【摘要】：聲紋識(shí)別技術(shù)是一種生物認(rèn)證方法,它從說(shuō)話人的語(yǔ)音中提取出能反映說(shuō)話人生理和行為個(gè)性的特征,再結(jié)合模式識(shí)別的理論,來(lái)判斷說(shuō)話人身份。本文主要針對(duì)基于移動(dòng)終端的聲紋識(shí)別系統(tǒng)的相關(guān)技術(shù)進(jìn)行了研究。在語(yǔ)音端點(diǎn)檢測(cè)方面,本文提出了改進(jìn)的能量-過(guò)零率兩級(jí)融合端點(diǎn)檢測(cè)法,該方法與傳統(tǒng)的能量-過(guò)零率端點(diǎn)檢測(cè)法不同,它可以將能量檢測(cè)和過(guò)零檢測(cè)分開(kāi)操作,使這兩種檢測(cè)的結(jié)果同時(shí)進(jìn)行又互不影響,從而實(shí)現(xiàn)多線程并行計(jì)算。此外,改進(jìn)的能量-過(guò)零率端點(diǎn)檢測(cè)法在檢測(cè)中運(yùn)用的是單門(mén)限,相對(duì)于傳統(tǒng)算法,改進(jìn)算法可將閾值參數(shù)減少一半,使算法過(guò)程更加簡(jiǎn)單。針對(duì)空間資源有限的移動(dòng)終端,本文將改進(jìn)算法與常用的單閾值能量檢測(cè)法進(jìn)行對(duì)比,發(fā)現(xiàn)運(yùn)用改進(jìn)算法的聲紋識(shí)別系統(tǒng)的識(shí)別率更高。因此,改進(jìn)的能量-過(guò)零率兩級(jí)融合端點(diǎn)檢測(cè)法在移動(dòng)終端上具有很高的應(yīng)用價(jià)值。針對(duì)傳統(tǒng)語(yǔ)音幀投票法無(wú)法突出每一幀語(yǔ)音判決結(jié)果的差異性的問(wèn)題,本文提出了基于似然概率的的加權(quán)投票法。此方法根據(jù)不同語(yǔ)音幀與概率模型之間的似然概率取值,對(duì)每一幀語(yǔ)音進(jìn)行加權(quán),使得似然概率大的語(yǔ)音幀權(quán)重更大,置信度更高,從而增強(qiáng)每幀語(yǔ)音判決結(jié)果之間的差異,使語(yǔ)音幀融合結(jié)果更準(zhǔn)確。同時(shí),通過(guò)多次的加權(quán)檢測(cè),本文驗(yàn)證了基于加權(quán)投票法的聲紋識(shí)別系統(tǒng)比基于傳統(tǒng)投票法的識(shí)別系統(tǒng)識(shí)別性能更優(yōu)。最后,本文設(shè)計(jì)了多種特征提取技術(shù)以及概率模型的組合方案,通過(guò)實(shí)際識(shí)別效果和算法復(fù)雜度的角度來(lái)分析它們?cè)谝苿?dòng)終端上的可行性,選出最可行的方案。并且根據(jù)最優(yōu)的聲紋識(shí)別系統(tǒng)方案,設(shè)計(jì)了一種基于移動(dòng)終端的聲紋識(shí)別系統(tǒng),并在MATLAB平臺(tái)上實(shí)現(xiàn)了該系統(tǒng),該系統(tǒng)可實(shí)現(xiàn)聲紋采集、模型訓(xùn)練、聲紋識(shí)別、聲紋注冊(cè)、聲紋確認(rèn)等功能。目前,該系統(tǒng)已經(jīng)成功移植于Android系統(tǒng)當(dāng)中。
[Abstract]:Voiceprint recognition is a biometric authentication method, which extracts the characteristics that reflect the speaker's physiological and behavioral personality from the speaker's speech, and then combines the theory of pattern recognition to judge the speaker's identity. This paper mainly focuses on the related technology of voiceprint recognition system based on mobile terminal. In the aspect of speech endpoint detection, this paper presents an improved two-stage fusion endpoint detection method with energy-zero crossing rate. This method is different from the traditional energy-zero-crossing rate endpoint detection method, and it can separate energy detection from zero-crossing detection. The results of these two kinds of detection are carried out simultaneously without affecting each other, so that multithreaded parallel computing is realized. In addition, the improved energy-zero crossing rate endpoint detection method uses a single threshold, compared with the traditional algorithm, the improved algorithm can reduce the threshold parameter by half, and make the algorithm more simple. For mobile terminals with limited space resources, the improved algorithm is compared with the conventional single threshold energy detection method. It is found that the recognition rate of the voiceprint recognition system using the improved algorithm is higher than that of the conventional single threshold energy detection method. Therefore, the improved energy-zero-crossing two-stage fusion endpoint detection method has high application value in mobile terminal. Aiming at the problem that the traditional voice frame voting method can not highlight the difference of the result of each frame, a weighted voting method based on likelihood probability is proposed in this paper. According to the likelihood probability of different speech frames and probabilistic models, each frame is weighted by this method, which makes the speech frames with large likelihood probability have greater weight and higher confidence, thus enhancing the difference between the results of speech judgment in each frame. The result of speech frame fusion is more accurate. At the same time, through multiple weighted detection, this paper verifies that the voice-pattern recognition system based on weighted voting method is better than that based on traditional voting method. Finally, this paper designs a variety of feature extraction techniques and probability model combination scheme, through the actual recognition effect and algorithm complexity to analyze their feasibility on the mobile terminal, select the most feasible scheme. According to the optimal scheme of voiceprint recognition system, a voiceprint recognition system based on mobile terminal is designed, and the system is implemented on MATLAB platform. The system can realize voice pattern acquisition, model training, voiceprint recognition and registration. Voiceprint confirmation and other functions. At present, the system has been successfully transplanted to the Android system.
【學(xué)位授予單位】：上海師范大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TN912.34

