Waves going above 0db

Greetings everyone!
When mixing multiple audio sources that were not previously clipped and exporting to 32-bit float, the final result always exceeds the 0db limit. As amplifying to a negative value or normalising to 0db bring the volume of the entire mix down until no clipping remains, I instead use audacity’s limiter (with input gain for both channels set to 0, limit to 0db, and hold set to 10ms) to only knock out the peaks going beyond 0db, and therefore let the overall volume unchanged. however I’ve been wondering about 3 things:
1. when both inputs gain are set to 0, are the clipped waves simply amplified down (simple volume change with minimal alteration done to the audio) or are they treated as if they were being boosted in volume (meaning that hold time could unnecessarily alter the audio in a way that distortion, and therefore quality loss, could potentially be introduced) ?
2. does the limiter only alter the waves that are going above 0dB or does it alter the entire audio?
3. Is using the limiter for that purpose recommended or is there a better way for achieving that goal?

In any case, Thank you so much for your time! I sincerely hope I was not too long :sweat_smile:
May everyone have a magnificent day !

That’s NOT how mixing is normally done!

Mixing is done by summation but analog mixers (which are built-around a summing amplifier) always have a master level control, and full-DAW applications also have a master volume control. So it ends-up as weighted average rather than a simple sum.

Since Audacity doesn’t have a master control, you can either export as floating-point, re-import and Amplify or Normalize, or you can lower the individual track levels. In reality, you often need to re-import and re-adjust the levels anyway (AKA “mastering” as a separate step after mixing.)

Or you can go to the Tracks menu, Mix and Render, then Amplify or Normalize the mix before exporting.

The limiter has a threshold (“Limit To”) and it has little or no effect below the threshold. i.e. If you limit to 0dB twice it should have no effect the 2nd time.

The input gain takes affect before limiting, so that will make the limiter kick-in sooner (at a lower level).

Make-up gain happens after limiting. The “traditional” way of making music “louder” is to apply limiting or compression to push down the peaks or loud parts, then amplify or normalize to bring-up the overall loudness.

In any case it DOES alter the sound… Either the dynamics or the waveshape… It’s a non-linear effect.

With hard & soft limiting it lowers everything near where you go over the threshold and the hold time affects that.

Limiting is an effect, or with make-up gain it can be used to boost overall loudness without hard-clipping the peaks.

I don’t know all of the details but Audacity’s limiter has “look ahead” so it kicks-in in advance of the peak and it lowers the level without distorting the waveform.

A “normal” limiter just works on the instantaneous level, so it kicks-in before the threshold and starts “pushing down” the levels and distorting the wave without hard-clipping. I think the limiter’s “soft clipping” option is more like a standard limiter.

Hey DVDdoug, thanks for your response! I now have a better understanding of how audacity’s limiter functions (expecially when it comes down to the its threshold and hold time) . However, I am still left with one problem: volume decrease. You advised to use amplify or normalise on the final mix - the master- in order to bring the “clipped” waves below the 0db limit. But, as I explained in my previous post, doing so would bring the volume of the entire mix down until every wave fit in the 0db limit (resulting in the master being excessively quiet, or, at least, not reaching the audio volume targeted), instead of only decreasing the waves that exceed that 0db limit in volume. If using the limiter for that particular situation is not recommended, therefore what other technique should I use to bring just the “clipped” waves down ?

Thank you for reading me :slight_smile:

That’s correct (depending on your “target volume”).

Amplification and normalization are linear effects. They adjust everything (actually everything selected) by the same dB amount as-if you adjusted the volume control before playback started.

Virtually all commercial music is compressed and limited (with make-up gain) and often it’s over-done and it destroys the musical dynamic contrast making it “constantly loud” (and IMO boring). That’s how you win Loudness War.

Note that all of the popular streaming services are applying loudness normalization. They will cut or boost the level to hit their target volume. But if you have a highly-dynamic track with loud peaks and quiet parts, their automatic processing will not boost into clipping and that track may sound quieter than the target and quieter than the other music on their website.

This is not just related to mixing… Something like a solo acoustic guitar is highly-dynamic and when normalized/maximized it will sound quieter than a whole band or an over-driven electric guitar, etc.

Mixing makes the peaks higher AND it makes the music “denser” so a normalized/maximized mix will sound louder than a solo instrument or vocal.

A classical orchestra is a complex mix and it can be rather loud in real life (depending on the music being played). But it’s also highly-dynamic music and usually very-little compression or limiting is used during production so if you buy a classical music CD it will usually be quieter than your popular music CDs. And if you want to get the realistic sound of an orchestra at home, you have to turn-up the volume!

Hey DVDdoug! Thank you for your reply and sorry for responding to it this late! (had a lot of Internet connection issues to address :frowning: )

Yes, I do know and understand (and truly despise) what the loudness war is, and unfortunately, I must indeed engage in it whenever I mix, however it is not my will to do so: As to provide a bit of context, I mostly make unprofessional, private video game music mixes done using lossless masters found in those game’s respective official soundtracks (most often the ONLY lossless masters available). And, needless to say the loudness war is far from being over in those kind of albums, with tracks sometimes so compressed that they constantly hit the 0db limit; form their early begging to their very end.

Therefore, considering what you said about mixing in audacity being simple summation, it is indeed only logical that putting one over-compressed track with another that has also been utterly butchered will fatally result in clipping.
Nevertheless
I belive the loudness war is not necessarily the culprit of audio clipping:
Let us imagine we got 2 extremely dynamic tracks; both being orchestral music, let us say. If we were to try to transition from one track to another using various effects that affect audio volume in some way or another (reverb, delay, low/high pass filter, echo… ex. ) right when both tracks are hitting the 0db limit, supposedly a very intense moment, a Climax, a passage we surly do not want to amplify down: Would we not still get audio clipping?

Regardless, Since clipping often occurs on relatively short section, and that you advised that limiter is not supposed to be used for avoiding clipping, I instead try to zoom in my final mix’s waveform and amplify down only the clipped parts, while trying to match the waves’ original form as closely as possible.

Ohhhh dear that was a long message! (My apologies if it was too long! I attempted to be as concise as possible) In any case, thank you for taking the time for helping the poor noob that I am X) May you have a superb day! :slight_smile:

True! If a track hits 0dB and you mix-in ANYTHING without adjusting the volumes it will be summed and go over 0dB.

Did I say that? I didn’t mean to say that… I might have I said “it will change the sound”. And if you have highly-dynamic music you probably won’t hear a “reasonable” amount of limiting/compression. If it’s highly dynamic with a few high-peaks you might not notice 6dB of limiting.

That’s fine. Adjusting the individual tracks as the mixed track is a normal-everyday part of mixing and audio production.

If you aren’t doing this already, the Envelope Tool allows you to fade-up & fade-down without any sudden jumps in volume.

But of course it also “changes the sound” and if you are evening-out the sound level, it’s a kind of dynamic compression. Automatic Volume Control (or “automatic gain control”) is a slow-kind of compression and similar to what you’re doing manually, minus the human judgement. Limiting is a fast-kind of compression.

That’s a creative decision, entirely up to you! :wink: And mixing two different tracks is “creative” so there are no rules.

No. Not if you are in floating point and you adjust-down before making a different format.

Uh… I get kinda’ wordy myself… Sometimes I think I should have chosen the screen name “WordyDoug”. :smiley:

Thanks for using paragraphs!!! I hate long posts with no paragraph breaks!

This topic was automatically closed after 30 days. New replies are no longer allowed.