PDA

View Full Version : Question about weird looking wave


Egg_Crisis
2010-12-11, 06:09 PM
Both screenshots represent the same piece of audio, both are ripped from the same boot by two different people. It represents the end of one song, some audience sounds then the start of the next song.

This first one is how the wave should look when ripped properly...

http://i53.tinypic.com/2psn3oz.jpg


The second one has had the levels increased, but it's not simply been normalised - you can see the audience bit has somehow been boosted to maximum volume so the entire wave is one messy splodge of horribleness.

http://i53.tinypic.com/2vsf56p.jpg

Question: How would someone have made the wave look like this?

Is there a setting on some cd ripping programs that through a users incompetence could accidentally make the wave look like this, or is it more likely the audio has been deliberately & maliciously destroyed?

paddington
2010-12-11, 06:44 PM
the setting you are asking about is "normalize" and is common on many CD burning programs. It should never be used.

however, I don;t think that is what was done here.
Generally, unless it sucks ass, a normalization algorithm will simply boost the modulation until the loudest sound peak hits 99% (or however you set it) which has the effect of bringing ALL the audio up the same amount - until that loudest peak hits max, then the amplification stops.

The only way for that process to modify the wav in a manner that matches your post is for some jackass to only normalize that quiet section as a separate selection, rather than the whole song / tape at once, as it should be done.

I do not think that was done, here.


That said, if they ran it through a limiter (which would not be part of a CD Burning app), the limiter, with enough drive, would do pretty much what you have there... notice the removal of the dynamics during the song and the huge amp of the banter sounds.
Limiting basically jacks up the volume of the wav, but limits how high it can reach, absolutely. A "line" is drawn and any audio peaks that go over are clipped off, so the loudest sounds never get louder, but the softest sounds come way up (depending on how much you drive it).

In effect, the dynamic range (difference between the softest and loudest sounds) is decreased, which looks to be exactly what has happened to your audio, here.

AAR.oner
2010-12-11, 06:47 PM
looks like they mighta used a compressor with normalization...few other possibilities as well, impossible to tell by just lookin at a wave form...I don't know of a CD extraction prog that'd do it automatically but its possible I guess

rspencer
2010-12-11, 07:50 PM
the setting you are asking about is "normalize" and is common on many CD burning programs. It should never be used.


While a common belief, this is incorrect. Or, at the least, too ambiguous to be useful.

Normalization can & often should be used, as long as it is "peak" normalization...which is basically what you go on to describe. The source is scanned, and then amplified so that the highest peak reaches a predetermined level (e.g., -0.1 dB). If the highest peak is -6 dB, then the entire wave would be amplified by 5.9 dB. This increases the volume while maintaining the dynamic range.

The "bad" normalization is "RMS" normalization. It sets the RMS average to a predetermined level, compressing any over peaks. This squashes the dynamic range. While many consider it "bad," it does have its uses and is sometimes the better way to go, but most often not.

While the second pic could illustrate normalization using a limiter, I don't think that's the case. The peaks are not as uniform as one would expect had a limiter been used. It looks to me more likely RMS normalization.

paddington
2010-12-11, 09:33 PM
I've never seen an RMS normalizer - either software or hardware... isn;t that pretty much just a compressor?

rspencer
2010-12-11, 11:33 PM
It's a feature in most audio software. Sound Forge, Audacity, Pro Tools...

Compression is often a facet of it, yes, but not necessarily.


I have used it; it has been the best option in certain situations. For example, in a show with not a wide dynamic range, say an acoustic solo performer, the greatest dynamic range will be the applause vs. the performer. So I raised the RMS level to get a good peak level, along with compressing the applause. It loses the dynamic range, but it wasn't preferable in a case like that.

roann
2010-12-13, 11:28 AM
I'm pretty sure this was caused by a compressor. It basically allows you to boost nearly anything that's lower than the highest peak to that maximum - if you a) have no clue how to use the compressor software properly, and b) cannot hear anything (even Pete Townsend will tell you how bad such a butchered recording sounds).

Using a compressor that moronic way the dynamic range will go down to zero. That's why you cannot get the second wave with normalisation. As said above this increases the volume with a given level while maintaining the dynamic range.

Five
2010-12-16, 02:47 AM
While a common belief, this is incorrect. Or, at the least, too ambiguous to be useful.

Normalization can & often should be used, as long as it is "peak" normalization...which is basically what you go on to describe. The source is scanned, and then amplified so that the highest peak reaches a predetermined level (e.g., -0.1 dB). If the highest peak is -6 dB, then the entire wave would be amplified by 5.9 dB. This increases the volume while maintaining the dynamic range.

The "bad" normalization is "RMS" normalization. It sets the RMS average to a predetermined level, compressing any over peaks. This squashes the dynamic range. While many consider it "bad," it does have its uses and is sometimes the better way to go, but most often not.

While the second pic could illustrate normalization using a limiter, I don't think that's the case. The peaks are not as uniform as one would expect had a limiter been used. It looks to me more likely RMS normalization.
biggest thing to pay attention to is that the 'normalization' is applied to the show as a whole, based on the highest peak of the entire program material. otherwise the volume changes at every cue stop & is really annoying. not sure if any progs do this anymore, iirc nero used to do this shit :rolleyes:



for the original question of the thread, it looks like compressor to me as well.

rspencer
2010-12-16, 04:01 PM
NERO burning app will do that if you check "normalize files." I've only used it maybe once or twice, just burning mix discs for the car. It scans & normalizes each track individually.

NERO wave editor, however, will do it as a whole file if that's what you highlight (just like any other editor).