隨著深度學習技術的崛起，在聲音辨識的領域上越來越多的應用逐漸被開發。許多在醫學上的研究可以利用深度學習技術的優勢，來達到更好的效果。因此本篇論文以呼吸音結合深度學習技術進行研究。
本篇論文的目的是開發一套聲音辨識系統，提供使用者自行上傳及辨識呼吸音。本文系統是以響應式網頁來設計，讓使用者可以在任何裝置上使用，透過使用者友善的界面，使用者可以更容易的進行操作。其中結合Google API的OAuth2.0管理帳戶登入，並利用Python的Selenium套件，來取得無法串接第三方聽診器應用的數據。
實驗數據使用ICBHI Challenge 2017科學比賽所提供的呼吸音開放數據，比較三種較熱門的CNN深度學習模型，即InceptionResNetV2、VGG16及MobileNetV2。在保持原有訓練及測試資料配比情況下，經過適度調整與搭配不同的特徵值擷取方法後，這三種深度學習模型的辨識率皆有所提升。

With the rise of deep learning technology, more applications in the field of sound recognition are gradually being developed. Many researches in the medical field take advantages of the deep learning technology to achieve better results. Therefore, this paper uses respiratory sounds combined with deep learning technology for research.
The purpose of this paper is to develop a cloud-based sound recognition system to provide users with the function of uploading and recognizing respiratory sounds. This system is designed with the technique of responsive web design, which make users accessible on any devices. The user-friendly interface design makes it easier for users to operate. This system combines the OAuth2.0 through Google APIs to manage the account login on this system, and uses the Selenium suite via python programs to obtain respiratory sounds that cannot be connected to third-party stethoscope applications.
In performance evaluations, we use the open data of respiratory sounds provided by the ICBHI Challenge 2017 science competition, and compare three popular CNN deep learning models, namely InceptionResNetV2, VGG16 and MobileNetV2. While maintaining the original training and test data ratio, these three existing deep learning models can be improved with some adjustments and suitable feature extraction methods.

誌謝 ii
摘要 iii
Abstract iv
目錄 v
圖目錄 viii
表目錄 x
第1章 前言 1
1-1 研究背景與動機 1
1-2 研究目的 1
1-3 論文綱要 2
第2章 相關研究 3
2-1 呼吸音介紹 3
2-1-1 呼吸音種類 3
2-1-2 呼吸音聽診 4
2-1-3 聽診器介紹 4
2-1-4 ICBHI Challenge 2017資料集 5
2-2 聲音辨識方法 6
2-2-1 前置處理 7
2-2-2 特徵提取 7
2-2-3 機器學習 8
2-2-4 深度學習 10
2-3 應用的技術 12
2-3-1 TensorFlow 12
2-3-2 Selenium Webdriver 12
2-3-3 RWD 響應式網頁設計 12
2-3-4 OAuth 2.0 13
第3章 雲端聲音辨識系統 15
3-1 系統架構 15
3-1-1 使用者端 15
3-1-2 伺服器端 16
3-1-3 系統運作流程 19
3-2 聲音訓練流程 20
3-2-1 原始音檔 20
3-2-2 聲音切割 22
3-2-3 縮小取樣 24
3-2-4 濾波處理 25
3-2-5 資料擴增 27
3-2-6 特徵提取 28
3-2-7 訓練模型 29
3-3 聲音辨識流程 31
3-4 使用者介面 32
第4章 實驗方法與結果 35
4-1 實驗環境 35
4-2 評估方法 35
4-3 實驗方法討論 36
4-3-1 未濾波及未資料擴增處理 37
4-3-2 經過濾波處理 39
4-3-3 經過濾波及資料擴增處理 41
4-3-4 實驗討論 43
第5章 結論與未來工作 47
5-1 結論 47
5-2 未來工作 47
第6章 參考文獻 48

[1] 黃俊卿, 基於物聯網之環境聲音辨識偵測平臺, 碩士論文，東華大學，2017.
