Slashdot Mirror
Lip Sync Problems with New Digital Displays?
An anonymous reader writes "With all of the new digital TV displays flying out the door, its easy to to think that life is good on the road to high definition. But, as Audioholics reports today, cheaper displays are using inexpensive processors that result in video delays of up to 60 milliseconds (that's about 2 frames of video). This means that the video processing (deinterlacing, video scaling, etc) delays the picture so that the audio is out of sync. Add to this inherent delays in some LCD and plasma units and the problem can be more than a little noticeable. As of right now only a few manufacturers are building audio lip-sync delay into their products to compensate."
I was wondering why Ron Jeremy's tongue was trailing behind the licking sound.
Trolling is a art,
Really? Damn. I was beginning to wonder why everything on T.V. was a badly dubbed German show.
Now the lips in my old gozilla movies will be in sync! Deaf people everywhere are rejoycing!
"As of right now only a few manufacturers are building audio lip-sync delay into their products to compensate."
So once again, another company is working around the problem instead of fixing it. This seems to be a bad trend in technology these days.
Work sucked, until it became unemployment, when it became slightly more tolerable. -Tet
"Let's Fight!"
<mouth keeps moving for several seconds>
Oh, wait. Those kung foo movies were always like that.
I only watch 1960's Italian westerns and old Godzilla movies. Who knows, it might actually help.
booo I say
The audio delay should be user configurable. We could turn boring stuff into something that's really funny. Almost as good as playing the old VHS backwards!
The Erogenous Zone
Video games depend on low latency between input (at the gamepad) and output (at the CRT and speakers). Video game systems manufactured for sale in the United States after 2006 will include some sort of digital TV output. These digital TV sets introduce a significant latency into the chain. So what will happen?
Will you really notice if the entire feed is delayed by a fraction of a second?
If the feed is coming from a video game console that's responding to live user input, I'll certainly notice llaagg. A delay of 60ms can spell the difference between a hit and a miss, adversely affecting game scores.
But, there is the delay from the sound traveling from the speaker to your ear (roughly 1 millisecond per foot of distance traveled). So one solution is simply to put the speakers about 60 feet away.
All is Number -Pythagoras.
Audio and Video processing happens asynchronously, so I don't know how you can avoid this. You can set a time limit, but then you will limit the amount of processing that can occur which sacrifices audio or video quality. I have a Panasonic 42" Plasma that does internal scaling. This is slower than doing Dolby Digital decoding. My Anthem AVM20 processor has an audio delay feature where now my audio and video are back in sync. Receivers are getting this feature so eventually it'll be commonplace.
Comment removed based on user account deletion
There's no reason to build expensive circuitry to correct the problem. You can use the laws of nature to resynch your video!
For a video lag of 60 milliseconds, you only need to step back 20.4174 meters from your TV before the speed of sound will correct the synchronization problem.
-3Suns
~~~~
The Revolution will be Slashdotted
I find that if I get ever so slightly drunk, the delay in my mental processing of the auditory information compensates nicely.
...or at least my future father-in-law did. My fiancee's parents recently bought a very nice Samsung HDTV system which had this problem. The audio was way out of synch with the video, and it was quite noticeable at times.
Samsung ended up sending someone to the house, and replacing a board in the TV with a newer model, and that seemed to fix the problem.
I don't understand why they couldn't have anticipated this problem before they shipped the TVs, though. Isn't that what QA is for?
You can accomplish anything you set your mind to. The impossible just takes a little longer.
...you can just view them from sixty feet away and the video delay will exactly compensate for the speed-of-sound delay. No problem.
And if you can afford one, you probably have a living room that big.
"How to Do Nothing," kids activities, back in print!
I frequently see audio delays on HDTV feeds being displayed on my DLP projector. Change the channel (to another station broadcasting at the same resolution), and the problem goes away.
It's either my crappy Scientific Atlanta HDTV receiver or the feed itself.
