NDI in the Cloud: Convert SRT to NDI Devices in Minutes

Moving live production to the cloud no longer means giving up NDI. With a modern gateway, you can ingest SRT from anywhere, convert it to NDI in the cloud, and expose those NDI sources to your mixers—locally or remotely.

Why SRT → NDI in the cloud

• Global capture: Field encoders push SRT through the public internet with ARQ and encryption.
• Production-ready I/O: Convert streams to NDI to feed switchers, graphics, replay, and multiviews.
• Elastic compute: Spin up instances per event, then shut them down to save costs.

Reference architecture

1. Ingress: Contributors send SRT (caller) to a public SRT server (listener) with main/backup ports.
2. Gateway: A cloud app receives SRT, normalizes audio/video, and exposes each feed as an NDI source.
3. Discovery: Deploy an NDI Discovery Server in the same VPC/subnet. Point adapters and mixers to it.
4. Production tools: vMix/OBS/wirecast-style switchers connect to the cloud NDI sources or pull return feeds.
5. Distribution: Encode ladders to HLS/DASH, or return SRT/RTMP to platforms and affiliates.

Step-by-step setup

1. Create two SRT listeners (main/backup). Share hostname/port/passphrase with contributors.
2. Map each incoming SRT to an NDI output. Name clearly (e.g., Cam-1-Paris).
3. Start the NDI Discovery Server. Set environment variables or config pointing all nodes to it.
4. Add NDI sources in your switcher. Confirm resolution, fps, and audio mapping.
5. Optionally publish a web player for QA so non-NDI users can monitor in a browser.

Latency & sync tips

• Use SRT latency that matches path quality (e.g., 80–200 ms). Don’t overtune.
• Align frame rates and color spaces across cameras to reduce conversions.
• For multi-cam sports, use timestamp offsets or genlock-like features to line up shots.

Security & reliability

• Protect SRT with passphrases and firewall allow-lists.
• Run dual gateways across AZs; route traffic automatically on failure.
• Use IAM roles for storage and API access; rotate credentials.

Right-sizing instances

Match CPU/GPU/FPGA to your workloads: CPU for light I/O, GPU/FPGA for dense ladders and HEVC. Keep NDI and discovery in the same zone to minimize jitter.

Result: You get the reliability of SRT for contribution and the flexibility of NDI for production—fully cloud-based, fast to set up, and easy to scale.