Whenever I try to synchronize separate-sourced audio with a video file, I find it very frustrating not to be able to make finer adjustments. The interface appears to force the clip to begin on a frame boundary.

It seems to me that there is no intrinsic reason for this, though, since the sound doesn't have frames.

Is there some trick for making finer adjustments?

Q: Is there some trick for making finer adjustments?
A. No, not yet.

Granjow's audio sync branch has been sync'd with git I think, to make the task of syncing external device audio to camera audio. Not sure if it's exposed in kdenlives interface yet.

http://quickgit.kde.org/?p=kdenlive.git&a=commit&h=f41ee253ade9dfda45752f232ea0ff35e26dce95

**EDIT**

Disabled by default in config:

http://quickgit.kde.org/?p=kdenlive.git&a=commit&h=98d539440441d18d24d9bcd78d755c050865091c

I've only been able to do this by adjusting the audio in a program that lets you get down to the sample level, then overdubbing the video with the new audio. It's very tedious, but if you are careful, you can do it this way.

Think about it: If you have a video frame rate of 25 fps (PAL standard) and an audio sampling rate of 48kHz (DVD standard), you will have 1920 audio samples in one video frame. If you cut out even a few audio samples (much less 1920 of them) from an audio clip, it is certainly going to be much more noticeable to the ear than it would be to the eye if you deleted 1 video frame.

If you are only trying to synchronize lip movements with singing, though, sometimes you can go about it the other way around (i.e. snip off a video frame or two here and there, perhaps add them back somewhere else, then overdub the audio again). Of course, if the whole video is earlier or later than the audio by a certain amount, you might be able to set an appropriate audio delay factor globally when overdubbing. I've seen this option in Kino, but haven't delved deeply enough into kdenlive to guide you straight to it (if it even exists).