I really have doubts about this article.
get nemulator
Any bets that they're just trying to build in delays to prevent more 'wardrobe malfunction' fiascos?
:)
My hunch is that they delayed the video on purpose, but forgot to touch the audio.
You can accomplish anything you set your mind to. The impossible just takes a little longer.
The old way was to read in a frame into memory. An other HW block or processor would perform the next operation, by reading that frame, process it and store it into memory again. The whole chain could be quite long.
This was not really a problem, they thought, because the audio was processed at the same time, and the delay was under full control of the soft and hardware.
until someone tried to use an external audio path...
As far as I know, they solved the problem, and the delay is minimal. And non existant if you route the audio over the same processor.
If anyone wants to see the real thing in action, just tune in to the Britney Spears concert that has been airing on ShowtimeHD. Her lips are definately out-of-sync with the music.
I was working on a video conference system a few years ago, where latency is everything. We got a huge $15k plasma display (at the time), only to discover it had something like 90 ms of latency! Since we were already pushing our latency budget by sending signals across the country and loosing frames here and there through video processing equipment and codecs, that 90 ms was more than enough to push us over the edge and make the system very difficult to use for natural conversation, and throw the audio/vidio sync visibly out of whack. The plasma had to be replaced. Three frames may not seem like a lot, but it is quite noticable.
I've delt with a lot of high-priced high-quality plasma systems over the years, and the lesson is definitely "Buyer Beware". The high quality 56" plasma systems can be stunning, but remember that you're also investing thousands in a device with a fairly limited lifetime, and no real industry-wide quality standards and more marketing buzzwords and cheap tricks than you can shake a stick at.
If you using plasmas as a computer display you will see even more artifacts. I've seen widescreen plasmas that could not accept any resolution modes of a correct aspect ratio. Many displays use a great deal of image processing to apply tricks to make the display look good, but sometimes the processing can seriously disturb things like computer text. I've seen apparent color segmentation problems on a lot of displays, and just a lot of artifacts in general.
-braddock
I used to develop code for digital set-top boxes, and I can tell you that this is not a trivial problem.
Because of the way MPEG-2 video works, there is an inherent delay in decoding (frame order in the bitstream isn't necesarily the display order because of the way P-frames and B-frames work.
Audio is slaved to the video through the use of timestamps, but the audio and video frame boundaries don't line up.
I'm not sure if the problem is really lip-sync delay, but building in enough buffering to account for video delay while not glitching audio.
Most people don't notice minor video problems, like repeating or dropping a frame, but they will hear lots of little audio glitches. Also, when a hardware audio decoder runs dry, you usually get a really bad artifact (it sound like stepping on a squealing mouse), and it takes 2 to 4 frames of audio to resync.
(S(SKK)(SKK))(S(SKK)(SKK))
I've seen this problem. I noticed it during the Britney Spears pay-per-view concert. Sometimes it seemed like her lips weren't even moving at all, but she was still singing.
It's good to use your head, but not as a battering ram.
Both original article and slashdot posting are quite inaccurate..
The delay is not caused by cheap processors, that is a myth. Just think about it, even delaying the video by 1 second will not reduce the required processing power...
In fact the delay is a technical neccessity for some of the algorithms employed in modern television. For example motion interpolation for 100Hz TV requires the knowledge of at least one frame in advance.
Also the "delay" in TFTs, as mentioned, has nothing in common with the delay due to video preprocessing....
The only remedy for this problem is to have an option to turn all the preprocessing off for video games and have an artificial audio delay, so it matches the video. Nothing that is out of bounds for an average TV...
