The Raspberry Pi makes for a crazy-cheap media center running XBMC, but it’s not all that powerful and occasionally struggles with HD video. Particularly when combining HD MPG video with multi-channel audio, you may find that occasionally people’s lips (video) don’t match their voice (audio) quite right.
The solution is pretty simple: XBMC has a feature to tweak the audio offset, allowing you to better synch up the audio and video. When watching a video, click the menu button and navigate to the audio settings:
And finally, use the left and right arrows to move the audio slightly in respect to the video until everything matches up. I’ve found with my setup most out-of-sync videos require the same treatment – backing up the audio by .25 seconds to match the video: