現在位置首頁 > 博碩士論文 > 詳目
論文中文名稱:應用語者辨識技術與網路社群平台增進輕微認知障礙銀髮族的人際關係 [以論文名稱查詢館藏系統]
論文英文名稱:Improving the Interpersonal Relationship of the Elderly with Mild Cognitive Impairment by Using Speaker Recognition and Social Network Platforms [以論文名稱查詢館藏系統]
院校名稱:臺北科技大學
學院名稱:電資學院
系所名稱:電機工程系
畢業學年度:106
畢業學期:第二學期
出版年度:107
中文姓名:林佳蓉
英文姓名:Chia-Jung Lin
研究生學號:105318045
學位類別:碩士
語文別:中文
口試日期:2018/03/20
論文頁數:61
指導教授中文名:譚旦旭;簡福榮
口試委員中文名:譚旦旭;簡福榮;段裘慶;劉宗瑜
中文關鍵詞:輕度知能障礙語者辨識高斯混合模型泛用背景模型
英文關鍵詞:Mild Cognitive ImpairmentSpeaker RecognitionGaussian Mixture ModelUniversal Background Model
論文中文摘要:輕度知能障礙(Mild Cognitive Impairment, MCI)患者最顯著的特徵是記憶力大幅衰退,因此會嚴重影響人際關係。本論文旨在應用語者辨識技術(Speaker Recognition)與網路社群平台的關聯功能開發一套照護系統來改進MCI銀髮族的記憶力,因而提升他們的人際關係,同時減緩失智的速度。首先,本研究將實現一套基於高斯混合模型(Gaussian Mixture Model, GMM) 及泛用背景模型(Universal Background Model, UBM)的語者辨識子系統,此子系統可以藉由親友/訪客的聲音辨識他們的身份,因此可以立即顯示親友/訪客的背景資料,此照護系統接著可連結到自建資料庫及社群平台擷取與雙方關聯的歷史資料,例如:一起聚會及出遊的照片,而透過關聯資料的懷舊療法,MCI 銀髮族可以立即回到雙方共同擁有的美好回憶,除熱絡氣氛,也可以刺激MCI銀髮族的記憶能力,因而延緩知能退化的幅度。
本論文採用中華電信研究所「單一國語數字」語音資料庫及另一套由本系416研究室團隊自行錄製的語料庫測試語者辨識之效能,實驗結果顯示,基於GMM-UBM模型的語者辨識子系統有最好的辨識效能,另外,使用環境匹配的強健式模型可降低雜訊對系統效能的影響。最後,我們邀請五位銀髮族實際操作我們所開發的系統,並以問卷調查了解銀髮族的使用情況。問卷調查的統計結果顯示銀髮族高度肯定此一系統,所獲之系統使用性(System Usability)達到前10%的分數,因此深具實用價值。
論文英文摘要:Significant memory decline is the most obvious symptom of the elderly persons with mild cognitive impairment (MCI) and this symptom is disadvantageous to their interpersonal relationship. This study will develop a care system to improve the memory of the elderly with MCI by employing the speaker recognition technique and association functionality of social network platforms. The proposed system aims to not only maintain a good interpersonal relationship, but also slow down the cognitive decline of the elderly with MCI. We will develop a speaker recognition subsystem based on the Gaussian Mixture Model-Universal Background Model (GMM-UBM) to identity the visitor/family via individual input utterance, and thus the personal background of the visitor/family can be immediately shown on the screen of the elderly with MCI. Furthermore, the proposed system can be linked to the private database and social network platforms to extract their associated information, such as the photos of parties and tours. Moreover, via the effect of the reminiscence therapy using association data, they can easily go back to the good old days and share beautiful things. Besides warming up their interaction in a very short time, this method stimulates the memory of the elderly with MCI, hence alleviating the tendency of cognitive decline.
This work employs two databases to evaluate the speaker recognition performance. Experimental results indicate that the speaker recognition subsystem based on GMM-UBM obtains the best performance. In addition, a robust model using an environmental match scheme can effectively alleviate the effect of background noise. Finally, five elderly persons are invited to measure the usability of the proposed system. A questionnaire is used to survey the five elderly persons, and it exhibits a very positive feedback of system usability, which reaches top 10%. Therefore, the proposed system is highly potentially applicable in improving the memory of the elderly with MCI.
