Topic: Attenuating clapping in live recordings - the easy way

[Ben Turnbull]

Tone, I made the attachments so as to avoid a long winded description of what the tool does, hoping the graphical representation would provide the "thousand words" business... that and I'm not really all that technically proficient as to what is supposed to be going on with the process, I just find that I seem to have been able to harness it for my purposes.  In your reply you suggest that I'm addressing each clap individually, but I'm not.  Why?  Well because it's time consuming that's why.  This provides a bit of automation and a result that I'm able to live with.  I select the stretch of offensive audience appreciation, make a judgment as to how hard to hit/limit the max clap level, apply the preset and BOOM!  Move on to the next offense.  It will work on individual incidences for sure, but it shines as a block editor.  (I used to remove the ass hat clappers altogether but in my old age I have less time to obsess over that so I now modify the offenders by making it seem as though I've moved them... over there ------>| )

The other part, I will try to coach you through it though, since you asked and since you haven't tried the tool yet.  Perhaps it will give you a little push...  Buckle up, this could get wordy and keep in mind that specific details apply only to the Audition DAW.

In the graphical shot of the dynamics processing tool you saw a graph which has dB across the bottom and on the right side.  These are sound levels, soft to loud, running LT to RT and Bottom to Top respectively.  These axis represent input levels (original signal) along the bottom and output levels (processed signal) up and down.... Here, this is from the HELP pages of Audition, which I have to say are written well enough to have given me all the insight I need to make me happy....

The graph depicts input level along the x-axis (left and right) and the new output level along the y-axis (up and down). A line that flows directly from the lower-left to the upper-right (default) depicts a signal that has been left untouched {this is my NULL part}, since every input value goes to the exact matching output value. Adjusting the shape of this line will adjust the input or output assignments, thereby altering the dynamic range.

For example, you can boost all input that has a level of around -20dB, leaving everything else unchanged. You can also draw an inverse line (a line from upper-left to lower-right) that will dramatically boost low amplitudes while dramatically suppressing high amplitudes (that is, all quiet sounds will be loud, and all loud sounds will be quiet).


So, my set of presets allows me to address different ranges of clap peaks relative to the other signal levels around them.  For example, the image I clipped represents an application where the peaks were relatively loud and I wanted to keep them relatively loud in comparison but tame the spikes.  Jazz drum solos can be especially troublesome in this regard.  In this example, everything in the selection to be processed, up to and including -16dB is unchanged or Null processed if you will.  Then it starts to compress downwards the input levels on a ratio of 2:1 such that input levels of -8.8dB are reduced to output levels of around -12dB.  From there input levels -8.8dB to 0dB are scrunched down at -2.6:1 such that the loudest output is around -15dB or so.  Open the tool hover over the graph area and you'll see a read out like  -8.8dB -> -12dB below the graph area.  These number pairs depict how the output levels will change depending on the input level.  Pick below the "null line" and the input is reduced; above it's amplified.  I've built other presets, specifically for ass clappers, that start much lower than -16dB and begin knock down the peaks harder.  The fact that the processing is based on ratios of input to output, instead of a hard limit value, makes for a bit more natural sounding applause which is, after all, the sure sign of a live performance.  Without a little clap in our tapes, how would anyone know our work from store bought studio multi track productions? 

In your most recent post re: DAW wish list, I'd agree that the best way to understand this tool set is to play with it, read the help pages, then read them again... and eventually you will start to understand the interconnectedness of the tool kit and how the various "edges" can be applied to shape your tapes for the better.  Also, YouTube is a great help in learning general concepts and terminology.  Have some fun!

[806 words via MS Word]

[capnhook]

^  LOL.  I have a taper dream that involves me owning a venue that books bands that everybody wants to see so much that they'd put on muzzles and mittens upon entering the venue.  I like the mitten method of clap attenuation.

Mittens?  Great idea for every taper's gear bag.....if we all carried them, breaking them out could soon become the universal sign to "stop f#%king clapping so loud"........

Worth a try..

