查詢結果分析
來源資料
相關文獻
- A Fuzzy Tying Technique for Mixture Autoregressive HMMs
- 以連續型隱藏式馬可夫模型來計算中文簽名之動態相似度值
- An Overview of RNN-Based Mandarin Speech Recognition Approaches
- 以色彩、紋理、和外形為內容之影像擷取方法
- 基於VQ/HMM之國語文句翻語音中音節音長與振幅參數產生之方法
- An Overview of Mandarin-Speech Tone Recognition
- A Survey on Automatic Speech Recognition with an Illustrative Example on Continuous Speech Recognition of Mandarin
- 中文連續音節辨識之研究
- 隱藏式馬可夫模型介紹
- 隱藏式馬可夫模型的基本概念
頁籤選單縮合
題 名 | A Fuzzy Tying Technique for Mixture Autoregressive HMMs=隱藏式馬可夫模型之模糊連結法則 |
---|---|
作 者 | 洪偉文; 林義楠; | 書刊名 | 明志學報 |
卷 期 | 38:1 民95.06 |
頁 次 | 頁7-11 |
分類號 | 312.85 |
關鍵詞 | 語言辨識系統; 隱藏式馬可夫模型; 特徵向量平均值和變異量; 模糊連結法則; Fuzzy tying technique; Gaussian mixture autoregressive density; Fuzzy membership function; Hidden Markov model; HMM; |
語 文 | 英文(English) |
中文摘要 | 語音辨識系統的建構中,隱藏式馬可夫模型經常被用來描述某個語音狀態之特徵向量的統計分布。然而,各個語音狀態又可以再細分成許多子狀態,而且每一個子狀態各自有自己的特徵向量平均值和變異量。為了簡化語音辨識過程中之計算量,因此,有必要將某個狀態內部之各個子狀態其特徵向量平均值(或變異量)相互連結成一個具有代表性之特徵向量平均值(或變異量)。本篇論文提出一種「模糊連結法則」,藉以連結各個子狀態之特徵向量平均值(或變異量),為每一個語音狀態找出一個最具有代表性的特徵向量平均值(或變異量)。實驗結果證明本篇論文所提出之「模糊連結法則」能夠有效強化語音辨識系統的準確性。 |
英文摘要 | In this paper, a fuzzy tying technique is developed to construct a framework for quantitavely formulating the uncertainty involved in the tying operation of Gaussian mixture autoregressive hidden Markov models (HMMs). For the proposed technique, the observation density in each Markov state is simply characterized by the convex combination of Gaussian mixture autoregressive densities that are weighted by a fuzzy membership function. By properly adjusting the fuzzy factor, we can achieve various extents of tying effect. Experimental results for recognition of continuous Mandarin telephone speech indicate that the fuzzy tying technique is useful in enhancing the robustness of HMM-based speech models. |
本系統中英文摘要資訊取自各篇刊載內容。