Closed haudiobe closed 2 years ago
Check if this test is already documented
The AVC content options (https://1drv.ms/w/s!AiNJEPgowJnWgbpZesbLvglzCXVlSg?e=4ZFRyB) say the following.
CMAF Fragment durations o 2 seconds o 5 seconds
Is this what you mean?
Again might need to consider minimum buffer level required on specific platforms. If a CMAF chunk corresponds to a single frame multiple chunks might be required for playback to start. Consequently, playback might stall for different platforms at different buffer levels.
Suppose an implementation requires a minimum of (say) 5s buffered and will stall when the buffer size drops below that. That's clearly not a low latency enabled implementation. Is it surprising if it would fail such a test? The test could be made configurable so that particular content providers could require the test to be passed with at least a certain value in order to have access to their content.
Suppose an implementation requires a minimum of (say) 5s buffered and will stall when the buffer size drops below that. That's clearly not a low latency enabled implementation. Is it surprising if it would fail such a test? The test could be made configurable so that particular content providers could require the test to be passed with at least a certain value in order to have access to their content.
I like that approach
part of new test cases.
Background: In low-latency, typically playback is done based on only a short buffer duration.
CMAF Content: Create CMAF Chunked content with chunk durations in the range of Low-Latency playback
Stimulus: Fill chunk by chunk on a frequent basis
Expected Playout Behaviour: It works