Project Overview

Welcome to the Blind Student Aid AI project. This tool leverages Google Cloud’s Vision AI and Gemini Pro to convert handwritten educational materials into structured, natural-sounding audio.

0.0.1 Quick Links

Read the Story: Learn about the UPSC motivation behind this project.
Explore the Code: See the end-to-end Python implementation.
GitHub Repo: View the source and contribute.

1 📝 Empowering Blind Students: Handwritten Note to Audio Converter

An AI-powered pipeline designed to bridge the accessibility gap for visually impaired students by transforming handwritten study materials into structured, consistent, and high-quality audio podcasts.

1.1 🌟 Motivation

In India, many bright and capable blind students—including those preparing for highly competitive exams like the UPSC—lack access to digitized study materials. They often rely on volunteers to manually dictate handwritten notes from other students.

This process has a significant hurdle: every volunteer has a different accent, cadence, and speaking style. For a student trying to focus on complex subjects for hours, this inconsistency is a major cognitive drain.

This project was born to solve that. By using AI to digitize handwriting and synthesize it into a singular, high-definition voice, we provide students with a consistent, “podcast-like” learning experience. Now, a student can listen to 100 pages of notes and know that the voice, speed, and quality will remain identical throughout.

1.2 🚀 Key Features

Handwriting to Digital: Uses Google Cloud Vision AI to extract text from dense, handwritten PDF scans.
Intelligent Refinement: Employs Gemini 1.5 to fix OCR errors and structure raw notes into logical Markdown.
Consistent Audio Experience: Converts text into lifelike speech using Google Cloud TTS (Chirp3-HD).
Podcast-Style Segmentation: Automatically splits long topics into manageable 12-minute audio “chapters” for easier navigation.

1.3 🛠️ Technical Stack

Core: Python, Google Colab
Vision: Google Cloud Vision API (OCR)
Intelligence: Google Gemini API (Text Correction & Structuring)
Speech: Google Cloud Text-to-Speech (Chirp3-HD - en-IN voice)

1.4 📋 How to Use

Prerequisites:
- A Google Cloud Project with Vision and Text-to-Speech APIs enabled.
- A Gemini API Key from Google AI Studio.
- A service account JSON key for Google Cloud authentication.
Setup:
- Upload your handwritten notes (PDF) to the environment.
- Store your GEMINI_API_KEY in your environment secrets.
- Update the PDF_FILE_PATH and GOOGLE_CLOUD_SERVICE_ACCOUNT_KEY_PATH in the configuration cell.
Execution:
- Run all cells in the BlindStudentAid.ipynb notebook.
- The workflow will output a cleaned_text.txt file and a folder of .mp3 audio segments.

1.5 📂 Project Structure

├── BlindStudentAid.ipynb   # Main execution notebook
├── audio_segments/         # Generated MP3 files (output)
├── extracted_images/       # Intermediate page scans for OCR
└── cleaned_text.txt        # The AI-corrected transcript

Created to ensure that a student’s ambition is never limited by their access to knowledge.

--- title: "Project Overview" --- Welcome to the **Blind Student Aid AI** project. This tool leverages Google Cloud's Vision AI and Gemini Pro to convert handwritten educational materials into structured, natural-sounding audio. ### Quick Links * [**Read the Story**](motivation.qmd): Learn about the UPSC motivation behind this project. * [**Explore the Code**](BlindStudentAid.ipynb): See the end-to-end Python implementation. * [**GitHub Repo**](https://github.com/RohanRanshinge/blind-student-aid-ai): View the source and contribute. --- # 📝 Empowering Blind Students: Handwritten Note to Audio Converter An AI-powered pipeline designed to bridge the accessibility gap for visually impaired students by transforming handwritten study materials into structured, consistent, and high-quality audio podcasts. ## 🌟 Motivation In India, many bright and capable blind students—including those preparing for highly competitive exams like the **UPSC**—lack access to digitized study materials. They often rely on volunteers to manually dictate handwritten notes from other students. This process has a significant hurdle: every volunteer has a different accent, cadence, and speaking style. For a student trying to focus on complex subjects for hours, this inconsistency is a major cognitive drain. **This project was born to solve that.** By using AI to digitize handwriting and synthesize it into a singular, high-definition voice, we provide students with a consistent, "podcast-like" learning experience. Now, a student can listen to 100 pages of notes and know that the voice, speed, and quality will remain identical throughout. ## 🚀 Key Features * **Handwriting to Digital:** Uses Google Cloud Vision AI to extract text from dense, handwritten PDF scans. * **Intelligent Refinement:** Employs **Gemini 1.5** to fix OCR errors and structure raw notes into logical Markdown. * **Consistent Audio Experience:** Converts text into lifelike speech using **Google Cloud TTS (Chirp3-HD)**. * **Podcast-Style Segmentation:** Automatically splits long topics into manageable 12-minute audio "chapters" for easier navigation. ## 🛠️ Technical Stack * **Core:** Python, Google Colab * **Vision:** Google Cloud Vision API (OCR) * **Intelligence:** Google Gemini API (Text Correction & Structuring) * **Speech:** Google Cloud Text-to-Speech (Chirp3-HD - `en-IN` voice) ## 📋 How to Use 1. **Prerequisites:** - A Google Cloud Project with Vision and Text-to-Speech APIs enabled. - A Gemini API Key from [Google AI Studio](https://aistudio.google.com/). - A service account JSON key for Google Cloud authentication. 2. **Setup:** - Upload your handwritten notes (PDF) to the environment. - Store your `GEMINI_API_KEY` in your environment secrets. - Update the `PDF_FILE_PATH` and `GOOGLE_CLOUD_SERVICE_ACCOUNT_KEY_PATH` in the configuration cell. 3. **Execution:** - Run all cells in the `BlindStudentAid.ipynb` notebook. - The workflow will output a `cleaned_text.txt` file and a folder of `.mp3` audio segments. ## 📂 Project Structure ```text ├── BlindStudentAid.ipynb # Main execution notebook ├── audio_segments/ # Generated MP3 files (output) ├── extracted_images/ # Intermediate page scans for OCR └── cleaned_text.txt # The AI-corrected transcript ``` *Created to ensure that a student's ambition is never limited by their access to knowledge.*