Hi there, I have a wav track extracted from an old movie which now I wished to fix
The main problem is that I have many “clicks” followed my a very small “gap”, around 30ms. Given that this gap literally sounds like a “silence” I tried to use the “Silence Finder” tool but failing because of those many false positive.
I am a totally newbie about sound editing and sound generally so I have no clue what values choose for this tool.
E.g. the “treat …as silence”: I thought that any sample closed to the “zero” line have a very high “-dB” so, lloking my wav track, my first option was to choose values around “-60dB” but instead it does work only if I choose values around -15dB…
later, if thre is a way to mark all these gaps, I guess that deleting them will result a in a too much short track putting my video out of sync. What could be the suggestion then?
The vinyl recording people have the same problem. The “silence” between the cuts isn’t silent. It still has vinyl noise and cat hair pops.
Pay attention to the green sound meters instead of the blue waves. You might consider making them a lot bigger so you can see what’s going on. Grab the right-hand edge of the meters and pull to the right.
That’s what you use to set the silent-sense, not the blue waves.
I’d be trying to fix the original clip.
Does the original work sound like that if you just play it and don’t try to capture and edit with it? That’s a lot of damage for a posted movie clip and if it’s repetitive, then something in the capture system may be broken.
I am probably doing wrong, but I have attached the spectral form of the selection of “Gap1”.
Ok, it’s not “silence”, but I thought that whatever under a threshold “-dB”, let’s say -60dB here , for N milliseconds, could be considered as silence.
by the way, the movie it’s not mine, got it from the Internet and I guess it’s a show acquired from a VHS recorded from a TV.
As you’ve found out already finding then and marking them is a bit of a problem - the Silence Finder and its counterpart Sound Finder aren’t always too accurate (I think Steve may be turning his attention back to this once he gets his head out of his current Fader project).
I think you are right to be worried about shortening the show by deleting them all (there is a tool in Audacity called Effect>Truncate Silence but I wouldn’t use that for this task as you do want to retain synch).
What you will need to do (after you have marked them) is to find a similar piece of audio from nearby and paste it in - actually you probably want to paste t into the right place in a new track under the existing track and then you can manage the cross-fades between the original and the repair. You will need to paste a bit that is slightly longer than the “silence” at both ends to work the fades. This technique is a bit like the “Stamp” tool in PhotoShop if you know that.
As you can probably guess this is pretty labour intensive and is a labour-of-love (I know, I used to do similar for vinyl click removal before more automated tools arrived on the scene).
The version that I was thinking of is rather more complex than you need, so here is a modified version of Silence Finder for detecting short gaps (optimised for 44.1 kHz tracks). When installed it will appear in the Analyze menu as “Finder Short Gaps…” gap-detect.ny (4.28 KB)
Fortunately there is already a more precise version of Silence Finder on this forum. I’ll have a look for it. If I can’t find it then it’s an easy modification to make Silence Finder more accurate.
Fortunately there is also a plug-in that can help to speed this up. Again I will need to search for it…
what to say… thank you they work specially ez-patch which would have been my next question
I will try to modify “find-gaps” to see if does make sense a “search for at least N milliseconds but not over Y milliseconds” these could filter out some false positive
I have a qeustion though. Again, I am a totally about sound editing, but what’s wrong with my idea of finding the threshold?
I see that even with “find gaps” the filter finds this gap if I choose a value of dB around -27dB, while I would expect that lower, around -60, would be correct.
How do I find this threshold then? I would understand the theory, of course I can manage just trying