【摘要】：語言是人們?nèi)粘＿M(jìn)行溝通最常見的方法之一,是一種不可或缺的技能。在全球化進(jìn)程中,人們對(duì)語言溝通的障礙日益凸顯。在這種背景下,迫切要求能夠?qū)崿F(xiàn)語言識(shí)別,因此,語言識(shí)別成為近幾年語音研究方向的重要研究課題。現(xiàn)有的語言識(shí)別系統(tǒng)仍然存在很多問題,比如在復(fù)雜的語音背景下提取出純凈的語音信息,從易混淆的語言中將具有語言屬性的信息剝離出來等,因此,語言識(shí)別仍有待繼續(xù)研究和探索。語言識(shí)別(Language Identification,LID)是根據(jù)語音對(duì)說話人所說語言所屬種類進(jìn)行自動(dòng)區(qū)分,從而進(jìn)行說話人語種鑒別的生物特征識(shí)別技術(shù)�；谝羲靥卣骱突诘讓拥穆晫W(xué)特征已經(jīng)被證明能夠非常有效的代表語言種類信息。雖然,通過機(jī)器學(xué)習(xí)能夠有效改進(jìn)了語言識(shí)別性能,但識(shí)別率依然達(dá)不到要求,尤其是對(duì)于短時(shí)語音段來說,識(shí)別性能仍然有待提高。近年來,基于DNN(Deep Neural Network,DNN)的語言識(shí)別更是由于DNN的興起和廣泛應(yīng)用以及良好效果,成為學(xué)術(shù)界以及工業(yè)界的一個(gè)研究熱點(diǎn)。本次課題以基于DNN的語言識(shí)別為研究重點(diǎn),致力于完成一個(gè)完善的且性能良好的語言識(shí)別系統(tǒng)。主要做了以下幾點(diǎn)工作:1.實(shí)現(xiàn)基于DNN的語言識(shí)別系統(tǒng)。2.采用一種基于底層聲學(xué)特征的音素特征向量,即DBF(Deep Bottleneck Features,DBF)特征,這一特征比底層聲學(xué)特征和音素特征更能夠?qū)φZ言特征進(jìn)行表述。3.使用一種采用DBF訓(xùn)練DNN統(tǒng)計(jì)量提取I-Vector的方法,將DBF代替UBM(Universal Background Model,UBM)用在GMM(Gaussian Mixture Model,GMM)模型中,獲得更加精確的統(tǒng)計(jì)量,進(jìn)而提高識(shí)別效率。4.對(duì)整個(gè)系統(tǒng)進(jìn)行測(cè)試和分析。首先,對(duì)DBF特征與SDC特征進(jìn)行性能對(duì)比,結(jié)果表明,DBF特征對(duì)語言有更強(qiáng)的表達(dá)能力,在短時(shí)語音任務(wù)、長(zhǎng)時(shí)語音任務(wù)和易混淆和方言識(shí)別任務(wù)上性能有顯著提高。然后,對(duì)基于DBF-GMM-TV的方法和基于DNN-TV的方法在性能方面做了對(duì)比分析。表明采用這種模型域能更有效的對(duì)模型進(jìn)行估計(jì)。最后,從本地測(cè)試和網(wǎng)絡(luò)在線測(cè)試兩個(gè)方面對(duì)系統(tǒng)性能做了測(cè)試。
[Abstract]:Language is one of the most common ways for people to communicate on a daily basis, and it is an indispensable skill. In the process of globalization, the obstacles to language communication are becoming more and more prominent. In this context, it is urgent to realize language recognition. Therefore, language recognition has become an important research topic in speech research in recent years. There are still many problems in the existing language recognition systems, such as extracting pure speech information from complex speech background, stripping the information with language attributes from confusing languages, and so on. Language recognition still needs to be further studied and explored. Language recognition (Language Identification,LID) is a biometric recognition technology which automatically distinguishes the language to which the speaker belongs according to speech, so as to identify the speaker's language. Phoneme based features and underlying acoustic features have been proved to be very effective in representing language category information. Although the performance of language recognition can be effectively improved by machine learning, the recognition rate still does not meet the requirements, especially for short-term speech segments, the recognition performance still needs to be improved. In recent years, language recognition based on DNN (Deep Neural Network,DNN has become a research focus in academia and industry because of the rise, wide application and good results of DNN. In this paper, DNN-based language recognition is the focus of research, and a perfect and good language recognition system is devoted to the completion of a perfect and good performance language recognition system. The main work has been done as follows: 1. Implement a language recognition system based on DNN. 2. A phoneme feature vector (DBF (Deep Bottleneck Features,DBF) feature based on the underlying acoustic feature is used, which is more able to express the language feature than the underlying acoustic feature and phoneme feature. Using a method of using DBF training DNN statistics to extract I-Vector, DBF is used instead of UBM (Universal Background Model,UBM in GMM (Gaussian Mixture Model,GMM) model to obtain more accurate statistics, and then improve the recognition efficiency. 4. The whole system is tested and analyzed. Firstly, the performance of DBF features is compared with that of SDC features. The results show that DBF features have stronger expression ability to language, and the performance of DBF features is significantly improved in short-term speech tasks, long-term speech tasks and obfuscating and dialect recognition tasks. Then, the performance of DBF-GMM-TV-based method and DNN-TV-based method is compared and analyzed. It is shown that the model domain can be used to estimate the model more effectively. Finally, the system performance is tested from two aspects: local test and network online test.
【學(xué)位授予單位】：電子科技大學(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9條

1 張翼飛;騰瀟琦;;一種改進(jìn)的有監(jiān)督訓(xùn)練的TV語種識(shí)別方法[J];計(jì)算機(jī)應(yīng)用與軟件;2016年09期

2 馬志強(qiáng);張澤廣;閆瑞;劉利民;馮永祥;蘇依拉;;基于N-Gram模型的蒙古語文本語種識(shí)別算法的研究[J];中文信息學(xué)報(bào);2016年01期

3 崔瑞蓮;宋彥;蔣兵;戴禮榮;;基于深度神經(jīng)網(wǎng)絡(luò)的語種識(shí)別[J];模式識(shí)別與人工智能;2015年12期

4 洪新海;宋彥;蔣兵;戴禮榮;;采用DBN的TV改進(jìn)方法在語種識(shí)別中的應(yīng)用[J];信號(hào)處理;2015年09期

5 楊緒魁;屈丹;張文林;;基于正則化i-Vector算法的語種識(shí)別[J];信息工程大學(xué)學(xué)報(bào);2015年02期

6 陳瑤玲;楊鑒;陳江;徐永華;;基于支持向量機(jī)的語種識(shí)別[J];信息技術(shù);2010年06期

7 劉明輝;戴蓓劏;解焱陸;;基于GMM多維概率輸出的SVM話者確認(rèn)[J];模式識(shí)別與人工智能;2008年01期

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

9 劉穎超,張紀(jì)元;梯度下降法[J];南京理工大學(xué)學(xué)報(bào)(自然科學(xué)版);1993年02期

相關(guān)會(huì)議論文 前1條

1 田W，

本文編號(hào)：2480760