Recitation of the Quranic verses correctly is an important deed for Muslims. To attain an accurate recitation of Quran, it is important to master the pronunciations technique of the Arabic letters. The traditional way of teaching and learning al-Quran is time consuming and requires a teacher who are expert in Quranic recitation to guide on the correct pronunciation of Makhraj and proper Tajweed, which mostly is done face-to-face. It is significant to provide an interactive platform to assist the teachers as well as expediting the students. This is proposed through the vocal tract shape estimation based on linear predictive analysis (LPC) of speech that can provide a visualized feedback to students during the practice session. This paper conducts the linear prediction analysis to estimate the vocal tract shape on two Arabic letters that are articulated from the deepest part of the throat. The results show that the vocal tract shape can be estimated using linear prediction analysis for these two letters and the shapes are similar to the 2D validated model that is used in the Quranic teaching.