Topic: Is there a limit to how much I should amplify a file in Audacity?

[page]

Yes. The applause is frequently louder than the music.

The easiest way in audacity then is to amplify the file to just under 0, then figure out how much louder the applause is compared to the music. For example sake, lets say that the applause is 3db louder than the music (so the music is around -3 on most of the loud songs. Select a section of applause between songs, and use the "hard limiter" under the effect's menu. To bring it to just slightly louder than the music and sound fairly natural, use -5db as your threshold, a wet level of 1, and a residual level of around 0.5. What the first and last numbers do is the limiter will start to kick in and shave everything louder than -5, and the residual value will take the average of the unlimited section and the -5 section to result in a new peak of -2.5 (it's akin to a "soft limit" as best I can tell from my tests). Play around with the residual value and threshold to see what works best for each between song section (and do them individually). It's not flawless, but when your doing small amounts of reduction, I haven't been disappointed. If I had to crank out like 10db with it, I'd use something else, but for purposes like this I'm happy with it.

[morst]

Page, are you suggesting to use the amplify effect in order to determine peak levels, but use the limiter to do the fix? If so, be sure to undo the amplify action before gaining up the whole thing using the limiter. If you are suggesting to amplify, then limit, I'll disagree, and say just use one process to do both. Though I don't really like the sound of audacity's hard limit plugin. 

[runonce]

Ok sounds like the applause levels are actually louder than the music.

If it was a quiet show and there are huge applause swells - I'd resort to manually selecting each applause segment and just bringing them down with the Amplify tool -

-5 or 6db at a time. As long as the level shift isnt bothersome. (And you can use the envelope tool to prevent the shift if you like)  You might easily use -10db...

Another thing I've done is cut a few seconds of the loudest applause...it can sound a little cooky - but a least it gets it over without being too obnoxious.

[Shadow_7]

You could try a hard limiter to cap the applause at a certain dB, then amp by the remainder.  You probably don't want to do that a lot, but if it's just a couple localized claps that are peaking hot, a nice way to tame the beast without invoking intentional clipping.  Effect -> 151 -> Hard Limiter... (on my system / -5dB is a good start most times, or whatever that might be relative to Effect -> Amplify)  It should give you a few more dB of gain. 

You can also apply gain in some sort of envelope thing, but I've not mastered that way yet.  One option is to get your mics higher at the point of capture, better balance of applause versus group.  Or just sit where there's an audience void (easily done for most things I record in the overly large venue relative to the audience).  You might also try the OMNI caps as they'll have less proximity effect.  But you wont isolate from the audience as much if there's more behind you than in front.

[page]

Page, are you suggesting to use the amplify effect in order to determine peak levels, but use the limiter to do the fix? If so, be sure to undo the amplify action before gaining up the whole thing using the limiter. If you are suggesting to amplify, then limit, I'll disagree, and say just use one process to do both. Though I don't really like the sound of audacity's hard limit plugin. 

I don't use the hard limiter as a gain function (period, at any time). I do in ozone because it doesn't give me much choice (so for precision edits, I have to use audacity's), but audacity's can function as a limiter without gain. The reason I specified gain first, was gain on the entire show to bring it up to levels that are easier to work with. Your end result is gain (entire show) > limit (selective portions) > gain (entire show). Technically you could limit the entire file but with the residue value set at half, you'd be effecting musical portions even as they are under what you want to resulting peak to be, hence the surgical method noted above. I find I have more trouble working with stuff that's around -15 then I do at -5, but I've also got my metering configured so that I only see the first 18db or so, but I get a lot of detail in that area.  Otherwise doing the limits first, and then gaining the entire show would work just as well.

As for the sound of the limiter, I think it's ok when you keep the residual value above 0.3, any lower then that and I agree it turns to ass (IMHO). The other time it falls apart is if you're knocking out more then 5-7db of the resulting file (so a limit of 10-14db with a residue of half).

Ok sounds like the applause levels are actually louder than the music.

If it was a quiet show and there are huge applause swells - I'd resort to manually selecting each applause segment and just bringing them down with the Amplify tool -

-5 or 6db at a time. As long as the level shift isnt bothersome. (And you can use the envelope tool to prevent the shift if you like)  You might easily use -10db...

Another thing I've done is cut a few seconds of the loudest applause...it can sound a little cooky - but a least it gets it over without being too obnoxious.

I've tried it both ways and I didn't like the effect on the background sound when that was done. That works best when you have applause that isn't mingled with music, at all (so no clapping during the trail end of a song) since you suddenly lose that musical content when you do the amplify function with a negative value. The envelope tool works better, but then suddenly your crowd gets quieter slowly, but so does everything else so it's similar to the effect of walking out of the concert hall backwards (sound decreases as does the loud stuff) which is better than the hard change, but still sounds meh to me.

[CTjazzfanatic]

Thanks for the help.

[morst]

> limit (selective portions)

How are you limiting selected portions? Is there an abrupt change at the start and end of these sections?

