【摘要】：語(yǔ)音特征集通常具有較高的維度,高維度的特征集不僅包含噪聲數(shù)據(jù)和冗余數(shù)據(jù),影響情感識(shí)別精度,而且在分類識(shí)別的過程中將花費(fèi)大量的計(jì)算開銷.如何從較高維度的特征集中選擇出規(guī)模更小,性能較優(yōu)的特征子集對(duì)語(yǔ)音情感識(shí)別系統(tǒng)具有重要作用.本文融合過濾式和封裝式兩種篩選策略,提出信息增益與和聲搜索算法相結(jié)合的方法進(jìn)行語(yǔ)音情感特征選擇.試驗(yàn)結(jié)果表明,采用過濾和封裝相結(jié)合的兩步策略進(jìn)行特征選擇,綜合了過濾策略的低時(shí)間開銷和封裝策略的高識(shí)別率的優(yōu)點(diǎn),而且可以選擇出較原始數(shù)據(jù)集維度更低且分類性能較好的特征子集.
[Abstract]:Speech feature sets usually have a higher dimension. The high dimension feature sets not only contain noise data and redundant data which affect the accuracy of emotion recognition but also spend a lot of computational overhead in the process of classification and recognition. How to select a feature subset with smaller scale and better performance from a higher dimension feature set plays an important role in speech emotion recognition system. In this paper, two filtering strategies, filtering and encapsulation, are combined, and a method combining information gain and harmonic search algorithm is proposed to select speech emotion features. The experimental results show that the advantages of low time cost of filtering strategy and high recognition rate of packaging strategy are combined with the two-step strategy of filtering and encapsulation for feature selection. Moreover, we can select a feature subset with lower dimension and better classification performance than the original data set.
【作者單位】： 合肥工業(yè)大學(xué)計(jì)算機(jī)與信息學(xué)院福祉科技實(shí)驗(yàn)室;合肥工業(yè)大學(xué)工業(yè)與裝備技術(shù)研究院;安徽建筑大學(xué)電子與信息工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金面上項(xiàng)目(61370219)資助 國(guó)家高等學(xué)校學(xué)科創(chuàng)新引智計(jì)劃項(xiàng)目(B14025)資助 情感計(jì)算與先進(jìn)智能機(jī)器安徽省重點(diǎn)實(shí)驗(yàn)室開放課題項(xiàng)目(ACAIM160103)資助
【分類號(hào)】：TN912.34

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 張小康;帥建梅;史林;;基于加權(quán)信息增益的惡意代碼檢測(cè)方法[J];計(jì)算機(jī)工程;2010年06期

，

本文編號(hào)：2250424