近年來物聯網(IoT)裝置已經漸漸普及至我們的生活當中，在如此多的IoT裝置當中智慧門鎖也占了相當大了一塊，但現今之智慧門鎖無論是感應式或是密碼式又或生物特徵辨識都需要直接與智慧門鎖接觸，對於手上有拿許多東西的使用者又或是行動不便的使用者可說是相當的不便。另外，如需要更好的辨識能力往往需要更多的器具與技術去搭配，但如果考量到成本與普及化之問題，現今的智慧型門鎖往往不適合此狀況。
本論文將基於上述提到的問題，進而提出一個方便、低成本、高安全性的智慧型語音者辨識於門鎖開關之機制，來解決上述所提及之問題。為建立上述系統之雛形，首先必須分別建立3個不同的子系統，分別為: 1.Distance Detection、2. Speech Recognition、3. Speaker Recognition，透過上述之3個子系統交叉驗證將大幅提高智慧門鎖之安全性與便利性，最後透過Robot Operating System(ROS)對於子系統進行溝通與管理，提升子系統與子系統間之平行化，並實現門鎖之控制來完成整個系統。
為了實現上述方法，本論文結合高效率之嵌入式開發版Raspberry Pi 3 Model B，進行距離之偵測並進行聲音之錄製，再透過深度學習進行字詞與語者之辨識。在系統實驗部分實驗結果能完全辨識正常情形之字詞與語者，在深度學習之訓練結果上本系統之準確度為99%。

In recent years, the smart door lock is more and more popular, but now the wisdom of the lock. Today's smart locks, whether inductive or cryptographic or biometrics, need to be in direct contact with the Smart Door Lock, which is comparable to a user who has a lot of things in his hand or who is inconvenient. In addition, if you need better identification skills often need more equipment and technology to match, but if you consider the cost and popularity of the problem, today's smart door locks are often not suitable for this situation.
This paper will solve the above mentioned problems, and then propose a convenient, low-cost, high-security intelligent voice recognition mechanism in the door lock switch. In order to establish the prototype of the above system, we must first establish three different subsystems, namely: 1.Distance Detection, 2. Speech Recognition, 3. Speaker recognition, through the above three subsystems cross validation will greatly improve the smart door locks and finally through the Robot Operating System (ROS) for the subsystems to communicate and manage, to enhance the parallel between the subsystem and subsystems, and to achieve the control of the door to complete the entire system.
In order to achieve the above method, this paper combines the high-efficiency embedded development version Raspberry Pi 3 Model B, the distance of the detection and voice recording, and then through the depth of learning words and language recognition. In the experimental part of the system experiment results can fully identify the words and words of the normal situation, in the depth of the training results on the system accuracy of 99%.

[1]BENEVO,(2017 June),"撬開智慧家居的第一道門鎖，真有智慧嗎?!",[Online],Available:https://goo.gl/NsnUYt.
[2]Mike Cronin,(2017 June), "Kwikset Kevo Smart Lock -Unboxing, Install & Review!",[Youtube],Available:https://www.youtube.com/watch?v=ag1MkSB22h4
[3]MUL-T-LOCK Website,(2017 June) , "ENTR®- Smart Lock Solution",[Online],
Available: http://www.mul-t-lock.com/en/site/mul-t-lock/for-home/entr/
[4]智慧產品圈,(2017 June), "紅透半邊天的智能門鎖，現在市場狀況究竟如何？｜智慧產品圈",[Online],Available: https://kknews.cc/zh-tw/tech/v9ymg4a.html
[5]KICKSTARTER,(2017 June), "Sesame. Your key, reinvented.",[Online],Available: https://www.kickstarter.com/projects/candyhouse/sesame-your-key-reinvented
[6]Alice Chang,(2017 June), "August 智慧鎖：讓你出門不用帶鑰匙",[Online],Available: https://www.bnext.com.tw/px/article/38087/BN-2015-11-30-135905-77
[7]戴客網,(2017 June), "再也不怕忘帶鑰匙了 10款必看的智能門鎖",[Online],Available: https://read01.com/7RjGBR.html
[8]Y. Bengio, "Learning deep architectures for AI",Foundations and trends® in Machine Learning vol. 2,p.1-127,2009
[9]G.E.Hinton,S.Osindero,and Y.-W.Teh, "A fast learning algorithm for deep belief nets",Neural computation,vol. 18,p1527-1554,2006
[10]維基百科,(2017 June), "AlphaGo", [Online] , Available: https://zh.wikipedia.org/wiki/AlphaGo
[11]資安人,(2017 July), "生物辨識技術應用新思維",[Online],Available: https://www.informationsecurity.com.tw/article/article_detail.aspx?aid=8342
[12]Variani, E., Lei, X., McDermott, E., Moreno, I. and Gonzalez-Dominguez, J. (2014). Deep neural networks for small footprint text-dependent speaker verification. 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
[13]Google,(2017 July), "Deep Learning", [Online], Available:https://trends.google.com/trends/explore?q=deep%20learning
[14]e絡盟指定台灣地區個人用戶經銷商,(2017 June), "[產品] Raspberry Pi 3 Model B (現貨)",[Online],Available: https://www.raspberrypi.com.tw/10684/55/
[15]"ROS Concepts", http://wiki.ros.org, 2016. [Online]. Available:https:// http://wiki.ros.org/ROS/Concepts. [Accessed: 15- Apr- 2017].