論文目次:摘 要 i
ABSTRACT iii
誌 謝 v
目 錄 vi
表目錄 ix
圖目錄 x
第一章 緒論 1
1.1 研究背景與動機 1
1.2 文獻探討 2
1.3 研究目的與方法 4
1.4 論文架構 5
第二章 背景知識 6
2.1 失智症(Dementia) 6
2.1.1 失智症的種類 7
2.1.2 初期失智症的症狀 8
2.1.3 失智症的病程 9
2.1.4 輕微認知障礙(Mild Cognitive Impairment, MCI) 9
2.2 懷舊療法(Reminiscence Therapy) 11
2.2.1 懷舊療法的功能 12
2.2.2 懷舊療法的目標 12
2.3 網路社群平台 13
2.3.1 社群網站 13
2.4 語者辨識技術(Speaker Recognition) 17
2.4.1 語者模型(Speaker Model) 17
第三章 研究方法 19
3.1 系統架構 19
3.2 語者辨識系統 20
3.3 語者區別(Speaker Identification) 21
3.4 梅爾頻率倒頻譜係數(Mel-Frequency Cepstrum Coefficients, MFCC) 23
3.5 基於GMM模型之語者辨識系統 29
3.5.1 高斯混合模型(Gaussian Mixture Model, GMM) 29
3.5.2 K平均值演算法(K-Means Algorithm) 31
3.5.3 期望值最大化演算法(Expectation Maximization Algorithm, EM) 32
3.6 基於GMM-UBM模型之語者辨識系統 33
3.6.1 泛用背景模型(Universal Background Model, UBM) 33
3.6.2 GMM-UBM模型 34
3.6.3 最大事後機率(Maximum A Posterior, MAP) 35
3.7 OAuth認證平台 38
3.8 Facebook Graph API 40
第四章 實驗結果 42
4.1 語者辨識系統效能實驗 42
4.1.1 實驗設定 42
4.1.2 語音資料庫 42
4.1.3 特徵參數 43
4.1.4 模型訓練 44
4.1.5 實驗結果 45
4.1.6 分析與討論 48
4.2 結合語者辨識與社群網站之照護系統實驗 49
4.2.1 系統操作流程 50
4.2.2 實驗介紹 53
4.2.3 實驗結果與分析 53
第五章 結論與未來發展 55
5.1 結論 55
5.2 未來發展 56
參考文獻 57
附錄 一 滿意度調查及回饋問卷 60
論文參考文獻:[1] 內政部統計處,來源: http://www.moi.gov.tw/stat/index.aspx
[2] 本部戶政司,國家發展委員會中華民國人口推估,來源: https://www.ndc.gov.tw/Content_List.aspx?n=5B78EEBCE18CBE9F
[3] 陳達夫,邱銘章,葉炳強(台大醫院神經部主治醫師),輕度知能障礙的研究現況,臺大校友雙月刊,第46期,民國95年7月。
[4] A. Busse, et al., “Subclassifications for mild cognitive impairment: prevalence and predictive validity,” Psychol Med. pp. 1029-38, 2003.
[5] A. Busse, et al., “Mild cognitive impairment: prevalence and predictive validity according to current approaches,” Acta Neurol Scand, pp. 71-81, 2003.
[6] C. DeCarli, “Mild cognitive impairment: prevalence, prognosis, aetiology, and treatment,” Lancet Neurol. pp. 15-21, 2003.
[7] D. S. Knopman, B. F. Boeve, and R. C. Petersen, “Essentials of the proper diagnoses of mild cognitive impairment, dementia, and major subtypes of dementia,” Mayo Clin Proc. pp. 1290-308, 2003.
[8] R. C. Petersen, et al., “Current concepts in mild cognitive impairment,” Arch Neurol. pp. 1985-1992, 2001.
[9] K. Ritchie, S. Artero, and J. Touchon, “Classification criteria for mild cognitive impairment: a population-based validation study,” Neurology, pp. 37-42, 2001.
[10] L. J. Thal, “Therapeutics and mild cognitive impairment: current status and future directions,” Alzheimer Dis Assoc Disord., pp. S69-S71, 2003.
[11] 臺灣失智症協會,105~150年臺灣地區失智症人口推計報告,來源: http://www.tada2002.org.tw/tada_know_02.html
[12] 國際失智症協會,2015年全球失智症報告,來源: https://www.alz.co.uk/research/worldalzheimerreport2015-traditionalchinese.pdf
[13] J. Vanus, J. Koziorek, and R. Hercik, “The design of the voice communication in Smart Home Care,” the 36th International Conference on Telecommunications and Signal Processing, 2013.
[14] H. Christensen, et al., “homeService: Voice-enabled assistive technology in the home using cloud-based automatic speech recognition,” the 4th Workshop on Speech and Language Processing for Assistive Technologies, 2013.
[15] M. Vacher, et al., “Evaluation of a context-aware voice interface for Ambient Assisted Living: qualitative user study vs. quantitative system evaluation,” ACM Trans. Accessible Computing, pp. 1-36, 2015.
[16] F. Portet, et al., “Design and evaluation of a smart home voice interface for the elderly: Acceptability and objection aspects,” Personal and Ubiquitous Computing, vol. 17, no. 1, pp. 127-144, 2013.
[17] M. Vacher, et al., “The SWEET-HOME project: audio technology in smart homes to improve well-being and reliance,” the 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 5291-5294, 2013.
[18] Y.-Y. Ou, et al., “Framework of ubiquitous healthcare system based on cloud computing for elderly living,” Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, pp. 1-4, 2013.
