Why Does the Picture Look Right but the Sound Arrive Late?

Of all the questions that reach us about Audio Sync and Mixing Support, one comes up more than any other: the picture looks fine, the sound is clean, yet the presenter's lips and voice do not quite agree. Producers often describe it as a feeling rather than a measurement, a slight sense that something is off, and viewers at home notice it long before the crew in the gallery does.

The short answer is that video and audio travel through different equipment, and each device in the path takes its own time to do its job. Vision passes through cameras, switchers, converters and graphics. Sound passes through microphones, a desk, embedded streams and perhaps a delay unit. Nothing is broken. The two simply arrive at the end of the chain having taken different journeys.

This note sets out how we explain that to crews, how we go about finding the gap, and what we suggest you correct first. It is written for these autumn weeks, when many teams are settling back into regular studio schedules and noticing small faults that went unremarked over the summer.

Video processing is usually the slower side. Every frame synchroniser, scaler, standards converter and screen graphics engine holds the picture for at least part of a frame, and sometimes several. A chain with four or five such devices can quietly add a noticeable lag, while the audio side, being lighter, runs ahead of it.

Digital audio has its own delays, though they tend to be smaller. Sample rate conversion, digital consoles with large processing paths, wireless microphone systems and embedding or de-embedding stages all take a little time. These add up, and they change whenever someone patches in a different box or swaps a mode on a unit.

The important point is that the error is rarely a single large fault. It is more often a handful of small, reasonable delays that were each harmless alone. That is why switching one thing off at random seldom cures it, and why a methodical walk through the chain works better than guesswork.

• Frame synchronisers and converters hold the picture for part of a frame or more
• Digital desks and wireless links add small audio delays of their own
• Changing a patch or a unit mode can shift the balance without warning

Our first step is always to measure rather than trust our ears. We use a test signal with a clear visual event, such as a flash or a clapper, paired with a sharp audio tone, and send it from the point where the programme is made. We then look at the output end of the chain on a monitor and a scope and compare when each arrives.

A clapperboard on a camera also works well when no generator is to hand. Record a clap in front of the lens, then step through the recorded footage frame by frame and see whether the sound lands on the frame where the hands meet. It is a simple, honest check that any crew member can repeat without special equipment.

We repeat the measurement at more than one point: after the vision mixer, after the encoder, and at the final delivery feed. If the gap is small at the first point and large at the last, we know the trouble was added somewhere in between, and we can narrow the search from there instead of rebuilding the whole signal path.

• Use a flash-and-tone test signal from the point of origin
• Measure at several points along the chain, not only at the end
• A clapper recorded on camera is a fair check with no extra gear

Once the gap is known, the usual remedy is to delay the audio to match the slower picture. Most desks and many embedders offer a fixed delay in milliseconds, and setting it to the measured value is typically quick. We prefer to apply the delay at one agreed place in the chain and write it down, so that nobody adds a second correction downstream.

We advise against chasing very small differences by ear. A lag of one frame or less is hard to see in normal viewing, and constant fiddling often creates new errors. Agree a tolerance with your producer, keep to it, and revisit only when equipment changes.

Leave the microphone and mixing balance alone while you do this. Level, tone and sync are separate jobs, and mixing them together makes it hard to tell which adjustment helped. Settle the timing first, confirm it with a fresh test, and only then return to the question of how the programme should sound.

Sync that is correct on Monday can be wrong by Thursday if someone has changed a route or borrowed a unit. The most reliable habit we know is a short written line-up: a few minutes before each programme, send the test signal, check the result, and tick it off. It costs very little and catches most faults before air.

Record the delay values you settle on, along with the devices that cause them. When a converter is replaced or a new graphics system arrives, you then know which number to revisit rather than starting again from nothing. Tom, our technical writer, usually turns these into a one-page checklist pinned beside the desk.

For live programmes the same care applies to remote contributors, whose feeds arrive with delays you do not control. Agree a way to measure those links at the start of a session, and allow the mixer a means to adjust each source separately so that one slow guest does not drag the whole programme out of step.

• Run a short test line-up before every programme
• Write down each delay value and the device behind it
• Treat remote contributors as separate sources with their own timing