Conversion of Image to Text as Well as Speech for Professional Use

A Python-based accessibility tool that converts a text-bearing image into machine-readable text (OCR), then converts that text into spoken audio (Text-to-Speech), with optional translation into another language.

This project was originally developed as a B.Sc. (Engg.) final year project in the Department of Information and Communication Technology, Comilla University (2020), aimed at helping visually impaired and speech impaired users access written information more independently. The original prototype was deployed on a Raspberry Pi 3 with a Pi camera, microphone, and speaker.

Motivation

Millions of people worldwide are visually impaired or unable to speak. This project explores how commodity hardware (a camera + a low-cost single-board computer) combined with open-source Python libraries can:

• Let a visually impaired person "hear" printed text (Image → Text → Speech)
• Let a speech-impaired person "speak" by typing/scanning text that is then read aloud (Text → Speech)
• Optionally translate the recognized text into another language before speaking it

How it works

 ┌─────────────┐     ┌───────────────┐     ┌────────────────┐     ┌───────────────┐
 │ Input Image │ --> │ Pre-processing │ --> │ OCR (Tesseract) │ --> │  Text file    │
 └─────────────┘     └───────────────┘     └────────────────┘     └───────┬───────┘
                                                                            │
                                                            (optional translation)
                                                                            │
                                                                            v
                                                                    ┌───────────────┐
                                                                    │ Speech (TTS)  │
                                                                    └───────────────┘

1. Pre-processing (src/image_to_text.py) — grayscale conversion, noise reduction, and contrast enhancement so the OCR step works reliably on both plain and colored/noisy-background images.
2. OCR — text is extracted via Tesseract OCR through the pytesseract Python wrapper, and saved to a .txt file.
3. Text-to-Speech (src/text_to_speech.py) — the text file is read and converted to speech offline via pyttsx3. Translation before speaking is supported via googletrans.

Project structure

image-text-speech-converter/
├── README.md
├── requirements.txt
├── src/
│   ├── image_to_text.py     # Image -> Text (OCR) module
│   ├── text_to_speech.py    # Text -> Speech (TTS) module
│   └── main.py               # End-to-end pipeline / CLI entry point
└── samples/
    ├── input_images/         # Put sample test images here
    └── output/                # Generated .txt / audio output

Installation

1. Install the Tesseract OCR engine (required by pytesseract):

   • Ubuntu/Debian: sudo apt install tesseract-ocr
   • macOS: brew install tesseract
   • Windows: download the installer from the Tesseract project

2. Clone this repo and install the Python dependencies:

   git clone <your-repo-url>
   cd image-text-speech-converter
   pip install -r requirements.txt

Usage

Run the full pipeline on an image:

python src/main.py samples/input_images/example.jpg

Translate the recognized text into Bangla before speaking it:

python src/main.py samples/input_images/example.jpg --dest-lang es

Or run each stage independently:

# Image -> Text only
python src/image_to_text.py samples/input_images/example.jpg -o samples/output/example.txt

# Text -> Speech only
python src/text_to_speech.py samples/output/example.txt -o samples/output/example.mp3

Sample Input & Output

Input: an image containing colored text on a colored background, demonstrating that OCR still works when text isn't simple black-on-white.

Figure: Image containing colored text with a colored background

Output: the recognized text, extracted via OCR and saved to a .txt file.

Figure: Extracted text saved to text_file.txt

Original project context

• Title: Conversion of Image to Text as Well as Speech for Professional Use
• Institution: Department of Information and Communication Technology, Comilla University
• Author: Sharmin Akter
• Supervisor: Md Ariful Islam — Assistant Professor, Department of Robotics and Mechatronics Engineering, University of Dhaka (Ex-Lecturer, Dept. of ICT, Comilla University)
• Supervisor: Md Ariful Islam — Assistant Professor, Department of Robotics and Mechatronics Engineering, University of Dhaka (Ex-Lecturer, Dept. of ICT, Comilla University)
• Original hardware: Raspberry Pi 3, Pi Camera, microphone, speaker
• Original tools: Python 3, Tesseract OCR, Pytesseract, PIL, pyttsx3, Google TTS/Translate

References

This project builds on the methodology described in the following works:

1. Shah, T., & Parshionikar, S. (2019). Efficient Portable Camera Based Text to Speech Converter for Blind Person. International Conference on Intelligent Sustainable Systems (ICISS 2019).
2. A Text Reader for the Visually Impaired using Raspberry Pi. Proceedings of the Second International Conference on Computing Methodologies and Communication (ICCMC 2018).
3. Rithika, H., & Santhoshi, B. N. Image Text To Speech Conversion In The Desired Language By Translating With Raspberry Pi.
4. Smith, R. (2007). An Overview of the Tesseract OCR Engine. International Conference on Document Analysis and Recognition.
5. Rice, S. V., Jenkins, F. R., & Nartker, T. A. (1995). The Fourth Annual Test of OCR Accuracy. Information Science Research Institute.
6. Bhargava, A., Nath, K. V., Sachdeva, P., & Samel, M. (2015). Reading Assistant for the Visually Impaired. International Journal of Current Engineering and Technology (IJCET), 5(2).
7. Devi, V. A., & Baboo, S. S. (2014). Optical Character Recognition on Tamil Text Image Using Raspberry Pi. International Journal of Computer Science Trends and Technology (IJCST), 2(4).
8. Kumari, K. N., & Reddy, M. J. (2016). Image Text to Speech Conversion Using OCR Technique in Raspberry Pi. International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering (IJAREEIE), 5(5).
9. Tesseract OCR Engine — github.com/tesseract-ocr/tesseract

License

MIT — see LICENSE.