Voice-Enabled AI Chatbot using Rasa 🎙️🤖

📌 Why I Made This Project

In an era of growing reliance on voice-driven technologies like Siri and Alexa, I wanted to explore the intersection of Natural Language Processing (NLP), conversational AI, and real-time voice interaction. This project aims to create a browser-based voice assistant that enables hands-free, intelligent conversations—providing an accessible and user-friendly solution for education, customer service, and daily assistance.

🌐 Project Overview

This is an end-to-end Voice-Enabled Chatbot built using Rasa (NLP framework) and Web Speech API, integrated with a custom JavaScript-based frontend and a high-performance backend using Sanic. The chatbot can:

• Recognize spoken queries via microphone 🎤
• Classify intents and manage dialogue using Rasa 📚
• Respond through text and convert it back to speech 🔊

💼 Business Problem

Typing long queries is inconvenient, especially for users who are visually impaired, multitasking, or using mobile devices. Businesses need:

• A fast and natural input method (voice)
• Scalable, intelligent assistants that can understand and respond accurately
• Cost-effective, open-source conversational solutions

✅ Solution

This project solves the above challenges by:

• Enabling hands-free voice communication via browser
• Using Rasa for flexible and accurate intent classification
• Implementing real-time interaction with minimal latency
• Supporting further expansion into domain-specific assistants (e.g., education, healthcare)

🛠️ Technologies Used

Layer	Tools/Frameworks
Frontend	HTML, CSS, JavaScript, Web Speech API
Backend Server	Python, Sanic (Async web server)
NLP Core	Rasa (DIETClassifier, RulePolicy, etc.)
Voice Interface	PyAudio (backend audio capture), Web APIs
Deployment Tools	Git, VS Code, Localhost (for testing)

🧠 Key Features

• Voice-to-text input and speech output
• 92%+ accuracy in intent recognition (30+ intents)
• Responsive web interface with real-time feedback
• Error handling and fallback responses
• Custom action support for dynamic replies

📈 Impact of the Project

• Enhanced accessibility for non-typers and visually impaired users
• Achieved under 1.2 seconds response time with async handling
• Delivered a scalable, modular chatbot architecture
• Gained real-world experience in NLP, full-stack development, and voice UI integration

🔮 Future Work

• Integrate advanced ASR models like Whisper or Vosk for improved voice recognition
• Add multilingual support and sentiment-based routing
• Improve noise filtering and optimize for mobile browsers
• Deploy via Docker with HTTPS and database integration
• Extend to domain-specific applications (e.g., medical assistant, student helpdesk)

---

🚀 Get Started (Local Setup)

# Clone the repo
git clone https://github.com/yourusername/voice-rasa-chatbot.git
cd voice-rasa-chatbot

# Install dependencies
pip install -r requirements.txt

# Start Rasa server
rasa run --enable-api

# Start action server (if using custom actions)
rasa run actions

# Start Sanic backend
python server.py

# Open index.html in browser