[19] Y. Zhao, “Speech-recognition technology in health care and special-needs assistance,” IEEE Signal Processing Magazine, pp. 87-90, 2009.
[20] M. Vache, et al., “Speech and speaker recognition for home automation: Preliminary results,” International Conference on Speech Technology and Human-Computer Dialogue, pp. 181-190, 2015.
[21] H. Li and B. Ma, “TechWare: Speaker and spoken language recognition resources,” IEEE Signal Processing Magazine, pp. 139-142, 2010.
[22] R. Togneri, and D. Pullella, “An overview of speaker identification: accuracy and robustness issues,” IEEE Circuits and Systems Magazine, pp. 23-61, 2011.
[23] 張世維,多重環境下語者辨識之效能評估與改進,國立台北科技大學電機系碩士論文,July 2003.
[24] D. A. Reynolds, T. F. Quatieri, and R. B. Dunn, “Speaker verification using adapted Gaussian mixture models,” Digital Signal Processing, pp. 19-41, 2000.
[25] S. Young, et al., “Hidden Markov model toolkit (HTK) version 3.4 user’s guide,” 2002.
[26] H. AlMazrua, et al., “Technology for psychosocial interventions for individuals with Alzheimers: reminiscence: therapy and monitoring progressive decline of cognitive abilities,” Science and Information Conference, pp. 171-175, 2013.
[27] B. Woods, et al., “Reminiscence therapy for dementia,” Cochrane Database Systematic Reviews, 2, 2005.
[28] E. Chapoulie, et al., “Reminiscence therapy using image-based rendering in VR,” IEEE Virtual Reality, pp. 45-50, 2014.
[29] 國家網路醫藥,來源: http://hospital.kingnet.com.tw/essay/essay.html?pid=28505
[30] 邱銘章、湯麗玉,失智症照護指南,原水文化,2009。
[31] R. C. Petersen, et al., “Mild cognitive impairment: clinical characterization and outcome,” Archives of Neurology, 56, pp. 303-308, 1999.
[32] R. C. Petersen, et al., “Practice parameter: early detection of dementia: mild cognitive impairment (an evidence-based review),” Report of the Quality Standards Subcommittee of the American Academy of Neurology, pp. 1133-1142, 2001.
[33] M. L. Lee and A. K. Dey, “Providing good memory cues for people with episodic memory impairment,” In Proceedings Assets ’07, ACM, New York, USA, pp. 131-138, 2007.
[34] 陳淑華,張曉雲,謝秀芳,懷舊療法之介紹,馬偕護理雜誌,6卷1期,pp. 7-16,2012。
[35] 維基百科,Facebook,來源:https://zh.wikipedia.org/wiki/Facebook
[36] 維基百科,Google+,來源:https://zh.wikipedia.org/wiki/Google+
[37] 維基百科,騰訊QQ,來源:https://zh.wikipedia.org/wiki/騰訊
[38] 維基百科,新浪微博,來源:https://zh.wikipedia.org/wiki/新浪微博
[39] 維基百科,Twitter,來源:https://zh.wikipedia.org/wiki/Twitter
[40] 維基百科,Instagram,來源:https://zh.wikipedia.org/wiki/Instagram
[41] S. Furui, “An overview of speaker recognition technology”, ESCA Workshop on Automatic Speaker Recognition, Identification and Verification, pp. 1-9, 1994.
[42] D. Martinez, et al., “Unscented transform for iVector-based noisy speaker recognition,” IEEE International Conference on Acoustic, Speech and Signal Processing, pp. 4042-4046, 2014.
[43] H. Ning, Y. X. Zou, and X. Hu, “A new score normalization for text-independent speaker verification,” the 19th International Conference on Digital Signal Processing, pp. 636-639, 2014.
[44] A. Jain, A. Ross, and S. Prabhakar, “An introduction to biometric recognition,” IEEE Trans. Circuits Systems Video Technol., pp. 4-20, 2004.
[45] 維基百科,梅爾刻度,來源:https://zh.wikipedia.org/wiki/梅爾刻度
[46] 王小川,語音訊號處理,全華圖書,第三版,2012。
[47] A. Dempster, N. Laird, D. Rubin, “Maximum likelihood from incomplete data via the EM algorithm,” Journal of the Royal Statistical Society 39, pp. 1-38, 1977.
[48] U. Bhattacharjee, and K. Sarmah, “GMM-UBM based speaker verification in multilingual environments,” IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 6, No 2, pp. 373-380, November 2012.
[49] L. Heck and N. Mirghafori, “On-Line unsupervised adaptation in speaker verification,” In Proc. ICSLP-2000, vol. 2, pp. 454-457, 2000.
[50] A. Bangor, “Determining what individual SUS scores mean: adding an adjective rating scale,” Journal of Usability Studies, pp. 114-123, 2009.
[51] J. Sauro, “Measuring Usability with the System Usability Scale (SUS),” 2011, February 2, Retrieved from https://measuringu.com/sus/
論文全文使用權限:不同意授權