AI-Based Intelligent Document Translator And Audio Reader For Telugu Language
Accurate machine translation for low-resource languages remains a significant challenge, particularly for technical documents where consistent terminology is essential. This work presents a dictionary-guided post-editing framework for English-to-Telugu document translation. The proposed system operates as a lightweight post-processing layer applied after translation and does not require modifications to the underlying translation model or additional training data. Terminology consistency is achieved through semantic matching with a curated Telugu glossary, ensuring uniform translation of domain-specific terms across the entire document. The system is evaluated using standard translation quality metrics and demonstrates improved performance compared to existing large-scale translation systems. The proposed method achieves a score of 9.23 on a widely used evaluation metric and 42.00 on a character-based evaluation measure, showing consistent improvement in both lexical and morphological accuracy. In addition, a speech synthesis module is integrated to convert translated text into audio form, enhancing accessibility for users. The results indicate that the proposed framework provides a practical and scalable solution for improving translation quality in low-resource language settings, particularly for technical and domain-specific content.