Six Audio Sync Mistakes We See and How to Avoid Them

Trusting one test and calling the chain fixed

The most common mistake we meet in an audio sync audit is a single clap test at the start of a project. Someone claps in front of the camera, the monitor looks right, and the chain is declared in sync. A clap shows you the state of one path at one moment, and live productions rarely stay in one state.

Routing changes between shows. A different graphics source goes through a frame synchroniser, a remote contributor arrives over a network link, or a monitor wall is switched to a new feed. Each change can add a few frames of video delay that the audio path does not share.

A better habit is to measure at several points: camera output, vision mixer output, and the final feed leaving the building. Record what you find and keep it with the run sheet, so the next person starts from numbers instead of a feeling.

• Test at more than one point in the chain
• Repeat the check after any routing change
• Write measured offsets down, in milliseconds

Ignoring where the clock comes from

Drift is the fault that appears an hour into a show and was not there at rehearsal. Nine times out of ten we trace it to devices running on their own internal clocks instead of a shared reference. Two units that each believe they are running at the correct rate will slowly disagree.

Audio and video equipment often need different kinds of reference, and a house sync generator may feed one set of devices but not the other. Embedders, converters and standalone audio interfaces are the usual gaps, because they sit quietly at the edge of a rack and nobody checks what they are locked to.

Walk the chain device by device and note the clock source on each. If you find one free-running unit, that is rarely a minor detail. Locking it to the shared reference is often the single change that removes the drift altogether.

• List the clock source for every device
• Look hard at converters and embedders
• Lock free-running units to the house reference

Fixing the symptom with a delay box

When lip-sync is out, the quickest repair is to add delay to the early signal until the picture looks right. It works on the day, and it is why so many chains carry mysterious delay values that nobody can explain a year later.

The trouble is that a compensating delay hides the cause. If the real fault is a processing stage that sometimes bypasses and sometimes does not, a fixed delay is right in one state and wrong in the other. The next engineer inherits a patch with no record of why it exists.

Where a delay is the correct answer, because some video processing genuinely cannot be avoided, keep it, but label it. State what it compensates for, who set it, and what the measured offset was before and after. A documented delay is a design decision, while an undocumented one is a trap.

• Find the cause before adding compensation
• Label every delay with its purpose and value

Forgetting that latency adds up across the chain

Each device contributes a small amount of delay, and each amount looks harmless alone. A frame store here, an up-converter there, a multiviewer, a network encoder: individually they are a frame or two, but together they can take a feed well outside what viewers tolerate.

The mistake is judging each device against its datasheet instead of adding the whole path. Datasheet figures are also given under particular settings, and a different scan mode or processing option can change them noticeably.

Draw the signal path on one page and put a delay figure beside every stage, taken from measurement where you can and from documentation where you cannot. Do the same for the audio path. Comparing the two totals tells you where the real mismatch sits before you touch a single control.

• Sum delay for both audio and video paths
• Measure with the actual settings used on air
• Revisit the figures when equipment is replaced

Leaving the people out of the audit

A technical audit that skips the crew misses half of what is going on. Operators know which camera has been temperamental, which input gets patched differently on late shows, and which monitor everybody has learned to ignore. That knowledge rarely appears on a drawing.

Ask the gallery and the floor team what they see go wrong, and when. A fault that only happens on outside inserts, or only after a restart, narrows the search far faster than another round of measurements. Write their observations down next to the numbers.

Crews also need to know what the corrected chain looks like. If the fix is a changed procedure, such as checking a reference lock before going to air, it only holds when the people running the show understand why it matters and where it is written.

• Interview operators before measuring
• Record when and where faults appear
• Share the corrected procedure with the whole crew

Skipping the written report

The last error is treating the audit as finished when the problem stops. Without a written record, the findings live in one engineer's head and the same fault returns when equipment is swapped or staff change.

A useful report is short and plain. It states what was measured and where, what the offsets and clock sources were, what was changed, and what the chain looked like afterwards. It also lists anything left unresolved, so that nobody assumes it was fixed.

Keep the report beside the system documentation and revisit it whenever something in the path changes. A one-page note with real figures will serve a new operator better than a long document nobody opens.