This article presents a sentence classification of the Tal-Patra script, the most complex challenge in the established research community. The main objective of this research to create a systematic catalog that allows for easy retrieval and references of manuscripts, to create a digital record of manuscript for long-term preservation and global accessibilty. The sentence of Tamil Tal-Patra manuscript was categorized using a Convolutional Neural Network with Bi-directional Long Short-Term Memory networks (CNN+BLSTM) with Natural Language Processing techniques (NLP), beginning with speech-to-text conversion, which used UTF-8 (Unicode Transformation Format- 8bits) Unicode for classification, tokenization, word embedding, and script categories. The dataset used for the sentence classification is the Agaporul Veena manuscript collected from Tamil Virtual Academy, Chennai, which includes Kurinji Pattu, Nedunalvaadai, Pattina Palai, Perumpan Atrupadai, Madurai Kanchi, Malaipadukadam and Mullai Pattu. The sentence classification of the Tamil Tal-Patra Manuscript obtained a cumulative accuracy of 96.5%, which led to the newest findings of the benchmark of the Tamil Tal-Patra manuscript analysis.