I have been using the Masterworks limiter in Digital Performer, and it's got nice clean gain stages, and then you can limit on top of that. One stop shopping!

[page]

> limit (selective portions)

How are you limiting selected portions? Is there an abrupt change at the start and end of these sections?

Pretend the middle blob is clapping or whatever you want to limit and the stuff on either side is stuff you don't want to limit:









Now select what you want to limit and bring up the Hard Limit window, after picking stuff, run it. (these are example settings for illustration)










Resulting wave form. Notice nothing on either side was touched. Also notice that the second blue area (symbolizing a more constant power), was (largely) unaffected, thats the difference between using an envelope and using the limiter and also why I advocate for the limiter in this situation.







So while this is sort of a crap example as to why someone would want to use the limiter with the above wave form (it was what I had open at the time, meh), but it's merely an illustration as to how it works. Now why I think it's appropriate to use in the original request is because if the applause is literally the loudest peak in the file, then once he limits it, he'll be able to amplify the entire file and the net gain will be signal during songs. Of course, it would be preferable to get that taken care of up front by recording in a location that yields more signal to clapping/noise (since this isn't a perfect solution, just the best one I can think of), but if that's what he has to work with, this the solution I recommend. Make more sense?

[Shadow_7]

I tend to always use hard limiting.  Why?  Because of drummers.  They are going to play as loud as they want, even if they're part of a flute ensemble.  Put the whole group in front of a brick wall and it's that much more out of whack.  So it's not just claps IME.  Getting a better vantage point helps, but oddities like door slams, fireworks, dropped books or drinks, and other things can and will give you those hyped peaks.  You don't want to just deamp the claps as they'll likely happen before the last note finishes.  For the image you posted, about -4.5dB or -4dB is where I'd hard limit to.  Depending on what percentage of the content that actually is if you zoom in on the waveforms.  Just gracing the peeks which can generally give you another 3dB or more of gain and bring the content of interest into the audible zone.

[rjp]

I tend to always use hard limiting.  Why?  Because of drummers.

Percussion in general can sometimes be tricky. The other night, I was recording an orchestra playing Shostakovich film scores, and one of them had some sort of "snapping" percussion in the duel and death scene from Hamlet. Nothing like seeing the "peak" light blink on the recorder when even the loudest passages earlier hadn't come close to full signal! As it turned out, the snaps didn't actually distort anything, and could be further hard-limited down without audible effect (so that I could amplify the recording in post). Other than the snaps, normal peaks were around -6 dBFS before final amplification.

I also routinely use hard limiting for applause (always previewing first). When I'm running cards ORTF or DIN, very little is needed, but when using in-ear omnis I generally need a lot more limiting.

[morst]

Make more sense?

Yes, and no. "The more you explain, the less I understand" (Mark Twain)

I understand how to operate audacity in order to apply effects to selected areas. My concern is that using this technique will result in abrupt audible changes at the point of the effect starting and stopping. Unless the effect is used VERY sparingly, I would think it would be rather jarring when the effect kicks in, though admittedly, I have not used the Hard Limit effect much.

I suppose if you targeted the selection to JUST the applause, it might be easy to do and have it be smooth-sounding. But if the applause covers any music, then rapid fades or limiting would affect the music too, no?

[page]

I suppose if you targeted the selection to JUST the applause, it might be easy to do and have it be smooth-sounding. But if the applause covers any music, then rapid fades or limiting would affect the music too, no?

Limiting would effect it if your music is loud enough to be picked up in that initial threshold, and specifically if it's the difference between nothing and the threshold based on your residue value, however fades/envelopes would every time. Do I need another set of pictures? 

If your music peaks at -18, and your clapping peaks at -2, then limiting as far down with a threshold of -16 (and a residue of .5 to make the math simple) should leave the music unaffected and your clapping reduced to about -9 peaks. (this is completely ignoring how good your limiter is for avoiding a metalic sound, that's a different problem, I'm just addressing the point of "does it affect quieter stuff") Remember, it won't take the meat out of the sound that resides in the decay or surrounding impact, just the spike. What this does is simply buy you some headroom to amplify the overall file (including your music) more than you would have originally.

Some limiters are better than others in this regard, and that's part of why I don't use the audacity one for as much as I would others (say, Ozone's which I think falls appart at about twice or three times the limiting used than Audacity's). Second, I recomend that you play around with it a lot to get familiar with it.

[mr qpl]

I like using an envelope to limit the louder applause. So, f'rinstance, go from 100 % to 70% and back up to 100% with splined curves. looks like a steep sided capital "U".  The change down from 100% is less noticeable than with a straight 70% amplify. (Cool Edit is the software I use for this)

[runonce]

If there are local/nearby clappers that persist as a song starts or ends - I've resorted to zooming in, selecting each individual clap and reducing them with amplify tool...effective but laborious...

Depends on how important the recording is...