Update README.md

README.md

@@ -1,68 +1,55 @@
----
-title: RecToTextPro
-emoji: 🚀
-colorFrom: red
-colorTo: red
-sdk: streamlit
-app_port: 8501
-tags:
-- streamlit
-pinned: false
-short_description: Recording to Text
-license: mit
-sdk_version: 1.55.0
----
 # 🎤 RecToText Pro – Intelligent Lecture Transcriber
 
-RecToText Pro is an AI-powered web application that converts mixed Urdu and English lecture recordings into structured text. It supports Roman Urdu and English output formats and allows Excel export.
-
----
+RecToText Pro is an AI-powered web application that converts mixed Urdu and English lecture recordings into structured, clean text output.
 
 ## 🚀 Features
 
-- Upload .mp3, .wav, .m4a files
-- Automatic Urdu + English speech detection
-- Whisper-based transcription
+- Upload MP3, WAV, M4A, AAC files (Up to 200MB)
+- Automatic Urdu + English detection
+- Long audio (30–60 min) supported
 - Roman Urdu or English output
-- Text cleaning & formatting
-- Excel export (.xlsx)
-- Word count & processing time
-- Professional Streamlit UI
-- Hugging Face Spaces compatible
-
----
+- Clean paragraph formatting
+- Excel export with timestamps
+- Word export with clean story formatting
+- Language detection
+- Word count
+- Processing time display
+- Hugging Face CPU compatible
 
 ## 🛠 Tech Stack
 
 - Python
 - Streamlit
-- OpenAI Whisper
+- faster-whisper (Whisper Open-Source Model)
 - openpyxl
+- python-docx
 - pydub
 
----
+## 📦 Deployment on Hugging Face
 
-## 📦 Hugging Face Deployment
-
-1. Create a new Space
-2. Choose Streamlit SDK
-3. Upload:
+1. Create new Streamlit Space
+2. Upload:
    - app.py
    - requirements.txt
    - README.md
-4. Commit changes
-5. Wait for build to complete
-
----
+3. Add a file named `packages.txt` containing:
+   ffmpeg
+4. Commit and wait for build
 
 ## 💻 Run Locally
 
-```bash
-pip install -r requirements.txt
+pip install -r requirements.txt  
 streamlit run app.py
-# Welcome to Streamlit!
 
-Edit `/src/streamlit_app.py` to customize this app to your heart's desire. :heart:
+## ⚡ Model Selection
+
+- base → Faster, moderate accuracy
+- small → Better accuracy, slightly slower
+
+For CPU deployment, base is recommended.
+
+## 📌 Notes
 
-If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
-forums](https://discuss.streamlit.io).
+- Supports long lecture recordings.
+- AAC format supported.
+- Optimized for Hugging Face CPU Spaces.