Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/19541
標題: 中文轉台語文轉音系統中一詞多音之預測
Solving the Polysemy Problem in a Chinese to Taiwanese TTS System
作者: 林金玉
Lin, Chin-Yu
關鍵字: polysemy
一詞多音
Chinese to Taiwanese TTS system
language model
decision tree and combined strategy
台語文轉音
語言模型
決策樹
組合式策略技術
出版社: 資訊科學與工程學系所
引用: [1] Frederick Jelinek , “Statistical Methods for Speech Recognition”,The MIT Press Cambridge Massachusetts,1997. [2] Ian Witten,Eibe Frank,”Data Mining practical machine learning tools and techniques with JAVA implementations ”,Morgan Kaufmann publishers,1999. [3] Jiawei Han,Micheline Kamber,”Data Mining Concepts and Techniques”, Morgan Kaufmann publishers,1999. [4] Weka 3, http://www.cs.waikato.ac.nz/ml/weka/。 [5] Yih-Jeng Lin, Ming-Shing Yu, and Chiung-Jung Huang,“The Polysemy Problems, an Important Issue in a Chinese to Taiwanese TTS System”,the IEEE proceedings of the International Congress on Image and Signal Processing (CISP2008), May 27-30, Sanya, Hainan, China, 2008。 [6] Yih-Jeng Lin, Ming-Shing Yu, Chin-Yu Lin, and Yuan-Tsun Lin,” A Multi-Layered Approach to the Polysemy Problems in a Chinese to Taiwanese TTS System” The IEEE International Conference on Sensor Networks, Ubiquitous, and Trustworthy Computing, SUTC 2008,June 11-13, 2008 Taichung, Taiwan。 [7] 游子宜,“台灣閩南語一字多音之研究上、下篇”,1997。 [8] 林日祥、林宏基,“台語破音多音詞彙”,2004。 [9] 林元淳,“中文文轉音系統中一詞多音之研究“,中興大學資訊科學研究所碩士論文,2005。 [10] 中研院中文斷詞系統 http://ckipsvr.iis.sinica.edu.tw/。 [11] 中央研究院平衡語料庫,ttp://www.sinica.edu.tw/ftms-bin/kiwi1/mkiwi.sh。 [12] 張子榮、 初敏, “解決多音詞字-音轉換的一種統計學習方法”, 中文信習學報, 2002, 第3期, pp.39-45. [13] 蔡文鴻, “語言模型訓練與調適技術於中文大詞彙連續語音辨識之初步研究”,國立臺灣師範大學資訊工程研究所碩士論文,2005。 [14] 李啓鑫, “資料探勘與統計技術的自由軟體整合介面”,淡江大學統計學所, 2004。 [15] 林震岩, “多變量分析SPSS的操作與應用”,智勝出版社,2007。 [16] 張大平、張一岑,“SPSS統計建模與分析程序”,文魁出版社,2007。
摘要: 一個中文轉台語文轉音系統可以將一連串的中文文字(句子)轉換成台語語音輸出。相對於中文的多音詞現象,台語的一詞多音問題更是嚴重,一詞多音現象會直接影響到輸出的台語文句之瞭解度(Intelligibility)。本研究針對中文轉台語文轉音系統中,一詞多音現象作探討,並利用語言模型、決策樹及貝式分類器來預測中文文句轉台語文句中,詞的正確台語發音。 本研究的目標是希望能夠針對中文文句轉台語文句中所有的一詞多音都能有正確的發音,並且提出一種組合式策略技術結合了語言模型及決策樹並搭配門檻值的設定,實驗的結果大部分都有提昇語言模型、決策樹的正確率,並可以套入現有的TTS系統中使用。
A Chinese to Taiwanese text to speech system (TTS) is a system that can output Taiwanese speech with some input Chinese text. Compared with Mandarin, the polyphone problems in Taiwanese is more complex than that in Mandarin. We call it is the polysemy problem in Taiwanese. We think it is an intent work should be done in a Taiwanese TTS system. This research focuses on the polysemy problems in a Chinese to Taiwanese TTS system. We use language models, decision trees and Bayes classifier to predict the correct pronunciation for words with polyesmy problem in Taiwanese. The main works we done is to predict the correct pronunciation for words with polyesmy problem in Taiwanese. We make good use of the abovementioned three methods. Results show that the combined approaches can achieve higher accuracies in predicting pronunciations of each word with polysemy problems.
URI: http://hdl.handle.net/11455/19541
其他識別: U0005-1808200812320900
文章連結: http://www.airitilibrary.com/Publication/alDetailedMesh1?DocID=U0005-1808200812320900
Appears in Collections:資訊科學與工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.