If I select a small area, it can detect silence correctly, if I extend the area selection (including the small area) and try again, the detection will be ridicolously inaccurate INCLUDING the small area. Any ideas why this is happening? If my settings proved to be working on the small area before, shouldn't the results at the very least be partially correct rather than getting ALL completely wrong? The file is 10 hours long, I attached a recording of this behavior.
Copy link to clipboard
Having looked at your video, no, the settings clearly won't be appropriate for the part further on where the threshold levels alter completely. If you can get the rest of the audio to be limited the way the first section is, then detection will work for the rest of the file. As it is, I'd guess that it's detecting silences early, and cutting off audio. Is that what's happening?