Detect Silence: how to achieve the opposite?

I’d like to strip away the sound of a dialog track, leaving only the “silence” or room tone, move them all next to each other and finally crossfade to use for filling in gaps in a scene. Is there a quick way of doing this or a macro I could use?

Interesting thought…

I’m betting you can’t, but you might be able to figure out a macro.

I found this from this link:

Make a copy of the track so that you have two adjacent tracks: original, copy.

Apply detect silence to copy – it will be chopped up with only non-silence remaining.

Select all of these non-silence events and run Silence on them from Audio>Process menu. They are now pure digital silence.

Constrain-position-drag them on top of the original, solo & set locators, and bounce down – now you have “silence” (room), punctuated by pure digital silence.

Run detect silence again. You will have to adjust the threshholds to discriminate between “silence” and pure digital silence.

What remains should be the “silence” (room).

But Nunedo, being the top of the line post software must have a better way of doing this…right? Like in this video:

How about this:

  1. Make a copy of the track, which I presume has dialog clips separated by silences.
  2. Detect silence on the copy, setting the threshold high enough that it drops roomtone but keeps dialog
  3. Invert the copy and mix the two together, phasing out the dialog sections
  4. Detect silence on the newly mixed result.
  5. Run a recursive macro that crawls through the mixed track, finding the next segment, selecting it, locating it to the end of the previous segment, repeat.
  6. Crossfade the result and bounce to a new piece of audio.

Other than setting the threshold in step 2, and writing the macro in the first place, it should go pretty fast.

But if you’re just looking for slightly longer roomtone to fill, why not take just lift one or two pauses and then do a c-loop?

Here is how. It seems longwinded, but is fast when you get the hang of it. It’s harder to write than to do!

First Create two new tracks, named EDIT and MAKE FILL (what they are named is irrelevant of course).
#1 Ctrl+alt drag dialog so you have the scene you need to create fill from on both EDIT and MAKE FILL tracks.
#2 Run detect silence on the Scene on the EDIT track.
Make sure settings are set low enough that not just dialogue is kept, but also large movments, clothes rustles and footsteps as well. Also make sure the pre/post roll is set at the minimum length you think might be needed to create good sounding crossfades. Click process.
Result: only the unwanted material is left.
#3 Check preferences, delete overlaps pref ON.
#4 Ctrl drag the selected unwanted material over to the MAKE FILL track (thus staying in sync).
Now all the unwanted material is SELECTED and all the fill parts are left between the selected events.
#5 Now delete all the selected unwanted material (just hit backspace).
Result only fill material is left but with a lot of holes in it.
#6 Enable snap+shuffle.
#7 Select all the fill events, but not the first event. drag the selected event to the end of the first event and you will now have your continous selection of fill material.

Now I would quickly listen through the transitions (with a slight preroll set in my transport panel), deleting events/trimming away parts that doesnt work well as fill.

Crossfade to taste. Dont apply crossfades longer than your set pre /post roll in detect silence or unwanted sound will start to creep back in during the crossfades.

Large parts of the above can be handled using macro and PLE making it a pretty easy and fast process.

Yes the explanation is longwinded and seems complex, but it isnt hard or complex when you’ve tried it a few times. The hardest part is actually setting the detect silence thresholds properly.

Here is a gif showing how to:

1 Like

Wouldn’t it be possible to simply “crawl” between the two tracks and cut the original up that way and omit the step mixing the two?

Thanks for the replies! Great work-arounds. Am hoping this can be programmed into a future version of Cubendo.

+1 on this. Another killer to dialogue workflow, much simpler, easier in PT to detect silence, and strip all but silence. The workaround certainly works, but something more akin to the functionality in PT would certainly be appreciated.

couldn’t you…

  1. take a section of the “Fill” room noise and use the copy X-number of times, to create a “Fill” track.
    put a gate on it, have the dialog track side-chain the gate.
  2. use detect silence on the dialog track so there is nothing in the holes where you want the “Fill” material

this would put fill material in all the holes of the dialog track.


Hey Sunshy,

did a proper version of this ever get released? I’m looking for the same…8 years later!

Haha, don’t think so. But now that izotope has ambience match, we don’t need it any more. Life moves so fast these days!