In this guide
List the four inputsCheck duration independently from positionRetain the inheritance and media evidenceA segment timestamp that looks enormous may simply be expressed in sample-time units. DASH can also offset that timeline relative to the start of a Period. Comparing raw integers with seconds is therefore a common source of false timing diagnoses.
Use the complete relevant addressing context and write each quantity with its unit. The example below is original explanatory arithmetic, not a complete playable MPD.
List the four inputs
Suppose Period start is 60 seconds, timescale is 1000 units per second, presentationTimeOffset is 5000 units and the sample position is 9000 units. The sample is 4000 units after the Period’s mapped sample origin. Dividing by 1000 gives four seconds; adding the Period start gives an MPD position of 64 seconds.
The order matters because Period start is already in presentation seconds, while the two sample values share the timescale. Adding 5000 directly to 60 or dividing the Period start again would mix quantities that do not have the same unit.
Check duration independently from position
A segment duration of 2000 units at this timescale is two seconds. That does not imply its start is two seconds or that the player is two seconds behind live. Position, duration and live availability are different questions.
If another representation uses 90000 units per second and a duration of 180000, it also describes two seconds. Comparing raw duration integers across those representations would incorrectly suggest a ninetyfold difference.
Retain the inheritance and media evidence
DASH metadata can apply at more than one hierarchy level. Preserve the effective SegmentTemplate/SegmentBase context, not just a copied S line. If the manifest arithmetic is coherent but playback fails, compare the actual fragment timestamps, resource availability and decoder evidence through appropriate source-side tools.
A useful report includes each input with its unit, the resulting presentation position, selected representation and observed failure. It does not fix a publishing defect by editing offsets blindly. State whether your conclusion concerns manifest arithmetic only or has been checked against actual media; those are different verification scopes.
| Calculation | Result |
|---|---|
| 9000 − 5000 | 4000 sample units |
| 4000 / 1000 | 4 seconds into Period |
| 60 + 4 | 64 seconds on MPD timeline |
| 2000 / 1000 | 2-second duration |
| 180000 / 90000 | Also a 2-second duration |