The call that started it
He had already tried the obvious fixes. He had replaced two cables, rebooted the vision mixer, and nudged the audio fader delay by eye until it looked about right in the edit. Each fix held for a day and then the problem came back, which told me the cause was not a single faulty item but something in how the chain was built.
Measuring before touching anything
My first request was that he change nothing for a week. Adjusting delays by eye is how most sync problems become permanent, because every correction hides the one before it. Instead I asked him to record a simple clap test at the start of each session: a presenter or crew member claps once, on camera, close to the microphone, and the take runs for a minute.
When Marcus sent the files, we stepped through them frame by frame. The sound of the clap landed consistently later than the picture of the hands meeting, by roughly twelve frames on one day and fourteen on another. A gap that varies slightly from day to day points to a processing stage with its own buffer, not to cable length, which adds almost nothing.
- Record a clap test at the start of every session
- Do not adjust any delay until you have a measurement
- Compare several days before drawing a conclusion
Walking the signal path
Thomas Whitcombe, our audio and sync engineer, joined for a visit to the studio. We drew the whole chain on a sheet of paper, from the presenter's radio microphone to the recorder, and wrote beside each box whether it handled audio, video or both. It sounds basic, yet Marcus had never seen his own gallery laid out like that.
The picture path ran through a frame synchroniser and an upscaler before reaching the mixer, and each of those holds frames for a short while. The audio path, by contrast, went from the radio receiver straight to a small desk and out. The video was being slowed down by several stages while the sound travelled freely, so the audio arrived early at the recorder, and an edit-time delay had been papering over it.
The fix, and why it was not a fader
The cleanest answer was to delay the audio at one fixed point, as late in the chain as possible, rather than at the desk where the operator could change it by accident. Marcus's desk offered a per-output delay, and we set a single value there, matched to the measured picture latency, then confirmed it with a fresh clap test and a spoken test sentence.
We also removed one needless processing stage. The upscaler had been left in the path from an earlier project and served no purpose for his current output. Taking it out cut the video delay by a few frames and made the remaining correction smaller and easier to trust. Fewer boxes between camera and recorder is nearly always the better first move.
Finally, we wrote the delay value and the date on a label fixed to the desk, with a note of what it was measured against, so the next engineer would know it was deliberate.
- Delay audio at one fixed, documented point
- Remove processing stages the programme does not need
- Verify with a fresh clap test after every change
What Marcus does differently now
Marcus now keeps the signal path drawing pinned beside the gallery door and updates it whenever equipment is added or removed. Any new box in the chain triggers a clap test before the next recording, which takes about two minutes. When a guest asked whether the picture and sound could ever drift again, he could point to a routine rather than offer a hope.
He told me the most useful change was cultural, not technical. The crew used to argue about whether sync looked right; now they look at the numbers from the last test. Disagreements about taste are hard to settle, but a measured offset of two frames is simply two frames, and everyone can see it on the paused clip.
A note on checking your own chain
You do not need a visit from us to begin. Pick one ordinary session, record a clap test, and step through the file frame by frame in any editor. If the sound and picture agree within a frame, leave well alone.
If they do not, draw your path, count the stages that hold video, and look for the place where a single, documented correction belongs. The measurement comes first every time.
Marcus's half-second drift was never a mystery, only a chain nobody had measured end to end. Once the clap tests gave us numbers, the cause showed itself and the cure was small: one removed stage and one labelled delay. If your presenter seems to speak out of step, resist the urge to nudge a fader. Measure, draw the path, correct once, and write down what you did.
Further reading
-
Checking Your Studio Cameras Before the Dark Months Begin
Shorter days bring heavier schedules and more live evening work. Thomas Whitcombe sets out how to review lens, tally and camera control settings in these weeks so pictures stay matched.
-
The Tools We Use to Review a Working Gallery
A look at the checklists, signal tools and monitoring methods we rely on when reviewing control room workflows, and why we keep the kit list short and the method clear.