Low Latency Streaming
Low-latency streaming minimises the delay between live action and viewer experience.
Where we use this- End to end: encoder settings, transport protocol and the player buffer.
- Interaction only works when the audience is close to real time.
- Latency is a trade against resilience. Buy only as much as the use needs.
What it means in live production.
Low-latency streaming minimizes the delay between live action and viewer experience. Standard streaming (HLS, DASH) introduces 6-15 seconds of latency because of how segmentation and playback buffers work. Low-latency protocols target sub-3 second latency, sometimes approaching real-time.
The technical trade-off: latency reduction requires different architectural decisions. HLS/DASH use fixed-size segments (10-second chunks) that must be complete before players can start playback. Lower latency requires smaller segments or stream types, but smaller segments increase codec overhead and make quality adaptation less smooth.
How it works on a live production.
SRT and LLHLS (Low-Latency HLS, a newer variant) achieve sub-3 second latency. For esports streaming, where viewers expect to see action nearly real-time and participate in chat about ongoing gameplay, low-latency is worth the technical complexity.
However, low-latency has limits. Physical laws prevent latency below the speed of light (traveling from venue to streaming infrastructure takes milliseconds). Encoding a frame takes milliseconds. Decoding at the viewer's device takes milliseconds. Add up the physics and you're fundamentally limited to 1-2 second absolute minimum latency with perfect optimization.
For remote speaker integration, 3-5 second low-latency is sufficient. speakers feel reasonably connected to audience response. For interactive broadcasts where viewers vote or participate in real-time decisions, lower latency (2-3 seconds) reduces felt delay.
The encoding design for low-latency is different. We use smaller Group of Pictures (GOP) intervals, more frequent key frames, and sometimes accept lower quality to maintain consistent bitrate (avoiding buffer fluctuations that increase latency). For esports, we accept slightly lower video quality at 1080p60 low-latency over perfect quality at 1080p30 standard latency.
At Creative Broadcast Agency, low-latency is a specialized service for esports and interactive broadcasts where latency matters. For corporate events and webinars, standard latency is often acceptable and cheaper to deliver.
Low-latency streaming
We scope latency against the use: auctions, betting and live Q&A need it; a keynote rarely does.
Low-latency streaming
Sub-500ms glass-to-glass streaming via SRT and WebRTC. Real-time interaction for esports, auctions, trading floors, and live Q&A events.
ReadNDI (Network Device Interface)
NDI is a protocol for real-time video transmission over IP networks, designed to be simpler and more flexible than SMPTE 2110 or SDI.
ReadIP-Based Broadcasting
IP-based broadcasting uses standard internet protocols (Ethernet, TCP/IP) to distribute broadcast signals instead of legacy specialised broadcast protocols (SDI, AES).
ReadSMPTE 2110
SMPTE 2110 is the standard for studio broadcasting over IP networks, replacing legacy SDI as the signal distribution method inside broadcast control rooms, OB trucks, and production facilities.
ReadSDI (Serial Digital Interface) explained
SDI has been the broadcast video standard for two decades.
ReadQuestions people ask about this.
How low can we go with latency?
1-3 seconds is practical. Below 1 second requires extremely aggressive settings that hurt quality or viewer experience. We target 2-3 second latency for esports, 5-8 seconds for events, 10-15 seconds for cost-optimized streaming.
Does low-latency increase bandwidth requirements?
Slightly. Smaller segments mean more per-segment overhead. But the bigger impact is that buffering strategy changes. low-latency players use smaller buffers, which means they need faster initial bitrate selection. Overall bandwidth is similar but consistency becomes more critical.
Can we use low-latency streaming for adaptive bitrate streaming?
Yes, but it's harder. Quality switches must happen within the lower latency budget, so the player has less time to assess network conditions before requesting the next segment. This means quality switches are more frequent and sometimes jarring. We usually narrow the bitrate ladder for low-latency (fewer quality options) to reduce adaptation complexity.
Is low-latency worth the complexity for corporate events?
Rarely. Corporate event attendees don't expect real-time interaction with the broadcast. Standard streaming at 8-10 second latency is fine and cheaper to deliver reliably. Low-latency is valuable for esports and interactive events, not for presentations.
Tell us the date, the venue and who is watching.
We scope every live production against its run of show, not a package.
CBA builds SRT streaming services for every show it produces, because transport decides whether the picture holds.