If you have a Samsung DLP and you are experiencing this problem, adding a fixed amount of delay does NOT guarantee a fix. The reason is that in Samsung DLP sets, the delay is intermittent. Often times it's barely noticeable (60 ms or less), but sometimes it spikes to as much as half a second. I used to have the Samsung HLN567W but I returned it before my 30 days guarantee was up. Picture Quality was great but the intermittent audio/video sync issue was driving me nuts. I first read about this problem (on Samsung DLPs) last August, I bought my TV last January, and apparently the problem still exists in new sets being sold currently. That makes this problem at least 9 months old... looks like the resolution is not easy otherwise Samsung could have fixed this a long time ago. With all the high tech circuitry being added to consumer electronics nowadays, regular consumers are now becoming beta testers for these consumer products too. Welcome aboard!
Thus today's syncing technology consists of:
The SMPTE LTC code is both recorded on an audio stripe (channel) of a multitrack audio recorder, and on an audio stripe of the film-camera. It is also possible to sync MIDI and DV-timecode to SMPTE time-code.
Now, there is equipment that creates all these signals, or you can build a chain of syncs. Blackburst is often the master, on which the word clock and SMPTE time-code is synced with a PPL (phase locked loop).
I just started in the video and film business and I was stunned by all of this.
There is a more sinister side to this - monitors that can't keep up, get into BIG problems when dealing with another venue that is pushing towards them more and more: VIDEO GAMES.
I'm actually serious. While the normal populace may scoff and deride those who play games like Soul Calibur or Street Fighter until they can actually count how many frames a particular move takes to execute - and how many frames from when the button is pushed to when the move reaches its damage point - everyone likes nice, crisp controls.
They want to know that when they push that button, it went into the system immediately.
Now you're talking about adding a possible 4-5 frame delay to the entire system - but you CAN'T make the video game system have the same delay, it'd have to recalculate everything backwards in time to compensate.
So what do you do there, huh? It's a pretty crappy workaround solution.
Slow processors can significantly delay the generation of the output video. Not only that, but the amount of work the processor has to do, which depends on how many changes from frame to frame take place, will cause varying delays.
The way the problem usually manifests itself is that the delta between video and audio gets biggere and bigger, the two slowly drift apart. The video is, of course, being backed up in memory. At some point it will run out of buffer capacity. The olde way of dealing with this was to just flush the buffer, which brings thing back into sync (for a while), but usually causes a nasty glitch in the video (blank screen for a few frames) in most cases.
Newer techniques involve dropping frames, more of them as the buffer fills up.
A good indicator that you are getting buffer overflow is when you change channel, then change back again and all is back in sync (for a while). This will have flushed the video stream buffer, and life will be good, untill it backs up again.
Faster processors can deal with the overall data rate without having to resort to these extremes, but the inherent delay caused by having to buffer a frame (or more) to be able to decode the next (because we are dealing with frame deltas in MPEG) will still cause varying delays in the video.
The real answer is to use adquate processing power, and to modify MPEG to insert timing marks into the video and audio streams, and allow the system to automatically and incrementally adjust the audio delay to keep it in sync with the video.
Expect to see a squadron of flying pigs before this happens ...
An even better answer, of course, is to scrap this digital TV crap. The best digital TV signal doesn't hold a candle to the best analog TV signal. All that digital buys is the ability to squeeze another 150 shopping channels onto every satellite at the expense of video quality - but that doesn't matter, its marginally better than VHS, so what will the consumers ever know?
Can you imagine how GranTurismo would be in a move-in-turns scheme?
>I don't understand why they couldn't have anticipated this problem before they shipped the TVs, though. Isn't that what QA is for?
Dilbert: We have a serious flaw in our product that can be corrected with a cheap, quick swap-out of the Model 9 Frammish Board with the Model 9A. You want me to recall all 495,000 units that have shipped to our value-added resellers and make the change, right? Remember, I told you about this six months before any of these units shipped.Pointy-Haired Boss: Noooo. Let's wait and see how many retail customers call and complain. We'll send some minimum wage guy out to swap out the card for the one's that somehow manage to call us.
Some mornings it's hardly worth chewing through the restraints to get out of bed.