I have some broken video files with an unknown delay between video and audio. In this given case I cannot guess what has been done to the streams in order to determine the delay. While watching I can see that the delay is there but can't quantify it, nor I can ask anybody with a better perception of rhythm and time.
What is the easiest way to acomplish this task?
For example, to determine pixel aspect ratio error some have suggested seeking perfectly round objects such as the Moon. What sections of audio do I need to look for? Can I increase the correction to some point and then step down until...?
Full Version: Experimentally determine audio delay
If it's the more common case of the video lagging the audio, you may be able to make use of the audio section of the ffdshow plug-in with your playback software to introduce a delay on the audio.
You can set a value in millisecond integer values to add audio delay. I have to use it all the time to compensate for the lag on the video resulting from heavy image post-processing by the video section of ffdshow when playing back DVD or MPEG-4 video. They both look hideous on a 60" screen without it.
If you take this approach, see if you can find a section of the offending video with portrait shot monologue or dialogue. It's fairly easy to lip-sync by noting people's lip movement.
Cheers, Slipstreem.
You can set a value in millisecond integer values to add audio delay. I have to use it all the time to compensate for the lag on the video resulting from heavy image post-processing by the video section of ffdshow when playing back DVD or MPEG-4 video. They both look hideous on a 60" screen without it.
If you take this approach, see if you can find a section of the offending video with portrait shot monologue or dialogue. It's fairly easy to lip-sync by noting people's lip movement.
Cheers, Slipstreem.
I will stick to speaking portraits then. But that generally won't work for commercial movies, due to use of looping/ADR.
About the technical side: After posting here I discovered an unutilized part of my sound card, the E-MU Digital Audio System. There is an effect called Stereo Delay, which can be used to add an arbitrary amount of positive delay to the audio, up to 1500ms (mono – 3000ms). Negative delay is possible if the audio is first delayed by a known amount (-500ms, -1000ms) using other means.
The advantages: delay is applied instantly, with insignificant CPU usage, the picture does not jerk at the moment of adjustment. In DirectShow you also have to open numerous dialogs to get to the setting, then press Apply, and Close to access the seekbar. Repeat. In the meantime you have missed a critical moment where a person speaks.
About the technical side: After posting here I discovered an unutilized part of my sound card, the E-MU Digital Audio System. There is an effect called Stereo Delay, which can be used to add an arbitrary amount of positive delay to the audio, up to 1500ms (mono – 3000ms). Negative delay is possible if the audio is first delayed by a known amount (-500ms, -1000ms) using other means.
The advantages: delay is applied instantly, with insignificant CPU usage, the picture does not jerk at the moment of adjustment. In DirectShow you also have to open numerous dialogs to get to the setting, then press Apply, and Close to access the seekbar. Repeat. In the meantime you have missed a critical moment where a person speaks.
The built-in Stereo Delay feature sounds very handy. I was forgetting that not everyone is running dual-screen setups. I had the movie on my projector and ffdshow open on my desktop monitor simultaneously while setting up my delay value which made it easy.
Good luck.
Cheers, Slipstreem.
Good luck.
Cheers, Slipstreem.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.