Low-Latency Live Streaming

Sources of Live Latency

Live streaming latency is the time差 between video capture and viewer display. Sources include:

• Capture latency: Camera/device processing time (milliseconds).
• Encoding latency: Time to encode a frame.
• 传输 latency: Data transfer from encoder to server to viewer.
• Segment latency: HLS needs enough data for a complete TS segment.
• Buffer latency: Players buffer segments before playback.

Traditional HLS Latency

Traditional HLS latency is 10-30 seconds. With 6-second segments and 2-segment buffer, segments and buffer alone add 12 seconds.

Low-Latency HLS (LL-HLS)

Apple released LL-HLS in 2019, reducing latency to 2-5 seconds through:

• Partial Segments: Smaller sub-segments for earlier playback.
• Preload Hints: #EXT-X-PRELOAD-HINT for提前 requests.
• Blocking Playlist Reload: Server blocks response until new content available.

Other Low-Latency Solutions

• WebRTC: UDP-based, 500ms-1s latency for small-scale互动.
• RTMP: 1-3s latency, requires Flash (deprecated).
• SRT: 1-2s latency for professional broadcast.