https://github.com/harsha-yuvaraj/iris-voice-ai
A voice-to-voice conversational AI built with Django, Deepgram, OpenAI, and Twilio—designed with smart time-wasting capabilities. Live now! Call & Chat at +1 956 952 7270!
https://github.com/harsha-yuvaraj/iris-voice-ai
ai-voice-assistant amazon-web-services asynchronous-programming deepgram django django-channels docker javascript openai python redis speech-to-text text-to-speech twilio-voice voice-chat websockets
Last synced: 3 months ago
JSON representation
A voice-to-voice conversational AI built with Django, Deepgram, OpenAI, and Twilio—designed with smart time-wasting capabilities. Live now! Call & Chat at +1 956 952 7270!
- Host: GitHub
- URL: https://github.com/harsha-yuvaraj/iris-voice-ai
- Owner: harsha-yuvaraj
- Created: 2025-03-16T20:24:36.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2025-04-03T01:03:28.000Z (3 months ago)
- Last Synced: 2025-04-03T02:19:47.339Z (3 months ago)
- Topics: ai-voice-assistant, amazon-web-services, asynchronous-programming, deepgram, django, django-channels, docker, javascript, openai, python, redis, speech-to-text, text-to-speech, twilio-voice, voice-chat, websockets
- Language: Python
- Homepage: https://irisvoiceai.tech
- Size: 59.6 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Iris Conversational Voice AI
**A voice-to-voice AI** built with **Django**, **Deepgram**, **OpenAI** and integrated with **Twilio**. Iris is designed to deliver a fun and engaging conversational experience with smart time wasting.
### **Check out the project live**: [irisvoiceai.tech](https://irisvoiceai.tech)
### **Talk to Iris**: Call +1 956-952-7270
# Update
Iris is currently down for upgrades — enhanced speech detection & response generation along with UI improvements.
## Features
- **Real-Time Voice Interaction**:
Django channels and asynchronous websockets for real-time voice exchange.
- **Speech-to-Text (STT) and Text-to-Speech (TTS)**:
Deepgram Nova-3 is used to transcribe user speech accurately in real time and Deepgram Aura model for converting text responses to speech.
- **Response Generation**:
OpenAI GPT-4o Mini tuned to process transcribed text to generate conversational responses.
- **Twilio Integration**:
Enabling inbound phone calls, allowing users to interact with Iris via phone.
- **Session-based Memory**:
Redis is used to simulate conversation memory by storing the conversation history mapped to a user session, providing context for response generation.
- **Responsive and Fast**:
Optimized asynchronous processing for low latency and a responsive experience for users.
## Deployment
- The application is containerized with **Docker** and pushed to **AWS Elastic Container Registry**.
- Hosted on **AWS EC2** with **Nginx** acting as a reverse proxy, ensuring secure access over **HTTPS**.
- **Redis** is used for session and cache management, with AWS ElastiCache in production.