this kind of parallel corpus; however, a monolingual title generation corpus of the same style can be built easily. Therefore, we apply a zero-shot strategy to generate a title in a different language and with a different style for an input document. A baseline model is Transformer consisting of an encoder and a decoder, pre-trained by several languages. The model is then equipped with multiple adapters for translation, languages, and styles. After the model learns a translation task from parallel corpus, it learns a title generation task from monolingual title generation corpus. When training the model with a task, we only activate an adapter that corresponds to the task. When generating a cross-lingual and style-based title, we only activate adapters that correspond to a target language and a target style. An experimental result shows that our proposed model is only as good as a pipeline model that first translates into a target language and then generates a title. There have been significant changes in natural language generation due to the emergence of large-scale language models. However, research to improve the performance of natural language generation using limited resources and limited data needs to continue. In this regard, this study seeks to explore the significance of such research.

• Phonetics and Speech Sciences
• /
• v.4 no.4
• /
• pp.29-35
• /
• 2012

The purpose of this study was to measure the first two vowel formants of the forty male and female speakers (twenty young vs. old male speakers and twenty young vs. old female speakers) from the Buckeye Corpus of Conversational Speech and to examine the vowel formant changes across two generations (younger vs. older). The results indicated that the vowel space of the younger generation (in their thirties or less) shifted to the lower left position compared to those of the older generation (in their forties or more) in both male and female speakers. When the results were compared to those of Peterson & Barney (1952), it appears that differences can be found in the size of the vowel spaces through time.

• Korean Journal of Cognitive Science
• /
• v.27 no.1
• /
• pp.27-41
• /
• 2016

Named entity recognition is to classify elements in text into predefined categories and used for various departments which receives natural language inputs. In this paper, we propose a method which can generate named entity training corpus automatically using knowledge bases. We apply two different methods to generate corpus depending on the knowledge bases. One of the methods attaches named entity labels to text data using Wikipedia. The other method crawls data from web and labels named entities to web text data using Freebase. We conduct two experiments to evaluate corpus quality and our proposed method for generating Named entity recognition corpus automatically. We extract sentences randomly from two corpus which called Wikipedia corpus and Web corpus then label them to validate both automatic labeled corpus. We also show the performance of named entity recognizer trained by corpus generated in our proposed method. The result shows that our proposed method adapts well with new corpus which reflects diverse sentence structures and the newest entities.

• Journal of information and communication convergence engineering
• /
• v.11 no.4
• /
• pp.268-273
• /
• 2013

In this paper, we propose a document classification model using Web documents as a part of the training corpus in order to resolve the imbalance of the training corpus size per category. For the purpose of retrieving the Web documents closely related to each category, the proposed document classification model calculates the matching score between word features and each category, and generates a Web search query by combining the higher-ranked word features and the category title. Then, the proposed document classification model sends each combined query to the open application programming interface of the Web search engine, and receives the snippet results retrieved from the Web search engine. Finally, the proposed document classification model adds these snippet results as Web documents to the training corpus. Experimental results show that the method that considers the balance of the training corpus size per category exhibits better performance in some categories with small training sets.

• Journal of Software Assessment and Valuation
• /
• v.7 no.1
• /
• pp.29-39
• /
• 2011

To evaluate the quality of clone detection tools, we should know how many clones the tool misses. Hence we need to have the standard code-clone reference corpus for a carefully chosen set of sample source codes. The reference corpus available so far has been built by manually collecting clones from the results of various existing tools. This paper presents a tree-pattern-based clone detection tool that can be used for automatic generation of reference corpus. Our tool is compared with CloneDR for precision and Bellon's reference corpus for recall. Our tool finds no false positives and 2 to 3 times more clones than CloneDR. Compared to Bellon's reference corpus, our tools shows the 93%-to-100% recall rate and detects far more clones.

• ETRI Journal
• /
• v.39 no.4
• /
• pp.455-466
• /
• 2017

This study focuses on a method for sequential data augmentation in order to alleviate data sparseness problems. Specifically, we present corpus expansion techniques for enhancing the coverage of a language model. Recent recurrent neural network studies show that a seq2seq model can be applied for addressing language generation issues; it has the ability to generate new sentences from given input sentences. We present a method of corpus expansion using a sentence-chain based seq2seq model. For training the seq2seq model, sentence chains are used as triples. The first two sentences in a triple are used for the encoder of the seq2seq model, while the last sentence becomes a target sequence for the decoder. Using only internal resources, evaluation results show an improvement of approximately 7.6% relative perplexity over a baseline language model of Korean text. Additionally, from a comparison with a previous study, the sentence chain approach reduces the size of the training data by 38.4% while generating 1.4-times the number of n-grams with superior performance for English text.

• MALSORI
• /
• no.53
• /
• pp.75-92
• /
• 2005

Prosody Generation Based on C-ToBI Representation for Text-to-SpeechSeungwon Kim, Yu Zheng, Gary Geunbae Lee, Byeongchang KimProsody modeling is critical in developing text-to-speech (TTS) systems where speech synthesis is used to automatically generate natural speech. In this paper, we present a prosody generation architecture based on Chinese Tone and Break Index (C-ToBI) representation. ToBI is a multi-tier representation system based on linguistic knowledge to transcribe events in an utterance. The TTS system which adopts ToBI as an intermediate representation is known to exhibit higher flexibility, modularity and domain/task portability compared with the direct prosody generation TTS systems. However, the cost of corpus preparation is very expensive for practical-level performance because the ToBI labeled corpus has been manually constructed by many prosody experts and normally requires a large amount of data for accurate statistical prosody modeling. This paper proposes a new method which transcribes the C-ToBI labels automatically in Chinese speech. We model Chinese prosody generation as a classification problem and apply conditional Maximum Entropy (ME) classification to this problem. We empirically verify the usefulness of various natural language and phonology features to make well-integrated features for ME framework.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.5
• /
• pp.489-495
• /
• 2022

It makes difference to LSTM D/L(Deep Learning) results for language model construction as the corpus preprocess changes. An LSTM model was trained with a famouse literaure poems(Ki Hyung-do's work) for training corpus in the study. You get the two wordvector sets for two corpus sets of the original text and eraised word ending text each once D/L training completed. It's been inspected of the similarity/analogy operation results, the positions of the wordvectors in 2D plane and the generated texts by the language models for the two different corpus sets. The suggested words by the silmilarity/analogy operations are changed for the corpus sets but they are related well considering the corpus characteristics as a literature work. The positions of the wordvectors are different for each corpus sets but the words sustained the basic meanings and the generated texts are different for each corpus sets also but they have the taste of the original style. It's supposed that the D/L language model can be a useful tool to enjoy the literature in object and in diverse with the analysis results shown in the study.