Topic: Processing to remove specific high end noise at intervals

[hoserama]

So basically I have some recordings that have some sporadic noise in the 4k-9k region that's very annoying. It's always at a very specific interval (like twice a second at equal distances). I have had excellent success in removing it using both Izotope RX Spectral Repair and/or Adobe Auditions 3.0 Auto-heal. However, it is *very* tedious to have to knock out all these individually.

I figure there's got to be a way to run a script to process it all through. If I find the exact specific interval that each click is on, how could I run a script with the Auto-heal or spectral repair? Any help would be appreciated.

I have Audition, Sony Vegas, and full versions of RX advance.

[Shadow_7]

There's a nyquist notch method.  In audacity Effect -> Nyquist Prompt...
(notch2 s freq q)

Parenthesis are part of the command.

(notch2 s 9000 q)

Some form of lisp or something.  Analyze -> Plot spectrum helps you ID what their frequency is, q, and result after you adjust.  Although it just makes them quieter so maybe not what you're looking for.  I generally replace q with (* q dB) where dB is the level shown in the plotted spectrum.  This seems to bring it down to the level of everything else.  Or I could be wrong.  Otherwise q can be a number and Edit -> Undo until it's close enough.  Also very tedious.  But you can make a plugin with a bunch of them and just run and edit that as needed.  And a bunch of other, if I only knew more math things with that nyquist stuff.

[runonce]

There's a nyquist notch method.  In audacity Effect -> Nyquist Prompt...
(notch2 s freq q)

Parenthesis are part of the command.

(notch2 s 9000 q)

Some form of lisp or something.  Analyze -> Plot spectrum helps you ID what their frequency is, q, and result after you adjust.  Although it just makes them quieter so maybe not what you're looking for.  I generally replace q with (* q dB) where dB is the level shown in the plotted spectrum.  This seems to bring it down to the level of everything else.  Or I could be wrong.  Otherwise q can be a number and Edit -> Undo until it's close enough.  Also very tedious.  But you can make a plugin with a bunch of them and just run and edit that as needed.  And a bunch of other, if I only knew more math things with that nyquist stuff.

If you have some audio that has only the noise(no music)...you could try Audacity's noise reduction tool.

But you really need audio with just the noise so Audacity can "learn" the noise.

Then you can apply that the whole wave at once...rather than one at a time.

Make sure to use the least aggressive settings...