Voice-to-FIR Generator

A system that automates First Information Report (FIR) generation from voice recordings using AI. This project integrates Whisper AI for speech-to-text conversion and Gemini AI for FIR generation. It also includes a case status search API and fine-tuned Qwen models with RAG implementation for Indian law-based queries.

---

Features

• 🎙 Audio-to-Text: Converts victim's voice recordings to text using Whisper AI.
• 📜 FIR Generation: Uses Gemini AI to generate structured FIR documents.
• 🔍 Case Status Search API: Allows searching case details using Case ID.
• 🤖 Fine-tuned Qwen (0.5 - 1.5) Models: Trained on Indian legal datasets for better FIR accuracy.
• 📚 RAG (Retrieval-Augmented Generation): Enhances model responses with legal knowledge retrieval.

---

Structure

• Main Branch (main): Contains Whisper AI and Gemini AI FIR generation.
• Fine-Tuning Repo: [https://github.com/ajf1016/Fine-Tuning-Qwen1.5-0.5B] [https://github.com/ajf1016/Fine-Tuning-Qwen1.5-0.5B]
• RAG Implementation Repo: [https://github.com/ajf1016/Qwen-RAG-for-legal-queries]

Installation

1. Clone the Repository

   git clone https://github.com/ajf1016/AI-Voice-to-FIR-Converter-.git
   cd voice-to-fir

2. Set Up the Virtual Environment

   python -m venv venv
   source venv/bin/activate  # On Windows use: venv\Scripts\activate
   

3. Install Dependencies

   pip install -r requirements.txt
   

4. Set Up Environment Variables Create a .env file and add:

   OPENAI_API_KEY=your_openai_api_key
   GOOGLE_API_KEY=your_google_api_key
   

5. Run the Django Server

   python manage.py runserver
   

API Endpoints

1️⃣ Upload & Convert Audio Endpoint: POST /api/upload/ Description: Uploads an audio file and converts it into text using Whisper AI.

Request Example (form-data):

audio_file: <file.mp3>

2️⃣ Generate FIR Endpoint: POST /api/fir// Description: Generates an FIR using Gemini AI from transcribed text.

Headers:

Authorization: Bearer <token>

3️⃣ Search Case Status Endpoint: GET /api/fir/CASE-<case_id>/ Description: Retrieves FIR details using a Case ID.

4️⃣ User Registration Endpoint: POST /api/register/ Request Body (JSON):

{
  "full_name": "Test User",
  "uid": "123456",
  "phone": "0000000000",
  "password": "securepassword"
}

5️⃣ User Login Endpoint: POST /api/login/ Request Body (JSON):

{
  "uid": "123456",
  "password": "securepassword"
}

Acknowledgment

This project utilizes datasets from another GitHub repository for model training. Special thanks to the original dataset creator. 🎖 https://github.com/civictech-India/Indian-Law-Penal-Code-Json/tree/main

Contributing

Contributions are welcome! Feel free to open an issue or pull request. For major changes, please discuss them in advance.