So I recorded voice commentary while playing a video game, and because of a bad mic set up, the video game’s audio also picked up in the voice recording. This has caused me to notice that the voice commentary doesn’t seem to be playing back at the exact speed of the original recording.
That is to say, if I sync based on the background/game audio at the start, then by the end (about an hour difference) it’s off by about half a second. I was able to correct this just by stretching it out, but obviously I’d like to avoid doing this in the future.
Any clue why this could be happening? My computer just lagging behind on it somehow? Or some setting I accidentally messed up?
The grownups in the movie industry go to extraordinary effort and expense to make sure that when they shoot sound separate from the picture, the quality of the timers in the camera and the sound recorders will sync within 1/24 of a second I think I once calculated for several days if needed, with no cables between them.
You, on the other hand are using lowest possible cost consumer grade timers in the computer and I’m guessing separate timers in the sound card running the microphone. If you used a USB microphone, that, too, has internal timers with their own quality problems.
Separate sound sync is a very special problem and in some cases where you don’t have multiple thousands of dollars to buy hardware, people make recording equipment that you can run a special sync cable between so they’re all preaching from the same timer hymnal. Those shots should come out exact.
Past that, you take what you can get which is drifting timebases. It’s worse, of course, if they go in different directions. The game drifts up and the soundcard goes down.
There is one fixed error we used to run into a lot more than now. US ColorTelevision picture frame rate was 29.97097… and it never came out even. It was intentional. It made the color look better. It’s not 30, so if you recorded sound at 30 and tried to marry it with a 29.97 television picture, you were reliably going to be off about a frame (1/30 sec.) per minute. It was such a common problem that we used to have easy, single-click, standard tools to correct it. But that would give you a much higher error than you have.
I think the digital sample rate error would give you a much worse problem, too. Audio CD sample rate is 44100 but digital television uses 48000. I think you can actually hear sound tone pitch changes if you cross those two.
So I think what you have is normal drift between pieces of consumer electronics.
By the way, this is exactly the same problem people have marrying two USB microphones. Some computers have the ability to record two separate USB microphones at the same time. There is a modest little setup panel which quietly asks which microphone would you like to use as master sync – since it can’t use both. The odd man out just slowly drifts away – out of sync.
That’s super interesting and useful. I did notice that my commentary was 44100Hz, while the audio extracted form the video was 48000Hz. Would changing my commentary audio to 48000Hz make a difference at all?
That’s super interesting and useful. I did notice that my commentary was 44100Hz, while the audio extracted form the video was 48000Hz. Would changing my commentary audio to 48000Hz make a difference at all?
No. That’s not the problem… The software takes care of that automatically, and computers are really good at math! However it is “good practice” to keep everything at the same rate (if possible).
The difference between 44100 and 48000 is about 9%, which would be about 7 minutes over an hour.
Your difference of 1/2 second over an hour is actually quite good for a setup without a master clock.