Questions Regarding hlscmafsink Latency Optimization

Hello GStreamer Community,

I’m exploring hlscmafsink for my HLS live streaming setup, aiming for ultra-low latency. I have a few questions:

  1. How does hlscmafsink help reduce startup latency compared to hlssink? Are there specific features that optimize for faster playback start times?

  2. Since CMAF packets can be sent during segment creation, how can we configure hlscmafsink to start streaming CMAF fragments immediately as they are generated? Are there specific parameters or pipeline setups required to enable this?

  3. What is the minimal startup latency achievable using hlscmafsink, assuming segments and playlists are configured optimally (e.g., 1-second segment duration, low buffer size)?

Any insights or best practices for using hlscmafsink effectively for low-latency streaming would be greatly appreciated.

Thank you!

as I understand hlscmafsink produce normal HLS manifest (not LL-HLS) with CMAF fmp4 fragments instead of MPEG-TS (.ts)

for LL-HLS you might need to generate the manifest yourself using cmafmux which support chunked fragments.

There is an example of generate HLS manifest using cmafmux might help. link