What this estimates
"Latency" for a live stream means glass-to-glass: the delay from the camera to the viewer's screen. It is the sum of four stages, and this calculator models each one so you can see which stage to tune:
- Encode + package - the encoder's own delay plus segmenting the output. Low-latency presets trade a little quality for speed.
- Player live-edge buffer - the biggest lever. Players hold a few segments (or, with low-latency modes, a few parts) before they start playing, so buffer = duration × count held.
- CDN / network - propagation from origin to the edge nearest the viewer.
- Device decode - entry-level smart-TV chips decode more slowly than phones or laptops.
Typical latency by protocol
| Protocol | Typical glass-to-glass | How it gets low | Best for |
|---|---|---|---|
| Standard HLS | 15-30 s | 6-10 s segments, ~3 held | VOD-style live, wide reach |
| Standard MPEG-DASH | 10-20 s | 2-4 s segments, ~3 held | Android / smart TV live |
| LL-HLS | 2-5 s | ~0.3 s CMAF parts | Sports, interactive TV at scale |
| LL-DASH | 2-5 s | chunked CMAF | Sports, low-latency at scale |
| WebRTC | < 1 s | no segmenting | Betting, auctions, two-way |
These are planning estimates, not a measurement of your specific pipeline - real latency depends on your encoder, CDN, player settings, and network. Use the calculator to compare setups and find the stage worth optimizing.
Frequently Asked Questions
What is glass-to-glass latency?
The delay from the moment something happens in front of the camera to the moment it appears on the viewer's screen. It is the sum of encode and packaging, the player's live-edge buffer, CDN and network propagation, and device decode.
What latency do I need for live sports or betting?
For live sports, aim for under 5 seconds so viewers do not hear a neighbour cheer before they see the goal. For live betting and interactive formats you want sub-second, which in practice means WebRTC or near-real-time delivery.
Is WebRTC or LL-HLS better for low latency?
WebRTC reaches sub-second latency but is harder to scale to huge audiences and has patchier CTV support. LL-HLS and LL-DASH reach roughly 2-5 s, scale over standard CDNs, and work across more devices. Most large live streams use LL-HLS/LL-DASH; WebRTC is reserved for true real-time cases.
Why is my HLS stream 20-30 seconds behind live?
Standard HLS players hold about three segments at the live edge, so with 6-to-10-second segments the buffer alone is 18-30 s. Shrink segment duration, hold fewer segments, or switch to LL-HLS with sub-second parts to cut it.
Does the calculator send my data anywhere?
No. The estimate is computed entirely in your browser from the values you choose; nothing is uploaded.