MomentX RT

Real-time moment detection for live streams.

MomentX RT is a lightweight server-side system built on Ant Media Server that detects important moments during a live broadcast and instantly generates timeline markers and short highlight clips.

Built for the Ant Media AI Hackathon 2026.

---

What It Does

• Monitors live streams in real time
• Detects moments using rule-based triggers
• Generates timeline markers instantly
• Creates short highlight clips during the broadcast
• Supports manual producer trigger (fallback)
• Stores confidence score and detection reason
• Provides post-stream ranked summary of moments

All processing happens while the stream is ongoing.

---

Architecture

Project Structure

/MomentX_RT/
  ├─ server/     # Core service & Ant Media integration
  ├─ triggers/   # Automatic + Manual detection logic
  ├─ clipper/    # Rolling buffer & Clip generation

---

MomentX RT operates as a modular backend service using stream IDs from Ant Media Server.

---

Technology Stack

• Ant Media Server (WebRTC streaming)
• Python-based backend service
• Real-time audio intensity detection
• FFmpeg for clip extraction
• JSON-based configuration and lightweight storage

---

Design Principles

• Low latency first
• Reliability over complexity
• Explainable detection (confidence + reason tagging)
• Modular and extensible structure

---

Hackathon

Ant Media AI Hackathon 2026
Team: error200
Members:

• Archisha Nigam
• Kshitij Pagar