DocJarvis- AI Medical Assistant

An AI-powered multilingual voice assistant for medical consultations. DocJarvis conducts initial patient assessments through voice interaction, supports 10 Indian languages, and generates consultation summaries.

Disclaimer: This is an AI-assisted tool for educational purpose only. Always consult qualified healthcare professionals for medical advice.

Features

• Voice-based interaction: Speech-to-text and text-to-spech capabilities

• Multilingual support: English +9 regional Indian languages

• AI-powered diagnosis: Uses Google's Gemini Pro for intellignet questioning

• Prescription generation: Automated consultation summary documents

• User-friendly interface: Gradio-based web UI

Supported Languages

Language	Code	Language	Code
English	en	Malayalam	ml
Bengali	bn	Marathi	mr
Gujarati	gu	Tamil	ta
Hindi	hi	Telugu	te
Kannada	kn	Urdu	ur

Quick Start

Prerequisites

• Python 3.10+
• Working microphone
• Internet connection
• Google AI Studio API key

Installation

1. Clone the repository

git clone https://github.com/singhdivyank/voice-assistant.git

cd docjarvis

2. Create virtual environment

python -m venv venv

source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Configure environment

cp .env .example .env

5. Run the application

python -m src.app

PyAudio installation on Mac:

brew install portaudio

python3 -m pip install pyaudio

PyAudio installation on Linux:

sudo apt-get install python3-pyaudio portaudio19-dev

pip install pyaudio

For audio playback on Linux, install one of these players:

# Recommended
sudo apt-get install mpg123

# Alternatives
sudo apt-get install mpg321
sudo apt-get install ffmpeg  # provides ffplay

PyAudio installation on Windows: Audio playback uses PowerShell's built-in Media.SoundPlayer (no additional installation required)

Configuration

Create a .env file with the following variables:

GOOGLE_API_KEY=your_gemini_api_key_here

# Optional (defaults shown)
GRADIO_SERVER_NAME=0.0.0.0
GRADIO_SERVER_PORT=7860

Usage

1. Open the web interface (default: http://localhost:7860)
2. Select your preferred language
3. Enter your gender and age
4. Click Submit to start the consultation
5. Speak your symptoms when prompted
6. Answer follow-up questions verbally
7. Receive your consultation summary

Project Structure

voice-assistant/
├── src/
│   ├── config/          # Configuration and settings
│   │   └── settings.py  # App config, prompts, enums
│   ├── core/            # Core business logic
│   │   ├── diagnosis.py # LLM-powered diagnosis
│   │   └── prescription.py
│   ├── services/        # External service integrations
│   │   ├── speech.py    # STT/TTS services
│   │   └── translation.py
│   ├── utils/           # Utilities and helpers
│   │   ├── exceptions.py
│   │   └── file_handler.py
│   └── app.py           # Main application
├── requirements.txt
└── README.md

API Reference

DiagnosisService

from src.core.diagnosis import DiagnosisService, PatientInfo
from src.config.settings import Gender

service = DiagnosisService()
patient = PatientInfo(age=30, gender=Gender.MALE)

# Create session with diagnostic questions
session = service.create_session(patient, "I have a headache")

# Add patient responses
service.add_response(session, 0, "It started yesterday")

# Get recommendations
recommendations = service.complete_session(session)

TranslationService

from src.services.translation import TranslationService
from src.config.settings import Language

translator = TranslationService(Language.HINDI)

# Translate for LLM (to English)
english_text = translator.to_english("सिरदर्द है")

# Translate for user
hindi_text = translator.to_user_language("Take rest")