[stevetoney]

^  LOL.  I have a taper dream that involves me owning a venue that books bands that everybody wants to see so much that they'd put on muzzles and mittens upon entering the venue.  I like the mitten method of clap attenuation.

Mittens?  Great idea for every taper's gear bag.....if we all carried them, breaking them out could soon become the universal sign to "stop f#%king clapping so loud"........

Worth a try..

Doubles as an emergency wind screen!

[stevetoney]

Tone, I made the attachments so as to avoid a long winded description of what the tool does, hoping the graphical representation would provide the "thousand words" business...

I understand about the screen shots.  It's just that these special tools just aren't very intuitive and I've tried to figure out how best to use them, but haven't been successful.  So, the extra detailed response that you provided is REALLY appreciated.  I know it takes time to write. 

I've got a recent recording with some really obnoxious clapping going on.  I'll bring that up and then re-visit your explanation.  If I have any more questions, I'll come back to this thread.  Thanks again man!

[Ben Turnbull]

Not a problem, I just don't like typing on my laptop. 

If you have any more questions just post back or PM me.  In the mean time just fool around with the tool.    Use my screen shots to set your tools and see if they come close to doing what you want.  Change the threshold or starting position/level and see what that does. 

Lots of youtube vids on dynamic processing but they may not be exactly what you need... I'll see if I can put together a bit more for show and tell later.

[nardo]

Thanks to this thread I now also use the click removal in Audacity. I highlight the section that contains the clapping, for this show I'm working on right now I set threshold 200 and spike width 30. This got most of the clicks, the ones that were still there I used the "Repair" effect on (for that you have to zoom in all the way, it only works on a few samples).

[bluntforcetrauma]

If you look at the waveforms in your editing software, the clapping sections are distinctive.  High narrow peaks - usually louder than the surrounding music or other recording.  It can take a bit of trial and error, but I usually start with the hard limit at or below the peak level level of surrounding program.  This way the loud claps are attenuated but the rest of the applause sounds pretty natural.  I find this takes me much less time than highlighting and adjusting each clap individually and I've often got claps that all don't respond to the same click removal settings.  If the overall applause is still too loud, the envelope tool is your friend.

This is why I use a limiter in general; once I figured out the loudest part I want to remain untouched, I shave off everything above that. I like that for clapping for two reasons; they don't have much sustain energy and part of what I find so offensive with them is the transient more so than the sound.

Your results will be based off of two things; you're ability to figure out that break-even point where you're not adversely harming the content you want to keep and second, the quality of your limiter plugin (they aren't all equal... Some can dig deeper without noticeable sonic impairment).

IMHO, noise reduction is never going to be the best choice for something like clapping because a noise reduction algorithm it's looking for steady-state noise to cancel (ex. HVAC drone, hiss, etc.) as opposed to intermittent random noise like clapping, etc.  That said, I've never been satisfied with Audacity's noise reduction - it always has given me the phasey "underwater" sound you experienced, or it sucked out percussive transients from the music.  BUT - the noise reduction in iZotope RX quite frankly is spectacular.  It really is able to cancel the background noise without harming the actual music, at least to my ears.  This is no help for claps or stand hits though.

I concur with all of this.

100ms is a long release for a hard limiter, especially on applause.  I'd try 1ms or less.

Good digital limiters are lookahead such that they effectively have no attack time--the attack will be applied in advance of the peak.

QFT. beat me to it.

How would one generally use a limiter?  it makes sense to use limiter the way you put it, you find the loudest part you want to remain untouched, then shave every thing off above that?  I understand the statement, but what would be a step by step instruction to get the result that is made in the statement?
I use PEAK which has Squeez limiters, but possibly maybe most limiters are close to being the same?
thanks for the help

[dean]

.marking thread

Ditto

[ButchAlmberg]

I just taped Warren Haynes @ Jannus Live and never realized how much I despise clappers. I'm convinced this one woman knew exactly what she was doing, too. This thread is just what I was looking for.

