We are submitting TimeDial to the following tracks:
- Integration: We utilize Google Vertex AI to access the exclusive Gemini 3.0 Pro model.
- Why it matters: Unlike standard chat apps, we leverage Gemini 3.0's superior reasoning and roleplay capabilities to maintain deep, historically accurate personas (Einstein, Cleopatra) that never break character, even when asked complex modern questions.
- Infrastructure: The entire application is containerized with Docker and deployed on Google Cloud Run for serverless scalability.
- Integration: We use the native audio generation capability of Gemini 3.1 Pro for ultra-low latency, natively integrated voice responses.
- Voice-First Experience: The app is designed as a "Voice-First" interface. We implemented a streaming architecture where Gemini generates audio responses natively, allowing users to have a fluid, natural conversation with history.
- Emotional Range: We utilized specific prompt and voice combinations to map characters to expressive Gemini voices.
TimeDial is an educational time machine. Instead of reading boring textbooks, students can "dial in" a historical figure and have a real-time voice conversation.
- Real-time Voice: Speak to the app, and it speaks back instantly.
- Visual Immersion: The UI pulses and reacts to the character's voice frequency.
- Fact-Checking: Powered by Gemini's grounding to ensure historical accuracy.
- AI Brain: Google Vertex AI (Gemini 3.0 Pro)
- Voice: Gemini 3.1 Pro (Audio Modality)
- Backend: Python FastAPI (Async/Await)
- Frontend: HTML5, TailwindCSS, Vanilla JS
- DevOps: Docker, Google Cloud Run
-
Clone the repo:
git clone https://github.com/engrtitooo/TimeDial.git cd TimeDial -
Set up Environment Variables: Create a
.envfile with your keys:
GOOGLE_API_KEY=your_api_key
PORT=8080- Run with Docker:
docker build -t timedial .
docker run -p 8080:8080 --env-file .env timedialMIT License - Open Source for the AI Partner Catalyst Hackathon.