[16]Torselllo,(2017 June), "Google can opensource: TensorFlow" ,[Online],Available: http://nlpx.net/archives/245
[17]Bappy, J., Barr, J., Srinivasan, N. and Roy-Chowdhury, A. (2017). Real Estate Image Classification. 2017 IEEE Winter Conference on Applications of Computer Vision (WACV).
[18]ELEC Freaks,(2017 June), " Ultrasonic Ranging Module HC - SR04",[pdf],Available: http://www.micropik.com/PDF/HCSR04.pdf
[19]Raspberry Pi Learning Resources,(2017 June), "Ultrasonic distance sensor",[Online],Available: https://www.raspberrypi.org/learning/physical-computing-with-python/distance/
[20]OwnTracks Booklet,(2017 June), "What OwnTracks does",[Online],Available: http://owntracks.org/booklet/guide/whathow/
[21]Python Tutorials,(2017 June), "Speech Recognition using Google Speech API",[Online],Available: https://pythonspot.com/en/speech-recognition-using-google-speech-api/
[22]OpenSLR,(2017 June), "TED-LIUM",[Online],Available:http://www.openslr.org/7/
[23]zouxy09,(2017 June), "语音信号处理之（四）梅尔频率倒谱系数（MFCC）",[Online],Available: http://blog.csdn.net/zouxy09/article/details/9156785
[24]TFLearn.org,(2017 June), "TFLearn: Deep learning library featuring a higher-level API for TensorFlow.",[Online],Available: http://tflearn.org/
[25]TFLearn.org,(2017 June), " Image PreLoader.",[Online],Available: http://tflearn.org/data_utils/#image-preloader
[26]Wikipedia,(2017 June), "Numeric_representations",[Wikipedia],Available: https://en.wikipedia.org/wiki/RGB_color_model#Numeric_representations
[27]Wikipedia,(2017 June), "Grayscale",[Wikipedia],Available: https://en.wikipedia.org/wiki/Grayscale
[28]CSDN,(2017 June), "理解長短期記憶網絡（LSTM NetWorks）",[Online],Available: https://read01.com/JoyNG.html
[29]TensorFlow.org,(2017 June), "LSTM",[Online],Available: https://www.tensorflow.org/tutorials/recurrent#lstm
[30]Geoffrey E. Hinton, Nitish Srivastava, Alex Krizhevsky, Ilya Sutskever, Ruslan R. Salakhutdinov , "Improving neural networks by preventing co-adaptation of feature detectors",[CV],arXiv:1207.0580
[31]FEIXIANG,(2017 June), "Tips for Deep Learning",[Online],Available: http://idatamining.net/blog/?p=2300
[32]TensorFlow.org,(2017 June), "MNIST For ML Beginners",Available:https://www.tensorflow.org/get_started/mnist/beginners
[33]ROS.org,(2017 June), "Creating a ROS Package",[Online],Available: http://wiki.ros.org/ROS/Tutorials/CreatingPackage
[34]cig01,(2017 June), "ROS (Robot Operating System) Tutorials",[Online],Available: http://aandds.com/blog/ros-tutorials.html
[35]T. Tan, Y. Qian, D. Yu, S. Kundu, L. Lu, K. Sim, X. Xiao and Y. Zhang, "Speaker-aware training of LSTM-RNNS for acoustic modelling", 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016.
[36]Andrey Kudryavtsev,(2017 July), "Automatic Speech Recognition Services Comparison",[Online], Available: http://blog-archive.griddynamics.com/2016/01/automatic-speech-recognition-services.html