[ilduclo]

not sure if I've posted in this particular thread or not, but the envelope ampilfy works great for this. I start at 100%, drop relatively quickly to 60% (usually) carry a flat 60% across the bottom of the envelope and then sharply back up to 100%, spline the curves (asymtotic). This gets a natural sounding 40% reduction in the applause. I have also combined this in a "reverse fashion" by amplifying quiet music using the envelope amplify function going from 100% up to 160%, then again back to 100 sharply. Combining these 2 has really worked well for me for a nice sounding recording of quiet music that was "enthusiastically appreciated" by the audience.

here's a screen shot of the applause lowering envelope, note the basis is 100% at the top, so the envelope goes lower than that below the top



and here's the 160% amplify upward. Note that the basis is 200% at the top and 100% in the middle, so the envelope amplifies when above the center



Of course, it goes without saying that the flat section in the middle where we are slope=0% should be "played with" until it sounds best to you. I use the 40% down and 60% up as what works for me and as sort of a maximum, above and below which I can really hear the amplification


and, of course, save your unedited files.

[hoserama]

If you're going to do that in Audition (that looks like Audition), just load it into the multitrack editor and automate the volume via envelop editing there. Non-destructive editing and then just export when you have the settings you want.

[ilduclo]

that's actually a soviet version of cool edit pro. I have those stored as presets, so it's pretty easy to do them, I don't generate those curves each time!

Question on this. The envelope works great for end of song clapping,  but what hard limiting values should be used for clapping during songs, like when people clap at the end of a good solo? I used to have a couple of presets that worked ok, but I lost them when migrating from one pc to another.  My hard limit has the following adjustable attributes

limit max amplitude to ____db
boost input by ________db
look ahead time _________ms (5....20) ------comes out of the box at 7 ms
release time _________ms (40...200)----comes out of the box at 100 ms


I'd like to be able to set it to achieve the same 40% reduction in clapping volume as I've done with the envelope

all suggestions would be MASSIVELY appreciated. I need to do this to something I recorded last night!

[voltronic]

^ I don't have cool edit, but try the suggestions I posted for Audacity at the beginning of the thread for this type of thing.  Or use the Hard Limiter with a Residue level of 0.7 as a starting point.

[Ben Turnbull]

Revisiting... The trouble with envelope attenuation is that, while you know you can knock down the peak claps with it by what ever percentage, you also lower the well behaved clappers close by and the room in general.  When you go too far, it sounds like the room has been moved well away from the action and you don't get a natural sounding reaction to the show.  These are live recordings after all.* That's why I've more or less gone to dynamic processing for most of my repairs.  It keeps the good local applause and limits the exuberant one or two at the same time...  YMMV

* I have prepared "studio" versions of live shows where I've cut the house out all together and cross faded the gaps.  They're alright if that's all you can salvage.

[voltronic]

Revisiting... The trouble with envelope attenuation is that, while you know you can knock down the peak claps with it by what ever percentage, you also lower the well behaved clappers close by and the room in general.  When you go to far, it sounds like the room has been moved well away from the action and you don't get a natural sounding reaction to the show.  These are live recordings after all.* That's why I've more or less gone to dynamic processing for most of my repairs.  It keeps the good local applause and limits the exuberant one or two at the same time...  YMMV

Yes that's a good point.  I tried envelopes once for this purpose but quickly abandoned them for the reasons you mentioned.  Now I use a combination of declicking for periodic claps and/or limiting for continuous washes of applause to reduce the loudest applause to that of the highest peaks, and the results are much more natural.

* I have prepared "studio" versions of live shows where I've cut the house out all together and cross faded the gaps.  They're alright if that's all you can salvage.

I did this once for a choral concert I performed in at the request of our conductor.  I recorded the dress rehearsal run in the afternoon and then was able to leave my setup exactly in place to record the concert that night.  The audience was exceptionally quiet and well behaved, and didn't jump in with applause immediately after the last cutoff.  This gave me two takes of the concert that sounded close enough where I could make "studio" comp edits.  I had plenty of clean hall tone to insert between tracks so the background was pretty seamless throughout the CD.  Not sure I'd ever do something like that again, but it worked out surprisingly well.