[2] S. Avutu, D. Bhatia and B. Reddy, "Voice Control Module for Low Cost Local-Map Navigation Based Intelligent Wheelchair," 2017 IEEE 7th International Advance Computing Conference (IACC), pp. 609-613, 2017.
[3] T. Erić, S. Ivanović, S. Milivojša, M. Matić and N. Smiljković, "Voice control for smart home automation: Evaluation of approaches and possible architectures," 2017 IEEE 7th International Conference on Consumer Electronics - Berlin (ICCE-Berlin), pp. 140-142, 2017.
[4] Z. Yan and S. Zhao, "A Usable Authentication System Based on Personal Voice Challenge," 2016 International Conference on Advanced Cloud and Big Data (CBD), pp. 194-199, 2016.
[5] Demir, F., Sengur, A. & Bajaj, V., "Convolutional neural networks based efficient approach for classification of lung diseases," Health Information Science and Systems volume 8, 2020.
[6] M. e. a. Grønnesby, Feature Extraction for Machine Learning Based Crackle Detection in Lung Sounds from a Health Survey, 2017.
[7] K. L. Kuan, A Framework for Automated Heart and Lung Sound Analysis, 2010.
[8] "聲音 | 維基百科，自由的百科全書," [Online]. Available: https://zh.wikipedia.org/wiki/声音.
[9] 吳雅婷, 文薜帷, 黃一城, 黃煜庭, 劉晟昊, "呼吸系統的身體檢查," 家庭醫學與基層醫療, vol. 30, no. 3, pp. 67-83, 2015.
[10] S. Lehrer, Understanding Lung Sound Third Edition.
[11] 麻生七緒, “つれづれ呼吸ケア日記,” 看護師の学び・仕事に役立つサイト, 26 12 2015. [線上]. Available: https://nursepress.jp/1208.
[12] T. Ferns, "Respiratory auscultation: how to use a stethoscope," in Nursing Times, vol. 103, 2007, pp. 28-29.
[13] R. T. H. Laennec, A treatise on the diseases of the chest and on mediate auscultation, 1835.
[14] S. Swarup and A. N. Makaryus, "Digital stethoscope: technology update," Medical Devices: Evidence and Research, vol. 11, pp. 29-36, 4 1 2018.
[15] "聽診器 | 維基百科，自由的百科全書," [Online]. Available: https://zh.wikipedia.org/wiki/聽診器.
[16] "Eko," [Online]. Available: https://www.ekohealth.com/.
[17] "Eko AI Press Release," 28 01 2020. [Online]. Available: https://assets.website-files.com/5d43b941a4923b9c4685f98d/5e2fc70cca409a6dadabe2de_Eko_AI_Press%20Release.pdf.
[18] D. Chamberlain, J. Mofor, R. Fletcher and R. Kodgule, "Mobile stethoscope and signal processing algorithms for pulmonary screening and diagnostics," 2015 IEEE Global Humanitarian Technology Conference (GHTC), pp. 385-392, Seattle, WA, 2015.
[19] Rocha, Bruno & Filos, D. & Mendes, Luís & Serbes, Gorkem & Ulukaya, Sezer & Kahya, Yasemin & Jakovljevic, Niksa & Loncar-Turukalo, Tatjana & Vogiatzis, Ioannis & Perantoni, Eleni & Kaimakamis, Evangelos & Natsiavas, Pantelis & Oliveira, Ana & Jácome, Cris, "An open access database for the evaluation of respiratory sound classification algorithms," Physiological Measurement, vol. 40, no. 3, 1 2 2019.
[20] J. Robert, "pydub api documentation," [Online]. Available: https://github.com/jiaaro/pydub/blob/master/API.markdown.
[21] Refaeilzadeh P., Tang L., Liu H., "Cross-Validation," in Encyclopedia of Database Systems, LIU L., ÖZSU M.T. Springer, Boston, MA, 2009.
[22] S. F. Pedro, "Crackle and wheeze detection in lung sound signals using convolutional neural networks," University do Porto, Portugal, 2019.
