When the Picture Arrived Before the Sound: Marcus and the Late Gallery Feed
A short account of one regional studio where presenters seemed slightly out of step with their own voices, and how a methodical review of the audio chain found the cause.
6 minutes to read
A call about faces that felt wrong
Marcus runs technical operations at a small regional studio, and his first message to us was brief. Viewers had not complained outright, but two presenters had begun saying that the evening bulletin looked slightly off when they watched it back at home. Nobody could point to a single moment. Lip movement and voice simply did not feel married together, especially on close shots of the anchor.
I took the call on a quiet afternoon in mid autumn, and I asked him the question I always ask first: has anything changed recently? He mentioned a new embedder installed in the summer, a replacement monitor wall in the gallery, and a firmware update on the main mixer. Three small changes, each reasonable alone, and none of them obviously guilty.
What struck me was how calm he was. Marcus did not want a quick fix. He wanted to know where the offset was coming from, so that it would not return the next time a piece of equipment was swapped.
Agreeing what the review would and would not cover
Before visiting, we agreed the scope in writing. The review would follow one programme path from camera to transmission feed, measuring timing at each stage where audio and video are handled separately. It would not include a redesign of the gallery or any change to the playout system. Keeping the boundary clear mattered, because timing problems tempt everyone to start adjusting things at once.
Priya Nair, our audio and sync engineer, joined me for the visit. She prepared a short list of what we needed from Marcus's team: a current signal flow diagram, access to the mixer settings, a quiet hour on air-off time, and someone who knew where each patch bay led.
Marcus supplied the diagram on the morning we arrived. It was a hand-marked printout, with two corrections in pen. That told us the drawing and the real wiring had drifted apart, which is common and not a criticism.
- A signal flow diagram, even a rough one
- Mixer and embedder settings exported or photographed
- One quiet hour with no live programme
- A named person who knows the patching
Following the signal stage by stage
Priya began with a test signal that carries a visible flash and an audible click at the same instant. We fed it in at the camera input and read it back at several points: after the vision mixer, after the embedder, at the monitor wall, and at the final output to the transmission link. At each point we compared when the flash and the click appeared.
The first three readings were clean. Picture and sound sat within a tolerance that no viewer would notice. The monitor wall, which Marcus's team had been watching all summer, showed a small delay of its own, which explained why the gallery had never spotted a problem: their reference screens were deliberately slowed by their own processing.
The difference appeared at the last stage. After the embedder, sound was arriving earlier than picture by a clear margin. The monitors were hiding it, and the output was exposing it.
What the embedder was doing
The new embedder had a delay setting that compensates for video processing inside the unit. It had been installed with the factory default, which assumed a processing path the studio did not use. The video side was being held back, while the audio went through untouched, so every programme left the building with sound slightly ahead of the face that spoke it.
This is a typical sort of fault, because nothing is broken. Every device works as designed. The error lives in the relationship between settings, and it only shows when someone measures the whole chain rather than each box alone. The firmware update on the mixer turned out to be a red herring, and we were glad not to have blamed it on the first day.
Priya adjusted the audio delay to match the measured video latency, then we repeated the flash-and-click test at the output until the two events landed together.
Writing it down so it stays fixed
A correct setting that lives only in one engineer's memory is a setting waiting to be lost. We wrote up the review as a short report: the measured offsets at each stage, the setting that was changed and its new value, and a one-page procedure for repeating the test whenever equipment is added or replaced.
We also recommended a habit rather than a purchase. Marcus's team now runs the test signal after any change to the chain, and notes the readings in the engineering log. It takes about a quarter of an hour. It would have caught this fault in the summer, on the day the embedder went in.
Marcus told me later that one of the presenters watched the next bulletin back and said nothing at all, which he took as the best possible review. Nobody had been able to say what was wrong before, and now nobody noticed anything.
- Measured offset at every stage, recorded
- The changed setting and its new value
- A repeatable test for future equipment changes
What the case shows about timing reviews
The lesson I take from Marcus's studio is that a gallery monitor is not a reference unless you know its own delay. Many teams trust what they see on the wall, and the wall is often the one device in the building with a reason to be late.
A review like this is slow work, and most of it is waiting for a test signal to travel through equipment that is behaving perfectly. The result is rarely dramatic. It is a number on a settings page, changed by a few frames, and a studio whose output finally matches what the people inside it believe they are sending.
Marcus's case ended without any new equipment and without a rebuild. One embedder setting, measured and corrected, put sound back where the picture was. If your own presenters seem slightly out of step with their voices, start by testing the full chain at the output, not at the gallery wall, and write down what you find. A small offset, once measured, is usually a straightforward thing to put right.
Ask for a quote
- Phone
- 01223 480162
- Write to us
- [email protected]