We've got a live streaming setup that has the HDMI video signal going through three (I think) components before going out onto the internet. The audio, meanwhile, comes directly from an analog audio board. Lack of sync appears to be on the order of 1 - 3 seconds. It's amazing how hard it is to tell while watching someone speak, just how far out of sync the audio and video are!

Anyway, can anyone suggest a one or two channel adjustable audio delay product for this purpose?

Thanks.

This piece works awesome

I used on an a theater with an older high end surround system that had no HDMI. Finding the right delay using an AVR for instance is frustratingly difficult I've found, but this piece made it easy.

The only thing is you'll have to dig into specs to see if you can get that long of a delay.

http://www.felston.com/

The Felston has a digital IO and up to 680ms of delay.

I use this Rane delay unit, which has analog IO and up to a second of delay:
[Link: ranecommercial.com]

You can also use nearly any old audio delay that a musician would use in a studio, for example.

I can't imagine what kind of processing would add several seconds of video delay. Can you provide some more details?

Well, first, it's a kluge.*
Second, it's REALLY hard to estimate how far out of sync it is. I'm guessing, but it's the worst I've ever seen.
And it's at a church.
A video system was rapidly planned and put in a year ago. A camera HDMI signal goes through a video switch, then through a 1x6 splitter and out to many locations. Lack of sync was already visible at this point.

Then they added a BlackMagic ATEM Mini that preps and combines audio and video to go to the internet. Its output goes to Facebook, which then sends it out... however they do it.

Due to COVID I have not carefully investigated the connection of that last piece. The audio feed is direct from an analog audio board, so there's no delay there.
As I said, I was guessing about the time, but I don't think 680 msec will do it.


*I will always insist on that spelling. "Kludge" rhymes with fudge, wouldn't you say?

Ernie,

Taking a cue from the film industry, have someone create a clapperboard sync event that you can view and time.

Hall Research has been mentioned several times on RC- IIRC, it's adjustable in 1mS increments.

time/money?

Vaddio makes a device that will take an hdmi input from your system and an audio feed from the mixing board and delivery them to the internet.

It's called the AV Bridge Mini, they're proud of it, my favorite supplier carries it.

There are a couple of times in the most recent event where the speaker claps his hands a couple of times. This still ends up with events very close in time and hard to suss out.

I just figured out how to determine the timing reasonably well. Record and monitor. Use a clapper as suggested, or even just clap hands. Clap repeatedly and adjust the time between claps until you hear the last clap exactly in time with the next clap. Once you've got that, it's easy to look at a time scale and get the timing.

Can you go audio in to your camera and use that audio for your stream instead of splitting the video and audio signal

ggarza,
The camera is some thirty feet of wire, both forward and below the level of the audio board. If I had that wired, I could run a line level signal up to the camera and inject it at that point. The camera should be made so that its audio and video are in sync, so that would improve things by one part, still leaving whatever delays are introduced by the switcher(s) and the DA. And does any delay get introduced by Facebook? I don't know. I doubt it, but the Facebook video is lower resolution than the camera output, so there's a layer of video processing there.

Still lacking info about everything in the chain, but a couple quick thoughts..

Broadcast standards call for a max lipsync delay of 2 frames (66ms). More than that will be noticed.

Assuming the audio goes through the mixer then to the black magic web encoder, there shouldn't be any audio delay.

Delay to the video can be introduced in several places:
- In the camera. Some (cheaper) cameras add 6 frames of delay, or more. Better cameras add a frame or less
- In the video switcher. Cheaper switchers add several frames. Better switchers add none.
- Scalers/converters. These typically add at least a frame or two.
- Frame sync. These add a frame of delay, at a minimum.
- Encoders. These can add lots of delay.

Generally speaking, a DA (or splitter) will not add any delay.

A web encoder, like the black magic encoder, will add quite a bit of delay, however, it is designed assuming that the video and audio coming into the device are in sync, and it should keep them in sync, with equal delay, throughout the process.

[Link: rticorp.com]

Oprah,
that RTI piece has a maximum delay of 340 msecs, so, half of the Felston unit. And I haven't looked at prices to see if that forces a decision.

340 and 680 seem strange values. I wonder where that number comes from. FYI and maybe irrelevant, neither product lists any frequency response specs. Hmmm.

I'd guess the number comes from the RAM installed. If sampling 16 bits at 44.1k the data rate is 1411.2 kbps. So it looks like about 1M of RAM is the limitation. My 3am math says that 680ms would use all of that up.

Your 3 am math was better than my 3 pm math!

Thanks.