[23] J. Acharya and A. Basu, "Deep Neural Network for Respiratory Sound Classification in Wearable Devices Enabled by Patient Specific Model Tuning," in IEEE Transactions on Biomedical Circuits and Systems, vol. 14, June 2020, pp. 535-544.
[24] F. Demir, A. Sengur and V. Bajaj, "Convolutional neural networks based efficient approach for classification of lung diseases," Health Information Science and Systems, 23 12 2019.
[25] Kochetov K., Putin E., Balashov M., Filchenkov A., Shalyto A., "Noise Masking Recurrent Neural Network for Respiratory Sound Classification," in Artificial Neural Networks and Machine Learning – ICANN 2018. ICANN 2018. Lecture Notes in Computer Science, 2018.
[26] G. Serbes, S. Ulukaya and Y. P. Kahya, "An Automated Lung Sound Preprocessing and Classification System Based OnSpectral Analysis Methods," in Precision Medicine Powered by pHealth and Connected Health. ICBHI 2017. IFMBE Proceedings, vol. 66, Springer, Singapore, 2018, pp. 45-49.
[27] N. Jakovljevic and T. Loncar-Turukalo, "Hidden Markov Model Based Respiratory Sound Classification," in Precision Medicine Powered by pHealth and Connected Health. ICBHI 2017. IFMBE Proceedings, vol. 66, Springer, Singapore, 2018, pp. 39-43.
[28] G. Chambres, P. Hanna and M. Desainte-Catherine, "Automatic Detection of Patient with Respiratory Diseases Using Lung Sound Analysis," in 2018 International Conference on Content-Based Multimedia Indexing (CBMI), La Rochelle, 2018.
[29] S. Butterworth, "On the Theory of Filter Amplifiers," in Wireless Engineer.
[30] "巴特沃斯濾波器 | 維基百科，自由的百科全書," [Online]. Available: https://zh.wikipedia.org/wiki/巴特沃斯濾波器.
[31] "Albumentations | Github," [Online]. Available: https://github.com/albumentations-team/albumentations.
[32] "Audiomentations | Github," [Online]. Available: https://github.com/iver56/audiomentations.
[33] Y. Ma et al., "LungBRN: A Smart Digital Stethoscope for Detecting Respiratory Disease Using bi-ResNet Deep Learning Algorithm," 2019 IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 1-4, 2019.
[34] S. S. Stevens, J. Volkmann and E. B. Newman, "A Scale for the Measurement of the Psychological Magnitude Pitch," Journal of the Acoustical Society of America, pp. 185-190, 1937.
[35] S. Davis and P. Mermelstein, "Comparison of Parametric Representations for Monosyllabic Word Recognition in Continuously Spoken Sentences," IEEE Transactions on Acoustics, Speech, and Signal Processing, pp. 357-366, 1980.
[36] "Keras Applications," [Online]. Available: https://keras.io/api/applications/.
[37] "TensorFlow," [Online]. Available: https://www.tensorflow.org/.
[38] "TensorFlow Lite," [Online]. Available: https://www.tensorflow.org/lite.
[39] "SeleniumHQ Browser Automation," [Online]. Available: https://www.selenium.dev/.
[40] "回應式網頁設計 | 維基百科，自由的百科全書," [Online]. Available: https://zh.wikipedia.org/wiki/响应式网页设计.
[41] J. Clark, "Content like water," 8 11 2011. [Online]. Available: https://twitter.com/bigmediumjosh/status/133587842654937088.
[42] B. Lee, "Be water my friend".
[43] "2018台灣網路調查-整體網路使用情況," [Online]. Available: https://report.twnic.tw/2018/OverallNetworkUsageStatus.html.
[44] "RFC 6749 - The OAuth 2.0 Authorization Framework," [Online]. Available: https://tools.ietf.org/html/rfc6749.
[45] "Using OAuth 2.0 to Access Google APIs | Google Identity Platform," [Online]. Available: https://developers.google.com/identity/protocols/oauth2.
[46] "Using OAuth 2.0 for Web Server Applications | Google Identity Platform," [Online]. Available: https://developers.google.com/identity/protocols/oauth2/web-server#php.