OCR → Voice

A small but handy tool that pulls text out of images and reads it aloud. Upload a picture, let Azure do the heavy lifting (OCR + speech synthesis), and get both the extracted text and a downloadable audio file — all from a single browser tab.

Built with Streamlit on the frontend and Azure Cognitive Services on the backend.

---

What It Does

1. You upload an image (screenshot, photo of a document, scanned page, whatever).
2. Azure's Computer Vision API reads every line of text in that image.
3. Azure's Speech Service turns that text into natural-sounding speech.
4. You get the extracted text on screen, word/character counts, and a .wav file you can play or download.

That's it. No accounts to create inside the app, no complicated steps.

---

Quick Look at the Stack

Layer	Tech
UI	Streamlit (Python)
OCR	Azure Computer Vision (Read API)
Speech	Azure Speech Service (Text-to-Speech)
Config	.env file via python-dotenv

---

Getting Started

Prerequisites

• Python 3.9+ installed on your machine.
• An Azure account with two resources provisioned:
  • Computer Vision — for the OCR part.
  • Speech Service — for text-to-speech.

1. Clone or download this repo

git clone https://github.com/prathamtagad/Readify-AI.git
cd Readify-AI

2. Install dependencies

pip install -r requirements.txt

3. Set up your Azure keys

Create a .env file in the project root (there's already a template in the repo). Fill in your own credentials:

VISION_KEY=your_vision_api_key_here
VISION_ENDPOINT=https://your-resource-name.cognitiveservices.azure.com/
SPEECH_KEY=your_speech_api_key_here
SPEECH_REGION=eastus

Heads up: Never commit real API keys to a public repo. The .env file should be in your .gitignore.

4. Run the app

streamlit run Abishek_sir_project_azure_ocr_to_speech.py

Streamlit will open a new browser tab at http://localhost:8501. That's your app.

---

How to Use

1. Click the upload area and pick an image (PNG, JPG, BMP, TIFF, or WebP).
2. Hit "✨ Extract Text & Generate Speech".
3. Wait a few seconds — the spinner will show progress.
4. Once done, the left panel shows the extracted text with stats (word count, character count), and the right panel has the audio player plus a download button.
5. The audio plays automatically in your browser — no extra clicks needed.

For best results, use images where the text is clearly visible. High-contrast, well-lit photos work best. Blurry or heavily stylized text might not come through perfectly.

---

Project Structure

.
├── Abishek_sir_project_azure_ocr_to_speech.py   # Main app — everything lives here
├── .env                                          # Your Azure API keys (not committed)
└── README.md                                     # You're reading this

It's a single-file project on purpose. Keeps things simple and easy to hand off or deploy.

---

Supported Image Formats

• PNG
• JPEG / JPG
• BMP
• TIFF
• WebP

---

Troubleshooting

"Missing Azure credentials" error on launch? → Double-check your .env file. Make sure the variable names match exactly: VISION_KEY, VISION_ENDPOINT, SPEECH_KEY, SPEECH_REGION.

OCR returns empty text? → The image might be too blurry, too small, or the text might be in a language/script Azure doesn't handle well. Try a cleaner scan.

Speech synthesis fails or gets canceled? → Usually a quota or region issue on the Azure side. Verify your Speech resource is active and the region in .env matches the one you created the resource in.

---

Deployment

The easiest way to deploy Readify AI is using Streamlit Community Cloud:

1. Push your code to a GitHub repository.
2. Go to share.streamlit.io and connect your account.
3. Click "New app" and select your repo/branch.
4. Important: Before clicking Deploy, go to Advanced Settings -> Secrets and paste your Azure credentials there:

   VISION_KEY = "your_key"
   VISION_ENDPOINT = "your_endpoint"
   SPEECH_KEY = "your_key"
   SPEECH_REGION = "your_region"

5. Click Deploy!

---

License

This project doesn't currently ship with a license file. If you're planning to share it, consider adding an MIT or Apache 2.0 license.

---

Built by Pratham Tagad — GitHub · Portfolio · prathamtagad0@gmail.com