I don't want to belabour the point I have tried to make in other posts but lip-sync error to some degree is present in essentially "all" broadcasts. According to research at Stanford University in the US over 20 years ago most people do not consciously notice 42 ms of lip-sync error.
For some of us the threshold at which we notice lip-sync error consciously might be 80 or even 100 ms. We don't see it (consciously) until something pushes it beyond our own personal threshold. Naturally we blame whatever does that. Whatever app or programme or piece of equipment that introduced the last increment of lip-sync error that went beyond what we could ignore. I've seen it called our "threshold of recognition" which varies tremenously from one person to another.
But the device that gets the blame "may have" added far less lip-sync error than was already in the signals previously added by others. That is to say a device or app that adds the last 20 ms and brings the problem into your conscious gets the blame but if the signal had not already had 40 ms error caused by others you may never have noticed it. And others with a threshold above 60 ms (in this example) may still not notice it.
This problem is exacerbated by the fact our threshold of recognition drops greatly once we notice the problem. The theory is that we "were" subconsciously looking away from the characters' lip movement to avoid the impossibility lip-sync error presents. It can't happen in the "real world" so our brains can't process sound before the event that creates the sound.
But once it exceeds our ability to look away and ignore and enters our conscious we look at the lips - we look "for" it - and can see tiny lip-sync error we never consciously noticed before. Even as little as 1 ms for some of us!
That theory - that we look away from the faces to avoid the impossible - helps explain the discovery at Stanford that even lip-sync error that we don't consciously notice causes negative feelings about the characters similar to those we have about people who don't make eye contact with us. But aparently it is the viewer who is not making eye contact with the characters when lip-sync is off.
If there is a silver lining to this aggravating problem at least once we notice lip-sync error we can take steps to correct it and eliminate the negative impact it has on our perception oif the characters documented in the Stanford report.
I've suggested this before but if you are interested in the Stanford research report Google "Reeves and Voelker Audio Asynchrony" and I am sure you will find it as it is often quoted.
Their 40 ms threshold that most people don't notice has influenced the video standards committees but what they failed to address 23 years ago and is still true today is that there is nothing in the video and audio signals to define when they were ever in-sync. The insustry has treated synchronizing audio and video with open loop controls: At each stage from content creation through broacsasting they are expected to measure the video delay each digital video effect (DVE) causes and add an equal audio delay to offset it.
In a perfect world that could work. If every video delay were measured perfectly and every audio delay added synchronization would be maintained. But it only takes "one" to drop the ball and there are hundreds of DVE's along the path and any error introduced anywhere cannot be automnatically detected downstream so errors are cumulative.
This is why you will see lip-sync error on one programme and perhspa not the following one.
And if that programme you are watching has a large error it may be the real culprint causing you to notice lip-sync error when some device or app with a smaller contribution to the error added enough to push it beyond your threshold.
I work in this area so forgive me for getting on my soap box but as I try to contribute on these forums I see so many devices getting blamed for causing the problem when in fact if there were less lip-sync error in the signals their slight addition to the error probably would not be noticed by many.
There is some light at the end of the tunnel as SMPTE has approved a standard that would add "signatures" of each video and accompanying audio frame to metadata to be passed with the signals. If this catches on it should eventually mean there will be signals that actually define when the content producer felt his ceation was in-sync and allow correcting lip-sync at any point in the broadcast chain no matter where the error originated.
I hope it will be accepted and all the eqipment required to utilize it will be purchased but my guess is that we're taling 10 yto 15 years or more if at all. The problem is "chicken or the egg" in that currently no programmes have the metadata signatures implanted so expensive equipment to read and realign audio with video would accomplish nothing. And content producers see no advantage to purchasing equipment to add signatures for which no broadcaster has equipment to utilize. And even worse - to my knowledge - no broadcast equipment manufacturer is producing and promoting products for broadcasters to utilize.
This same situation occurred almost 20 years ago. Tektronix had developed their AVDC100 which would watermark audio and video signals and use those watermarks to realign lipsync downstream. It was a fantastic product in my opinion but the induustry did not accept it and tektronix finally gave uo and exited the market. they cost about $8000 for each end which is not much to a TV station so that wasn't the problem. It was the same problem the new signature scheme has: It's worthles until "everybody" buys in to the scheme. and invests in the equipment.
But at least it is "possible" that someday the signals will be there to correct lip-sync automatically.
I think the hype around the HDMI "auto lip-sync correction" feature announced with HDMI 1.3 has caused a lot of confusion as most consumers seem to think that means it can somehow automatically correct lip-sync. Of coures it can't - since there are presently no signals in the video or audio upon which to base such automatic correction - but it makes consumers think it is possible and if it were possible then they naturally blame manufactures of equipment they think could alos maintain lip-sync automatically.
I don't envy manufacturers and broadcasters like Sky who constantly catch the blame for something they can't possibly correct. ALl they can hope to do is reduce their contribution to the problem and hope that's enough to keep it below most viewers threshold.
Incidentally all HDMI's "auto lip-sync cortrection" feature which has caused so much confusion can do is declare via EDID session with each source (like plug and play for computers) two latency values for audio and video. One set for interlaced and one for progressive video. They will be about 20 ms apart.
They are fixed values for the video delay of the TV and have nothing to do with any lip-sync error in the arriving signals. When this was announced (HDMI 1.3) most AV receivers aready offered fixed audio delay settings to compensate for a TV's video delay which did the same thing but consumers assumed it could dynamically correct for changing lip-sync which still isn't possible.
It's a worthless feature which can be confirmed by the number of TV manufacturers who actually support it. Basically NONE. Most major TV manufacturers don't put those latency values in the EDID data. And those who do (like some Sony and Toshiba TV's I've seen) make it worthless by adding audio delay intrenally the user can't turn off. , etc. I had a Toshiba that sent the data but it applied the delay (100 or 116 ms) onit's audio as well as it's s/pdif output so the user has no option to turn it off and let another device like an AV receiver or external lip-sync correction product provide the delay. That deprives the user of the TV's video delay which in cojunction with and audio delay allows correcting for audio that arrives delayed or is delayed by external speakers like sonos, etc.
