Waveform inversion and the 0 dB ceiling

I guess I’ll start off by asking if the zero dB ceiling is the same actual volume before playback no matter what digital system you are using, which brings me to my next more involved question: When I import my wavs into audacity, the inversion of the waveform is such that the peaks relative to the ceiling are at the bottom and if invert the waveform for the purpose of easy viewing, does this affect my headroom?

The “zero dB ceiling” refers to “signal level”.
“Volume” refers to the “sound pressure level” (SPL) of the sound after it comes out of the speakers/headphones.
The “volume” (how “loud”) depends as much on the audio amplifier and speakers/headphones as it does on the actual signal level.

Not sure what you mean. Can you post a screenshot?

The dB scale goes up and down from the centre line of a track (though in Audacity it is only shown above the centre line.
The centre line represents -inf (minus infinity) dB.
The top OR bottom of the track (marked as +/- 1.0 on the standard Audacity track view) represent 0 dB.
Inverting a waveform does not change the dB level.

That makes a world of sense to me now. While we’re on the topic of zero dB, I’ve read that 32-bit floating point gives a lot more headroom. The wav files I import into audacity were mixed in reason, which uses 32-bit float. Some of these mixes clip, but it’s not audible. Is this the 32-bit float?

By the way, I should mention they clip in Audacity regardless of the format I switch to and I no longer have Reason to go back and check my meters. Most of my mixes are not too difficult to fix and clean up in Audacity, but my older wavs exported from Reason before I knew the golden rules of mixing are too far gone to really do anything with.

32-bit float format is able to handle values over 0 dB.
Sound cards do not use “float format” they use “integer formats” so they will always clip at 0 dB.

IF the files that you have are 32 bit float format then it may be possible to bring them back down to within the 0 dB limit by using the Amplify effect in Audacity.
The Amplify effect can “amplify” up or down (to a higher or lower level).
By default the Amplify effect will amplify to 0 dB.

I don’t think that it is likely that your files were saved in 32 bit float format as the format is rarely used for audio outside of “internal processing”.

32 Float is handy if you do processing or effects that result in your show going seriously over “0” (remarkably easy to do, by the way). Watch all the red overload indicators, reach over to Effect > Amplify and bring everything back down again. No harm done.

If you exceed 0 at the capture or conventional export step, that’s the end of the show.

And oddly, there is no fixed relationship between audio and sound. You can play your rock music recorded correctly with peaks at “0” and send it through a powerful sound system, boost it up to +125 dBSPL and blow out windows and terrify the cat.

That one is obvious, but this is much more serious the other direction. You just bought your new USB microphone and it doesn’t have volume controls on it. In rough numbers you can produce a show with volume variations of about 40 - 60dB roughly the loud/soft range of human hearing in the daytime drinking coffee and reading the paper.

Human hearing in the real world goes from 0 to +180 dBSPL. Dead quiet to thunderclaps. Which 60dB – or 40dB portion would like like to record – you only get one segment and you can’t move it during the show. Also, most people are spoiled rotten by professionally produced music and want to hit their MP3 performance to the nearest two or three dB. You see the disconnect?

People who are very serious about recording their live performances are author’s of some very long forum postings as we slowly correct all the misconceptions and errors. You can and we do use Audacity to produce some very nice recordings, but you do have to pay very careful attention to what you’re doing.


So do I really need to pay attention to the format when exporting material amplified to zero, some of which may have clipped in the original version before using Audacity, before I burn them to CD?

There is no “only clipped in the original version.” Once you take the tops (or bottoms) from a waveform, you can never accurately put it back. Here’s a transmission I got from NPR with serious and permanent clipping. I reduced the volume of the show to pull the waves down from the yellow line (clipping), but the damage is done.

You only get the magic of 32-floating inside Audacity unless you found a sound card that will capture that way, and I don’t know of any.

Yes, you can help with Steve’s Clip Fix tool, but don’t expect it to turn raspy, overloaded music into Warner Brothers Media.

Screen shot 2011-12-28 at 6.28.25 PM.png

So nothing you do is going to change that music and I think the elves are agreed that using 0 as a target isn’t a good idea, either. Too many things can go wrong with a music track constantly on the edge of overloading. Use -1 for everything.


So my understanding is that clipping won’t necessary be audible until playback? It makes me think of Chevelle’s “Wonder What’s Next” – beautiful production overall and really loud, but turn it up and give it the tiniest treble or bass on your system and it’s just plain overwhelming. I’m sure they didn’t exceed zero, but maybe just a little too close? Using Audacity lets me fix some basic mistakes I made with unnecessary EQ and compression in my mixes, and my sound is noticeably better and my perceived loudness is a bit more now, but I can’t help but wonder just how much any of those clipped songs are going to come back to bite me.

To clarify, I meant to ask if clipping becomes more apparent at varying volumes and qualities of playback systems.

On a low quality playback system, a small amount of clipping may go unnoticed, but immediately become irksome when listened to on a high quality system.