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 屈丹,王炳錫,魏鑫;基于GMM-UBM模型的語(yǔ)言辨識(shí)研究[J];信號(hào)處理;2003年01期

2 甄斌,吳璽宏,劉志敏,遲惠生;語(yǔ)音識(shí)別和說(shuō)話人識(shí)別中各倒譜分量的相對(duì)重要性[J];北京大學(xué)學(xué)報(bào)(自然科學(xué)版);2001年03期

3 胡光銳,韋曉東;基于倒譜特征的帶噪語(yǔ)音端點(diǎn)檢測(cè)[J];電子學(xué)報(bào);2000年10期

4 趙雪芬 ,江肇蓮;頻譜分析儀的諧波測(cè)量技術(shù)[J];國(guó)外電子測(cè)量技術(shù);2001年02期

5 陳芬菲;;基于GMM的說(shuō)話人識(shí)別系統(tǒng)[J];微處理機(jī);2006年04期

6 燕繼坤,鄭輝,王艷,曾立君;基于可信度的投票法[J];計(jì)算機(jī)學(xué)報(bào);2005年08期

7 王娜;鄭德忠;張淑清;;基于混沌振子的低信噪比語(yǔ)音端點(diǎn)檢測(cè)新方法[J];儀器儀表學(xué)報(bào);2009年07期

8 韓志艷;王旭;王健;;基于短時(shí)能零積和鑒別信息的語(yǔ)音端點(diǎn)檢測(cè)[J];東北大學(xué)學(xué)報(bào)(自然科學(xué)版);2009年12期

9 陳業(yè)仙;張歆奕;毛杰;;基于GMM-UBM的語(yǔ)言辨識(shí)算法研究[J];五邑大學(xué)學(xué)報(bào)(自然科學(xué)版);2010年03期

10 蔣曄;唐振民;;GMM文本無(wú)關(guān)的說(shuō)話人識(shí)別系統(tǒng)研究[J];計(jì)算機(jī)工程與應(yīng)用;2010年11期

相關(guān)博士學(xué)位論文 前1條

1 張晶;聲紋識(shí)別魯棒性技術(shù)及應(yīng)用研究[D];廣東工業(yè)大學(xué);2015年

相關(guān)碩士學(xué)位論文 前6條

1 李煒?shù)h;基于Android的有身份識(shí)別功能的流媒體播放器的設(shè)計(jì)與實(shí)現(xiàn)[D];電子科技大學(xué);2014年

2 路娜;孤立詞語(yǔ)音識(shí)別系統(tǒng)的研究與設(shè)計(jì)[D];曲阜師范大學(xué);2014年

3 陳衛(wèi)強(qiáng);基于DSP的孤立詞語(yǔ)音識(shí)別系統(tǒng)的研究與實(shí)現(xiàn)[D];南昌航空大學(xué);2013年

4 張慧珊;基于聲紋識(shí)別和動(dòng)態(tài)密碼的雙因素身份認(rèn)證系統(tǒng)的研究與實(shí)現(xiàn)[D];武漢理工大學(xué);2013年

5 胡政權(quán);說(shuō)話人識(shí)別中語(yǔ)音參數(shù)提取方法的研究[D];南京師范大學(xué);2013年

6 郝艷莉;基于DM6446的音頻信號(hào)識(shí)別系統(tǒng)的研究[D];哈爾濱理工大學(xué);2012年

，

本文編號(hào)：2173475