AI Interpret — real-time speech translation for events

Challenge

The goal was to make interpretation accessible without requiring attendees to install an app or use dedicated hardware. The system combines event rooms, streaming audio, speech recognition, translation and speech synthesis with low latency.

Solution

• Speaker and Listener roles with QR-based access
• Deepgram STT and streaming speech processing
• LLM translation and Azure Speech TTS
• Real-time room state over WebSockets
• PWA interface for fast mobile access

Outcome

The product is designed as a service for conferences, training sessions and multilingual events, with room to extend event management, limits, analytics and local deployments.