Is 0.0036% time inaccuracy on a consumer-grade camcorder expected?

Question

I did an experiment on our church's camcorder (Sony Handycam HDR-CX380) recently to test whether it was the source of a time inaccuracy between our video recordings and PC audio recordings of sermons (see Identifying source of audio drift between camcorder and recording PC).

What I discovered is that video files recorded by the camcorder are about 0.0036% shorter than a stopwatch measurement of the same events. (Example: 3:56:52.81 of stopwatch time led to a file with duration 3:56:52.30.) It's not noticeable when watching the footage on its own, but when I try to match a high-quality audio recording to the video, the bad lip-syncing is noticeable.

Is this kind of inaccuracy in recording times typical for consumer-grade camcorders? Or is our camcorder "broken" even by consumer-grade standards?

Answer 1

IME, 0.0036% is rather better than average. Consumer gear operates on quite inexpensive crystal oscillators. These oscillators or crystals are not tuned precisely (or they would cost too much), and they don't have extraordinary stability over temperature ranges.

Professional gear uses special techniques to keep audio and video gear in synchronization. They use premium-quality sync generators with oscillators that cost more than your whole camera. And even then, they must be re-synchronized several times per day.

And traditional broadcast gear would have a master sync generator, and all the cameras, switcher, etc. would "slave" to that central reference so that everything would be in lock-step.

It sounds like your gear is operating properly. Nothing is "broken". That is simply normal and expected performance.

If you need to sync video to your reference audio track, it is MUCH better to "stretch" or "shrink" the video to fit the audio than vice-versa. Slightly changing the length of the video track with simply insert (or remove) an extra frame every few minutes, which almost nobody will notice. But slightly changing the length of an audio track is likely to create quite noticable and annoying artifacts.

It is simple enough to try it for yourself. Try slightly changing the length of a long video clip and watch the results. And then try the same with an audio track. You may be lucky that you can get away with it. Only you can try the experiment with YOUR content and YOUR editing system, and YOUR expectations.

Answer 2

Keep in mind that the h.264 codec (seems your camera uses it), although versatile, should only be used for re-viewing purposes. It's not a steady-as-a-rock pro-level codec, one used to record sync sound. It can work, and does (like millions of people, I've used it many times with DSLRs), but in the end be ready for sync issues.

More over, your camera is really low-grade. Use a DSLR instead with an external audio recorder.

I would consider using an external audio recorder with your current camera. Sync the video by using a clap-board or even doing a hand-clap right after the recording has started. Test your setup that way and see what you get. If after trying different permutations you still have sync issues, shoot with something else.

Answer 3

If you are talking about a physical stopwatch I would doubt your ability to make timings in the milliseconds domain with it.

Usually the best way to avoid audio drift when using consumer equipment is to record audio and video together, so a simple solution might be to just feed the audio to your camera.

The problem.with audio drift is sometimes complex, if it is constant, it is easy to fix since you can do a slate at the beginning and end and modify the audio clip to fit the videos length using something like audacity.