Stream latency test: how we measure stream delay and slide timing.
A stream latency test measures the delay between something happening in the room and a viewer seeing it. We measure it by putting a clock in the picture: the viewer's own clock minus the clock on screen is the glass-to-glass latency on that network. This is how we run that test on a rehearsal, and how the number sets slide timing and cues.
- Reading time
- Updated
- Glass to glass
- 0.1 s
- 1 kHz
- Rehearsal
How do you run a stream latency test?
A stream latency test measures the delay between something happening in the room and a viewer seeing it. We measure it by putting a clock in the picture. The viewer's own clock minus the clock on screen is the glass-to-glass latency on that network.
We run this on every rehearsal, with the StreamCertified Sync Clock. It is the only honest way to get the number. Estimates add up the parts and always miss one.
What glass-to-glass latency includes
"Glass to glass" means from the camera lens to the viewer's screen. It includes every stage the picture passes through:
- the camera, switcher and graphics;
- the encoder;
- the contribution link out of the venue, such as SRT or bonded cellular;
- the platform's transcode into viewer renditions;
- the CDN;
- the player's buffer on each viewer's device.
The last stage varies the most. Segment-based delivery such as HLS buffers several seconds by design. Two viewers on different devices and networks can sit seconds apart. So we measure on more than one viewer.
Why latency matters for slides, polls and cues
Latency on its own rarely upsets a viewer. Things that ignore it do.
- Slides. If slides reach viewers as data, alongside the video, they arrive before the picture. The viewer sees slide 12 while the speaker is still on slide 11. The fix is a slide delay equal to the stream latency.
- Polls and Q&A. Open a poll when the presenter asks, and remote viewers see it before they hear the question.
- Remote speakers and interpreters. Anyone working from the stream is behind the room by the full latency.
- Replays. Slides and chapter marks on the recording must line up with the picture, not with the wall clock in the room.
We delivered a two-day, eleven-session live broadcast for a global fragrance house, with slide-synced replay published per session. The timing on that show came from this measurement.
How the sync clock works
The Sync Clock is a wall-clock timecode built to be streamed.
- Time to a tenth of a second, large enough to read on a phone viewing the stream.
- A sweeping bar and a ten-dot ring, so a frozen frame is visible as a picture, not just a number that stopped.
- A 1 kHz beep on every second, so audio-only paths can be measured too.
- A moving grain layer, so a hardware encoder never starves on a near-static image and freezes the far end.
- A clock check. Everything runs on the laptop's own clock, so it must be synced to network time. The page compares itself with the server and warns when they are 2 seconds or more apart.
| What we read | What it tells us |
|---|---|
| Viewer clock minus clock in picture | Glass-to-glass latency on that viewer |
| Spread across several viewers | How much player buffering varies |
| A missing beep | A dropped frame or audio gap |
| A doubled beep | A repeated frame |
| Clock on the recording | The exact wall time of any moment, for placing slides and cues |
How we run the test on a rehearsal
- Put the clock in the programme. A laptop runs the clock fullscreen, and its HDMI goes into the switcher like any other source.
- Stream it for real. Same encoder, same venue line, same platform and player the audience will use. A test on a different path measures a different show.
- Read it on several viewers. A laptop on office broadband, a phone on 4G, a smart TV if the client expects one. Each next to its own synced clock.
- Photograph both clocks in one frame. The photo removes reaction time from the reading.
- Set the delays. The slide delay and cue timing are set from the measured figure, then checked again on a second pass.
We repeat it if anything in the chain changes: a new encoder setting, a switch to the bonded 5G second path, or a different player setting. Each path has its own latency, and the delays must match the one that is live.
Measure the network first
Latency on the day depends on the line under it. Before a rehearsal we run the venue readiness check: upload, loss, jitter and round-trip time. A jittery line forces a bigger transport buffer, and that buffer is part of the latency the clock will show. For the settings that keep delay down, see low latency streaming.
We run this routine on our live streaming in Dubai shows. If your event has slides, polls or remote speakers that must land on time, talk to us before the rehearsal.
Questions we get asked.
What is glass-to-glass latency?
The time from light entering the camera lens to the picture appearing on a viewer's screen. It includes encoding, the contribution link, transcoding, the CDN and the player buffer.
How do I measure the delay on my live stream?
Stream a clock, then compare the clock in the picture with a synced clock next to the viewer. The free StreamCertified Sync Clock is built for this.
Why are my slides ahead of the video on the stream?
Slides sent as data travel faster than the video, which is buffered. Delay the slides by the measured stream latency and they land with the picture.
Why do two viewers see different latency?
Each player buffers differently, and each viewer's network adds its own delay. Measure on several devices and set delays for the typical viewer.
01 / 06 How do you run a stream latency test?
Jitter test for live streaming: what jitter is and how we check a venue.
A jitter test measures how much network delay varies. Why jitter makes a live stream stutter, the thresholds, and how we check a venue.
Read the articleStreaming diagnostic: is your internet good enough for live streaming?
Is your internet good enough for live streaming? The streaming diagnostic we run before we quote or go live: every check, its threshold, and what it decides.
Read the articleWhat is SRT streaming? Secure Reliable Transport explained.
SRT (Secure Reliable Transport) is the video streaming protocol behind modern live broadcast. How SRT works, latency and encryption settings, SRT vs RTMP, and when to use it.
Read the articleHow to record a live stream professionally.
Recording a live stream professionally: encoders, ISO recording, archive management, and post-event delivery.
Read the articleLive from anywhere. No fibre. No landline. No excuses.
Go live from anywhere with bonded 4G/5G cellular streaming. Aggregated connections for reliable outdoor broadcasts, events, sports, and IRL streaming.
See the serviceReal-time streaming. Under 500ms, where the moment needs it.
Sub-500ms glass-to-glass streaming via SRT and WebRTC. Real-time interaction for esports, auctions, trading floors, and live Q&A events.
See the serviceConference production. Multi-stage. Multi-day. One standard.
Conference live streaming in Dubai, and summit production across the GCC. Multi-stage coverage, speaker integration, session recording and audience analytics.
See the serviceYOUNGO at COP28: the youth constituency's daily broadcast.
We produced the YOUNGO daily broadcast inside COP28. Four languages, four channels, every day for two weeks, fed by stand-up interviews and bonded 5G reporters.
Read the case studyIRENA: ministers on air inside three minutes.
The stand-up interview broadcast position at IRENA: ministers on air with a broadcast cut inside three minutes, feeding the live pipeline and same-day packages.
Read the case studyA global fragrance house: eleven sessions, replay packaged per session.
Eleven sessions over two show days for a global fragrance house: five cameras, every camera recorded on its own track, and a slide-synced replay packaged per session.
Read the case studyCDN (Content Delivery Network)
A CDN distributes video streams across geographically distributed servers, ensuring viewers connect to nearby servers for faster delivery and lower latency.
Read the definitionHLS (HTTP Live Streaming)
HLS is an adaptive streaming protocol where video is divided into small segments and a playlist tells players where to find them.
Read the definitionLow Latency Streaming
Low-latency streaming minimises the delay between live action and viewer experience. Target sub-3 second latency.
Read the definitionSRT (Secure Reliable Transport)
SRT is a modern streaming protocol designed to reliably transmit video over unpredictable networks.
Read the definitionTell us the date, the venue and who is watching.
We scope every engagement against its run of show, not a package.
CBA runs bonded cellular streaming for venues with no fibre of their own.