Audio Transcription & Summarization App

A full-stack application that lets you record or upload audio, transcribe it via OpenAI’s Whisper model, and generate structured summaries using Google’s Gemini API. Transcriptions and summaries are managed by a FastAPI backend and presented through a modern React frontend.

---

🚀 Features

• Record Audio
  Capture system audio via a loopback device (e.g., “Stereo Mix” on Windows).

• Upload Audio
  Upload local audio files (.wav, .mp3, etc.) for transcription.

• Automatic Transcription
  Speech-to-text powered by OpenAI Whisper.

• Translation
  Non-English audio is auto-translated into English before transcription.

• Session-Scoped History
  Only show and summarize transcriptions from your current session.

• Summarization
  Generate a concise, structured summary of the transcription using Google’s Gemini API.

• Download Summary
  Export your summary as a PDF.

• Search
  Full-text search within your current session’s transcripts.

• Modern Frontend
  Built with React for a smooth, responsive UI.

• Robust Backend
  Powered by FastAPI and SQLite for quick development and easy deployment.

---

📦 Tech Stack

• Backend: FastAPI, SQLite, Uvicorn
• Frontend: React (Create React App)
• Transcription: OpenAI Whisper
• Summarization: Google Gemini API
• Languages: Python 3.8+, Node.js 16+

---

🔧 Installation

1. Clone the Repository

git clone https://github.com/yourusername/soundcard_testing.git
cd soundcard_testing

2. Backend Setup

cd backend
python -m venv venv
# Activate the venv
# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activate

pip install -r requirements.txt

Create a .env file in backend/ with your Gemini API key:

GEMINI_API_KEY=your_gemini_api_key_here

Start the backend server:

uvicorn main:app --reload

The backend will be available at:
http://127.0.0.1:8000

3. Frontend Setup

Open a new terminal and run:

cd frontend
npm install
npm start

The frontend will be available at:
http://localhost:3000

---

⚙️ Usage

1. Start the Backend:

   uvicorn main:app --reload

2. Start the Frontend:

   npm start

3. Open the App:
   Navigate to http://localhost:3000 in your browser.
4. Interact:
   • Click Record to capture system audio.
   • Or Upload any audio file.
   • View live transcription (auto-translated if non-English), then generate a summary.
   • Download the summary PDF or search within your session’s transcripts.

---

📄 Requirements

• Python 3.8 or higher
• Node.js 16 or higher
• A system audio loopback device (e.g., “Stereo Mix” on Windows)
• A valid Google Gemini API key
• Internet access (to download the Whisper model on first run)

---

📝 Notes

• Session-only Data: Only transcriptions made during the current server run are shown. Restarting or deleting the SQLite file clears history.
• Persistence: The SQLite database file (e.g., db.sqlite3) lives in backend/ and persists between runs unless manually deleted.
• Loopback Audio: Ensure your OS has a loopback/mix device enabled if you want to record system audio.

---

🙏 Credits

• OpenAI Whisper
• FastAPI
• React
• Google Gemini API

---

📜 License

This project is licensed under the MIT License. Feel free to use, modify, and distribute!