發(fā)布時(shí)間：2018-11-28 16:01

【摘要】：語音作為人們交流和表達(dá)情感的一種重要媒介,在日常生活中卻總是受到噪聲的干擾,因此我們需要對(duì)混入背景噪聲的干凈語音進(jìn)行語音增強(qiáng)。語音增強(qiáng)算法的最終目標(biāo)就是對(duì)背景噪聲進(jìn)行抑制,改善語音聽覺質(zhì)量,同時(shí)保證一定的語音可懂度。人們對(duì)語音增強(qiáng)算法的研究已有半個(gè)多世紀(jì)的歷史,這期間涌現(xiàn)過很多經(jīng)典的語音增強(qiáng)算法,如譜減法、維納濾波法、幅度譜最小均方誤差算法等,且一直為人們所研究。這些算法在高信噪比平穩(wěn)噪聲下,通�？梢匀〉昧己玫恼Z音增強(qiáng)效果,但是在低信噪比非平穩(wěn)噪聲下,語音增強(qiáng)效果卻不盡人意,還有很多需要攻克的難題。所以,在低信噪比非平穩(wěn)噪聲場(chǎng)景下對(duì)帶噪語音信號(hào)進(jìn)行語音增強(qiáng)仍是當(dāng)前國(guó)內(nèi)外學(xué)者研究的一個(gè)熱點(diǎn)。本文主要針對(duì)對(duì)數(shù)譜最小均方誤差(Log-Spectral Amplitude Minimum Mean-Square Error,LSA-MMSE)算法以及信號(hào)子空間算法在低信噪比場(chǎng)景下存在的缺陷提出改進(jìn)。主要研究工作如下:首先,提出了低信噪比場(chǎng)景下改進(jìn)的LSA-MMSE算法。針對(duì)傳統(tǒng)LSA-MMSE算法在強(qiáng)噪聲環(huán)境下語音信息完整保留效果不佳,本文將Loizou等人提出的大部分語音增強(qiáng)算法對(duì)帶噪語音進(jìn)行增強(qiáng)處理后普遍存在兩種不同類型失真,這一理論應(yīng)用到LSA-MMSE算法中�；谶@一理論對(duì)LSA-MMSE算法提出了改進(jìn)。以往學(xué)者總是將區(qū)域Ⅰ的衰減失真和區(qū)域Ⅱ小于或等于6.02dB的放大失真所對(duì)應(yīng)的幅度譜歸為一類處理,認(rèn)為這樣不會(huì)對(duì)語音信息的完整保留造成影響,研究表明這樣反而會(huì)產(chǎn)生更多殘留噪聲�；谶@一點(diǎn),本文對(duì)衰減失真對(duì)應(yīng)的幅度譜、小于等于6.02dB放大失真對(duì)應(yīng)的幅度譜、大于6.02dB放大失真所對(duì)應(yīng)的幅度譜分別采取不同程度的向下約束。另外,低信噪比場(chǎng)景下先驗(yàn)信噪比和增益函數(shù)的估計(jì)誤差對(duì)語音增強(qiáng)效果有很大影響,改進(jìn)的LSA-MMSE算法中分別對(duì)它們進(jìn)行了調(diào)整。實(shí)驗(yàn)結(jié)果表明,低信噪比場(chǎng)景下本文算法更好地保留了語音的主要信息,同時(shí)有效抑制了低頻部分的背景噪聲。其次,提出了低信噪比場(chǎng)景下改進(jìn)的信號(hào)子空間語音增強(qiáng)算法。子空間算法有著良好的去噪效果,但在低信噪比環(huán)境下仍然殘留較多噪聲。本文首先把濾除小于零的特征值及與之對(duì)應(yīng)的特征向量,這一方法應(yīng)用到傳統(tǒng)子空間算法中,以達(dá)到優(yōu)化信號(hào)子空間的效果。同時(shí)提出使用共享正弦多窗譜的協(xié)方差估計(jì)方法減小估計(jì)誤差和計(jì)算復(fù)雜度。最后對(duì)估計(jì)的干凈語音引入維納濾波函數(shù)進(jìn)行修正。實(shí)驗(yàn)結(jié)果表明,在5種常見噪聲的低信噪比場(chǎng)景下,改進(jìn)算法能有效去除背景噪聲,改善語音聽覺質(zhì)量,其語音增強(qiáng)效果整體優(yōu)于改進(jìn)前的算法。
[Abstract]:As an important medium for people to communicate and express their emotions, speech is always disturbed by noise in daily life. Therefore, we need to enhance the voice of clean speech mixed with background noise. The final goal of speech enhancement algorithm is to suppress background noise, improve the quality of speech hearing, and ensure a certain degree of speech intelligibility. Speech enhancement algorithms have been studied for more than half a century. During this period, many classical speech enhancement algorithms have emerged, such as spectral subtraction, Wiener filter, amplitude spectrum minimum mean square error algorithm and so on. These algorithms can usually achieve good speech enhancement effect under high SNR stationary noise, but in low SNR non-stationary noise, the speech enhancement effect is not satisfactory, and there are still many difficult problems to be solved. Therefore, speech enhancement of noisy speech signal in low SNR non-stationary noise scene is still a hot research topic at home and abroad. This paper focuses on the improvement of the logarithmic spectrum minimum mean square error (Log-Spectral Amplitude Minimum Mean-Square Error,LSA-MMSE) algorithm and the signal subspace algorithm in low SNR scenarios. The main research work is as follows: firstly, an improved LSA-MMSE algorithm in low SNR scenario is proposed. Because the traditional LSA-MMSE algorithm can not preserve the speech information completely in the environment of strong noise, there are two different types of distortion after most of the speech enhancement algorithms proposed by Loizou et al are used to enhance the noisy speech. This theory is applied to LSA-MMSE algorithm. Based on this theory, the LSA-MMSE algorithm is improved. In the past, the attenuation distortion of region I and the amplitudes of region 鈪，

本文編號(hào)：2363391