頁籤選單縮合
題名 | Consonant Classification for Mandarin Syllables=中文字音之聲母分類 |
---|---|
作者 | 林明灶; 李清坤; Lin, Ming-tzaw; Lee, Ching-kuen; |
期刊 | 大同學報 |
出版日期 | 19981100 |
卷期 | 28 1998.11[民87.11] |
頁次 | 頁211-216+461 |
分類號 | 312.23 |
語文 | eng |
關鍵詞 | 中文字音辨認系統; 聲母分類演繹法; |
中文摘要 | 本篇論文中,主要發展中文字音辨認系統中之聲母分類演繹法。我們的研究主題包括聲母類別分配、語音參數之選擇與分類規則之設計。我們將21組之聲母分配於七大類,並使用聲音之平均能量、長度區間、平均過零率、週期數及高頻頻帶能量與低頻頻帶能量的比值,作為聲母分類演繹法中語音參數之選擇。基於我們的實驗資料,將每一聲母群組聲音特性,建構於鈴形之歸屬函數。因此,模糊理論作為聲母分類演繹法之決策規則。透過我們的實驗結果,在前一候選之正確分類率是87.2%,前二候選之正確分類率平均達到98.44%。 |
英文摘要 | In this paper, we developed an algorithm for consonant classification, which can be used as a subsystem for Mandarin syllable recognition. Our main research topics included: class assignment, selection of acoustic features, and the design of classification decision rules. The total 21 Mandarin consonants were assigned to seven classes. The averaged energy, duration, zero-crossing rate, pitch period, and energy ratio between low and high frequency bands were selected as the acoustic features for our consonant classification algorithm. Based on our experimental data, a bell-shaped membership function was constructed for each acoustic feature of every consonant class. Then, a fuzzy rule based algorithm was designed for decision making of consonant classification. The experiment results showed that the proposed algorithm has an average correct percentage of 87.2% for the first selection, and can achieve an average correct percentage of 98.44% for the top two selections. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。