Menu
▾
▴
#6 Thumping sound in output file
When trying to create a sound file from a film I get a very loud beating sound in the resulting audio file. It sounds very similar to an optical soundtrack reader running accross sprocket holes. My thought is that the software is seeing the black blanking area around each frame. The film scan I have is both pillarboxed and letterboxed, that is to say that the soundtrack has black bars across the top and bottom, so the image does not take upthe whole frame. However, when setting up the job I was sure to make the red box appear only where the image of the soundtrack exists. Is there anything I am missing here?
Here is a screencap of the file I'm using, which should illustrate the cropping I was described in my earlier post. Would I have better luck if i tried to crop the frame to reduce the size of the blanking area?
It looks like your scan doesn't have sufficient overscan -- there's no trace of the previous and/or next frame in the scan.
Without sufficient material to calculate the overlap between frames, the overlap calculation will fail, and the failure will sound like low frequency buzz.
Ah, and yes, the letterbox will definitely corrupt the audio signal, as it effectively introduces a wide block of slience at the top and bottom of each "frame" of audio (exactly as sprocket holes would do). The audio track must be continuous.
So two adjustments are needed: continuous audio (no black letterbox area in the soundtrack column) and overscan (scan bits of the previous and next frame with each frame so that the soundtrack snippets extracted from each scan overlap, and the overlap is wide enough for the algorithm to lock it down).
Last edit: L. Scott Johnson 2017-04-05
Sorry for the late reply. I recaptured the film with a larger overlap and cropped the video better and the thumping went. However, now I have an issue where the audio from AEO-Light is significantly slowed down. I've included audio samples, I'm not sure how to deal with this.
Slow sound (usually with a corresponding drop in the pitch) usually indicates that the frame timing is off, either because the fps setting is off or the frame overlap is consistently miscalculated so as to select too many (or too few) samples per frame (or perhaps the output coding misidentitifies the sampling rate in the output).
Are the frame pitch lines set to corresponding positions in consecutive frames within the scan window?