0

I have two video clips of the same music video. Both are made of the same video and audio tracks but of different sources and quality.

One of them is a rip from an officially released media, so the video/audio synchronization in it can be used as reference.

The other was made by muxing video and audio streams from different unrelated sources. Whereas they appear perfectly in sync to human eye and ear resembling the original, I would like to somehow gauge it with sample-perfect precision to see how close the sync to the original actually is.

I can use some video players to nagivate frame-by-frame to the same frame in both clips. But I don't know how to see the accurate position in the audio waveform which that frame corresponds to (to be perfectly clear, the brackets encompassing the audio fragment corresponding to a particular video frame). What tools can I use for this?

Ideally I want the tool to tell me something like "your second clip's audio is muxed 0.0005 seconds behind comparing to the reference clip". I will then use that 0.0005 seconds result to correct the second clip by re-muxing it with the right offset.

  • I don't understand the question. You said audio is perfectly in sync judging by your eyes and ears. Anyway, you cannot get more precise than 0.01 - 0.04 seconds depending on fps of the video. To me it seems you're chasing a ghost problem. – Matt Jul 5 at 13:07
  • @Matt Let's say I'm chasing 0.04 s for pure perfectionism purposes. Eyes/ears won't detect 0.04 s out-of-sync, will they? I want to detect and negate that. – Greendrake Jul 5 at 14:06
  • @Matt Actually, why would audio sync precision depend on fps? While a frame is frozen on the screen for a split of a second, the audio keeps playing continuously. You can move it backwards or forwards with its sampling precision (e.g. 1/48000 s). – Greendrake Jul 5 at 14:10
  • Humans can detect off-sync audio from around 0.045 s or 0.1 s depending on who you ask. You're right, you can align the audio signal as precisely as the audio sampling rate. But you need to decide to what frame to align but you only get a new frame every 0.04s for example. What if you want to sync to a drum stick hit on a hi-hat but the stick is in one frame far from the hi-hat and in the next one already touches it? The hit is somewhere in between those two frames. You can't tell by how much to move the audio signal back or forth. That's what I meant. – Matt Jul 5 at 18:20
  • @Matt The stick hit will be somewhere inside the frame previous to the first one where the stick touches the hit-hat. And yes I can tell how much to move the audio signal because I can see where exactly the hit is in the frame in the reference video. Just need a tool to magnify and measure it with precision, which is what this question is about. – Greendrake Jul 6 at 1:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.