# OvenMediaEngine — Sub-Second Latency Live Streaming Server

> An open-source, enterprise-grade live streaming server that delivers WebRTC and low-latency HLS streams with sub-second latency for large-scale broadcasting.

## Install

Save in your project root:

# OvenMediaEngine — Sub-Second Latency Live Streaming Server

## Quick Use
```bash
# Run with Docker
docker run -d --name ome 
  -p 1935:1935 -p 3333:3333 -p 3334:3334 
  -p 8080:8080 -p 10000-10005:10000-10005/udp 
  airensoft/ovenmediaengine:latest
# Push RTMP stream
ffmpeg -re -i input.mp4 -c:v libx264 -c:a aac -f flv rtmp://localhost:1935/app/stream
# Watch via WebRTC at http://localhost:8080
```

## Introduction
OvenMediaEngine (OME) is a sub-second latency live streaming server built for large-scale, high-definition broadcasting. It ingests streams via RTMP, SRT, or WebRTC and distributes them to viewers through WebRTC, LLHLS (Low-Latency HLS), and LLDASH with latency as low as 0.3 seconds.

## What OvenMediaEngine Does
- Ingests live streams from encoders via RTMP, SRT, MPEG-TS, and WebRTC protocols
- Distributes content to viewers through WebRTC, LLHLS, LLDASH, and HLS
- Transcodes streams on the fly using hardware acceleration (NVIDIA NVENC, Intel QSV)
- Records live streams to MP4 or HLS segment files for on-demand replay
- Scales to thousands of concurrent viewers with built-in origin-edge clustering

## Architecture Overview
OME is written in C++ for high performance. It uses a modular pipeline architecture: ingest modules receive encoder streams, a media router directs them to transcoder and publisher modules. The transcoder handles codec conversion and adaptive bitrate ladder generation. Publisher modules format output for each delivery protocol. An origin-edge architecture lets multiple edge servers pull from a single origin to scale viewership horizontally.

## Self-Hosting & Configuration
- Deploy via Docker with a single command or build from source on Ubuntu/CentOS
- Configure applications, streams, and output protocols in `Server.xml`
- Enable hardware transcoding by installing NVIDIA drivers and setting GPU options in config
- Set up TLS certificates for HTTPS-based protocols (WebRTC, LLHLS)
- Add edge servers that connect to the origin for horizontal scaling

## Key Features
- Sub-second latency via WebRTC delivery to standard web browsers
- Adaptive bitrate streaming with server-side transcoding and bitrate ladder configuration
- Origin-edge architecture for scaling from dozens to millions of viewers
- Built-in recording to MP4 and HLS segments for VOD replay
- NVIDIA NVENC and Intel QSV hardware transcoding support for reduced CPU load

## Comparison with Similar Tools
- **Nginx-RTMP** — lightweight RTMP server; simple to set up but no WebRTC output or adaptive bitrate transcoding
- **SRS (Simple Realtime Server)** — feature-rich streaming server; strong RTMP/HLS support but less focus on sub-second WebRTC delivery
- **Janus Gateway** — WebRTC gateway; more flexible plugin model but requires more custom development for streaming workflows
- **Wowza** — commercial streaming platform; more polished management UI but proprietary and expensive

## FAQ
**Q: How low can the latency get?**
A: WebRTC delivery achieves 0.3-0.5 second latency end-to-end. LLHLS typically delivers 2-3 seconds. Standard HLS is 6-10 seconds.

**Q: How many viewers can a single server handle?**
A: Depends on transcoding load and output protocol. Without transcoding, a single server can handle thousands of WebRTC viewers. Use edge servers for larger audiences.

**Q: Does it support adaptive bitrate?**
A: Yes. Configure a transcoding profile with multiple resolution and bitrate renditions. The server generates all variants and lets clients switch based on network conditions.

**Q: Can I use it for video conferencing?**
A: OME is optimized for one-to-many live streaming rather than many-to-many conferencing. For conferencing, consider Janus Gateway or LiveKit.

## Sources
- https://github.com/OvenMediaLabs/OvenMediaEngine
- https://airensoft.gitbook.io/ovenmediaengine/

---
Source: https://tokrepo.com/en/workflows/asset-1167dad1
Author: AI Open Source