Abstract—The paper introduces TaAl–Lip: Application of Artificial Neural Network in Recognition of Select Tagalog Alphabet through Lip Reading. It is an experimental study designed to determine the accuracy of the recognition of select Tagalog alphabet or also known as the ABAKADA Alphabet. It utilized Artificial Neural Network (ANN), Computer Vision (CV), Macropixelling, and Image Processing (IP) to develop a tool that can recognize mouth’s movement by means of lip reading. After the tool was developed the degree of accuracy of the recognition of the application was evaluated by the proponents according to the: (a) light orientation; (b) viewing angle and (c) the user’s distance from the camera. Based on the experiment conducted, the researchers concluded that the mouth’s movement is most recognizable with a front side light orientation with an average of 70.34%. Index Terms—Lip reading, artificial neural network, macropixelling, image processing, computer vision. I. INTRODUCTION There are two aspects of human speech: audio and visual. The audio speech signal refers to the acoustic that uses the articulatory organs together with the muscles a speaker produce speech. The problem of acoustic signal is that the accuracy of speech recognition technique is not good enough in noisy condition. [1] The process of identifying the words or letters from the movement of lips is called lip-reading. Lip-reading can solve this problem because lip-reading uses only visual features, which are not affected by noise. The lip movements contain enough information for the visual categorization of speech but these features vary from one language to another because the pronunciation is different for different languages. As for spoken language learning for hearing impaired people, aside from residual listening, lip reading is an important channel to understand the information. Lip reading or speech reading is a tool applicable for person with hearing disability. This is a way of understanding speeches by interpreting the movement of the lips and tongue when there is noise interference or normal sound is not available. With the Manuscript received December 29, 2013; revised June 20, 2014. This work was supported in part by the Polytechnic University of the Philippines. The authors are with the College of Computer and Information Sciences, Polytechnic University of the Philippines, Sta. Mesa, Philippines (e-mail: olivermembrere@yahoo.com, reysupan7@yahoo.com, jackilynmagtoto@yahoo.com, Miles_kev21@yahoo.com, bennycomendador@yahoo.com, rmmontaril@yahoo.com). complexity of the Artificial Neural Network (ANN), nowadays, many computer systems use ANN for it is developed with a systematic step-by-step procedure which optimizes a criterion commonly known as the learning rule. Artificial Neural Network refers to computing systems whose central theme is borrowed from the analogy of biological neural networks. The use of the term “Artificial” is to provide an imitation of the real thing by the use of computer technology. [2] A neural network’s ability to perform computation is based on hope that we can reproduce some of the flexibility and power of human brain by artificial means. Basically, [3] a neural network is machined that is designed to model the way in which the brain performs a particular task or function of interest. The network is usually implemented by using electronic components or is simulated in software on a digital computer. This paper is closely related to the existing study entitled “Lip-Reading using Neural Networks” wherein [4] the researchers of this study implemented this project using an evaluation version of the software NeuroSolutions5. They have used the evaluation version of the software; they got the maximum accuracy of 52%. This research project is the first attempts to use Neural Networks classification (clustering) for addressing this challenging problem that combines two different application domains of classification and predicting which brings out the much desired output. However, none of these include the effect of the variable extraction of the lips, distance, viewing angle and the light orientation while recognizing the speech. The developed tool is an application that recognizes the select alphabet by interpreting the movement of the lips and tongue where noise interference and normal sound is a not a factor. [5] Filipinos speak nine more indigenous languages all belonging to the Malayo-Polynesian group namely: Tagalog, Cebuano, Ilocano, Ilonggo, Bikolano, Waray, Visayan, Pampango and Pangasinan. Each of the nine has a number of dialects; hence, there are 87 dialects in all. Some dialects of the same language are mutually unintelligible to each other. Each of the nine languages has its own extensive literature. The oldest and the richest is Tagalog, the language extensively used in Central and Southern Luzon, and considered the basis of the national language of the Philippines. The ABAKADA alphabet is the traditional Filipino alphabet that is being used as guide in writing and speaking of Filipino words. Since the said alphabet is the starting point of learners herein the Philippines, the researchers decided to make a lip reading system using this TaAl-Lip: Application of Artificial Neural Network in Recognition of Select Tagalog Alphabet through Lip Reading Oliver M. Membrere, Reynaldo M. Supan, Jackilyn C. Magtoto, Miles Kevin B. Galario, Benilda Eleonor V. Comendador, and Ranil M. Montaril Journal of Advances in Computer Networks, Vol. 2, No. 4, December 2014 293 DOI: 10.7763/JACN.2014.V2.128