TODAY

Today is a FastAPI-based application that enables users to transcribe and store their diary entries through an online speech-to-text system.

Features

• Speech-to-Text Transcription: Uses Whisper ASR for converting spoken words into text.
• Real-time Processing: Utilizes WebSocket connections to process audio streams in real time.
• Diary Saving & Retrieval: Stores diary entries with timestamps and enables users to retrieve past entries.
• Grammar Correction: Uses an LLM to correct grammatical mistakes in diary entries you can use any open source LLM you like.
• privacy: completly private to you No APIs used

Technologies Used

• Whisper ASR: Automatic Speech Recognition model for transcriptions.
• VLLM: Integration for LLM-based text processing.
• FastAPI: API framework for building scalable applications.
• WebSockets: For real-time communication with the transcription backend.

Installation

Prerequisites

Ensure you have the following installed:

• Python 3.12
• FFmpeg
• Whisper ASR dependencies
• Hugging Face API Token (for LLM access)
• cudnn
• Docker and Docker Compose (for containerized setup)

Setup

Option 1: Docker Installation (Recommended)

1. Make sure you have Docker and Docker Compose installed
2. clone the repo

git clone https://github.com/Mahmoud-ghareeb/today.git

3. Create a .env file with your Hugging Face token:

cp .env.example .env

4. Add your HF token to the .env file
5. Build and run the container:

docker compose up --build -d

6. The application will automatically start when you run docker-compose up. Access it at:

• Web Interface: http://localhost:8008

Option 2: Local Installation

1. create conda env

conda create -n today python==3.12
conda activate today

2. install pytorch

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126

3. install dependencies

sudo apt install portaudio19-dev && python3-pyaudio

4. install the requirements

pip install -r requirements.txt

5. install cudnn

6. Set up environment variables: make a copy of .env.example file rename it to .env and fill the required information

7. To start the FastAPI server:

python main.py

API Endpoints

WebSocket Endpoint

• /asr: Accepts audio streams and transcribes them in real time.

REST Endpoints

• GET /: Serves the static HTML interface.
• POST /save: Saves a diary entry.
• GET /get: Retrieves a diary entry by timestamp.
• POST /diary: Converts raw transcription into a diary format.
• POST /correct_mistakes: Corrects grammatical mistakes in the text.

Usage

1. Open the application interface in a browser.
2. Start speaking, and the app will transcribe in real time.
3. Save the transcribed text as a diary entry.
4. Retrieve and edit past entries as needed.
5. Use the AI-powered grammar correction and diary formatting features.

Under Development

• Correct Mistakes
• Diary Formatting
• Docker Support

Todos

• Dark mode
• Chat feature

Future Enhancements

• Support for multiple languages.

Contributions

Contributions are welcome! Feel free to open an issue or submit a pull request.

Contact

For any questions or support, reach out to mahmoudghareeb11111@gmail.com.