🧠 NostalgiQ — Talk to the Past, Speak to the Future
🌟 Inspiration
We wanted to preserve memories in a more human way—by letting people have real conversations with AI recreations of their past selves, loved ones, or historical figures.
As AI becomes more personal, we asked: What if technology could help us reconnect with the voices, faces, and personalities we've lost to time?
🛠️ How We Built It
- DistilBERT + TF-IDF: Predicted personality from small text samples
- Whisper: Real-time speech-to-text
- ElevenLabs: Voice cloning
- SadTalker / D-ID / HeyGen: Talking avatars
- Python + FastAPI: Backend + async processing
- React.js + Tailwind CSS: Frontend
📚 What We Learned
- Simulating people with limited data is challenging but doable
- Syncing multi-modal AI systems in real time is non-trivial
- Users form emotional connections—ethics and consent are critical
🧩 Challenges
- Syncing video, audio, and text in real time
- Managing third-party API rate limits
- Designing fallback systems for sparse user data
- Handling ethical concerns around identity recreation
💡 Applications
Memory preservation, time capsules, legacy AI, influencer libraries, long-term collaboration, and multi-self interactions.
