DocuVoice: An AI-Powered Intelligent Document Analyzer for the Visually Impaired

Pankaj Ahirrao

 Department of Computer Engineering Pune Institute of Computer Technology

Pune, India pdahirrao25@gmail.com

Abhishek Chavan

Department of Computer Engineering Pune Institute of Computer Technology

Pune, India abhishekchavan9394@gmail.com

Roshan Patil
Department of Computer Engineering Pune Institute of Computer Technology Pune, India roshanvpatil2004@gmail.com

Sahil Patil
Department of Computer Engineering Pune Institute of Computer Technology Pune, India

ssp221004@gmail.com

 Prof. P. T. Kohok
Assistant Professor, Department of Computer Engineering Pune Institute of Computer Technology

Pune, India ptkohok@pict.edu

Abstract—Access to written information remains a persistent challenge for visually impaired individuals. Traditional screen readers and text-to-speech (TTS) systems support linear playback but fall short of enabling genuine comprehension, interactive navigation, and contextual understanding of complex documents. This paper presents DocuVoice, a full-stack, AI-powered docu-ment analysis platform designed to bridge this accessibility gap. The system integrates Optical Character Recognition (OCR), document layout analysis, Natural Language Processing (NLP), Retrieval-Augmented Generation (RAG)-based question answer-ing, and a voice-centric interface into a unified pipeline. Key fea-tures include extractive and abstractive summarization, Named Entity Recognition (NER), sentiment analysis, multilingual sup-port, automatic quiz generation, and full voice control. DocuVoice accepts diverse document formats (PDF, DOCX, scanned images) and delivers intelligent, context-aware audio responses, empow-ering visually impaired users to read, understand, and interact with documents independently.Index Terms—Accessibility, visually impaired, document anal-ysis, OCR, NLP, text-to-speech, retrieval-augmented generation, sentiment analysis, assistive technology, artificial intelligence, DocuVoice