【摘要】：針對(duì)非結(jié)構(gòu)噪聲難以去除的問(wèn)題,基于字典訓(xùn)練和稀疏表示提出一種無(wú)監(jiān)督語(yǔ)音增強(qiáng)算法。該算法通過(guò)構(gòu)造過(guò)完備字典并使用帶噪語(yǔ)音樣本對(duì)其進(jìn)行訓(xùn)練來(lái)實(shí)現(xiàn)。首先指出K-奇異值分解算法(K-SVD)存在的不足并提出一種新的改進(jìn)的字典訓(xùn)練算法:K-雙邊隨機(jī)投影算法(K-BRP);然后使用K-BRP算法不斷更新字典矩陣和相應(yīng)的增益系數(shù)矩陣,從被非結(jié)構(gòu)化噪聲所污染的帶噪語(yǔ)音中提取出結(jié)構(gòu)性強(qiáng)的純凈語(yǔ)音。大量實(shí)驗(yàn)結(jié)果表明,由于訓(xùn)練樣本考慮到了語(yǔ)音信號(hào)的時(shí)頻域局部結(jié)構(gòu)特征,所提算法能夠很好地消除隨機(jī)噪聲,并且在低信噪比情況下仍然能夠保持較高的語(yǔ)音質(zhì)量和可懂度。
[Abstract]:An unsupervised speech enhancement algorithm based on dictionary training and sparse representation is proposed. The algorithm is implemented by constructing a complete dictionary and using noisy speech samples to train it. Firstly, the shortcomings of the K-SVD algorithm (K-SVD) are pointed out, and a new improved dictionary training algorithm: the K-bilateral Random projection algorithm (K-BRP) is proposed. Then the dictionary matrix and the corresponding gain coefficient matrix are updated by K-BRP algorithm to extract the strong structural pure speech from the noisy speech contaminated by unstructured noise. A large number of experimental results show that the proposed algorithm can eliminate random noise very well because the training samples take into account the local structural characteristics of speech signals in time and frequency domain. And it can still maintain high speech quality and intelligibility under low SNR.
【作者單位】： 解放軍理工大學(xué)指揮信息系統(tǒng)學(xué)院;
【基金】：江蘇省自然科學(xué)基金資助項(xiàng)目(BK2012510)
【分類(lèi)號(hào)】：TN912.3

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 劉曉山;付國(guó)蘭;;基于脊波變換的圖像壓縮[J];電腦與信息技術(shù);2007年02期

2 劉曉山;付國(guó)蘭;;基于脊波變換和SPIHT算法相結(jié)合的圖像壓縮[J];江西師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2007年06期

3 王華丹;劉海林;;稀疏盲源分離問(wèn)題的恢復(fù)性研究[J];廣東工業(yè)大學(xué)學(xué)報(bào);2008年02期

4 談華f，

本文編號(hào)：2317577