I understand about dynamic range compression and the loudness war, but after lots of searching it's hard for me to get a good grasp of it.

Let's compare two tracks, the first one is newer. It's called "Sowing Season." It seems louder and more impactful in it's production, compared to the second older track "Airbag." Each song is posted along with it's "Dynamic range compression" stats. Dynamic range is rated out of 20. Click on the link below the song to see those stats. Each song is the first song listed so just look at the first number in the sequence of numbers.


Sowing Season:
http://grooveshark.com/#!/s/Sowing+Season/1Xs0nh?src=5

http://dr.loudness-war.info/album/view/34836

Airbag:
http://grooveshark.com/#!/s/Airbag/4CCzFv?src=5

http://dr.loudness-war.info/album/view/55030



Now if you notice on Sowing Season,  they are able to get the song extremely loud about a minute in, in a way that on Airbag they are not. Also in general the first song seems crisper and more impactful in general. Why is this?

Stats are given at the bottom of each page

Sowing Season:

DR       Peak         RMS         
---------------------------------------------------

DR4               over         -6.53 dB        



Airbag:

DR        Peak        RMS             
---------------------------------------------
DR7      -0.12 dB    -9.00 dB     

So, given those stats, why does the first song that is more recent sound the way it does. Was it technologically impossible in 1997 to get Airbag to sound like Sowing season? Or was it just as possible but they just decided not to in order to not lose dynamic range?

Also in general the first song seems crisper and more impactful in general. Why is this?

It's clearly a stylistic choice.

"If you are patient enough (and have sufficient understanding) to set up each processor section carefully, the results are nothing short of superb — but in the wrong hands, the Finalizer can wreak more havoc on your music than a second‑hand Chinese cassette in a well‑used 4‑track!”

I don't think it's just a stylistic choice, I know of no songs that sound like Sowing Season made in the 90's. Play it at the same time as the loudest 90's song and it'll still completely drown the sound of the 90's song out. You can't tell me that there's not a clear difference in general crispness between any 90's song and that song that's a result of something different being done in the production.

DSP technology became affordable and popular in the 1990's and created a true boost in audio compression. One of the first examples is the TC Electronic Finalizer.
The music recording magazine SoundOnSound wrote in 1996:

Quote

"If you are patient enough (and have sufficient understanding) to set up each processor section carefully, the results are nothing short of superb — but in the wrong hands, the Finalizer can wreak more havoc on your music than a second?hand Chinese cassette in a well?used 4?track!”

[/size]We now know how spot on this was.

For reference here is the waveform image of the last song linked above: http://grooveshark.com/#!/s/Degausser/2gEx83?src=5

BTW. This image was only generated from an analog stream capture from the flash player in the above link.

Edit: @peterh, your two candidates:
http://paste.debian.net/plain/106659
To me it looks almost like "brand new" was artificially produced for some sort of "fake" high LRA.

I know, but I mean how do they get such a high loudness range?

How are they able to start it fairly loud (or at least average volume) then get it far more loud?

I know, but I mean how do they get such a high loudness range? How are they able to start it fairly loud (or at least average volume) then get it far more loud?

I don't know exactly what these waveforms mean since i'm not an expert, but I can see it's very bricked during the loud sections. What about the quieter sections, is that what being bricked looks like too?

Quote

I don't know exactly what these waveforms mean since i'm not an expert, but I can see it's very bricked during the loud sections. What about the quieter sections, is that what being bricked looks like too?

You can keep boosting the volume until every "wave" and every little ripple is "bricked"! 

There is a limit and at some point you simply cannot go louder.  A 2kHz 0dB square wave is probably as loud as you can get.    If you boost a 0dB clipped square wave, absolutely nothing happens (if the result is also clipped at 0dB).  (Your ears are most-sensitive around 2kHz). 

When done properly, limiters are used to "round over" the peaks, so they are not clipped (squared-off).  But if you push compression/limiting hard enough you'll still get clipping.  And the harmonics introduced by clipping make it sound even louder!

If you download Audacity, you can look at some waveforms.  You can zoom-in to see the individual samples and individual "waves" if you wish.  And you can boost the volume to see what happens.  You can use the Amplify effect to boost the volume by more than 50dB!  You'll have to check the box to allow clipping, and the volume won't really be boosted by 50dB because it will get clipped.*  But, you will get an idea of what happens to the waveform and what clipping does to the sound.  (You may need turn-off "Show Clipping" if you boost volume a lot, and you still want to see the waveform.  "Show Clipping" makes the display red so you can see where the clipping is, but if there's a lot of clipping it makes it harder to see the waveform.)

Audacity also has compressor & hard limiter effects, or you can download optional compressors & effects.  The hard-limiter is really a "clipper" as it doesn't round-over the waves. 

BTW - You can also find expander effects, or some compressors can be confgured to expand.  Expansion increases the dynamic range by boosting the loud parts and/or reducing the quiet parts.  Expansion is rarely used in audio production except as a special kind of downward-expander called a noisegate.  A noisegate kills the audio completely when the level falls below a preset threshold.

Note that expansion cannot be used to un-do compression, because we never know the compression parameters and often compression is also applied to the individual tracks before mixing.  And, when the signal is limited or clipped there is no way of knowing the original height or shape of the waveform.




*Audacity won't actually clip the waveform because it uses floating-point and there's no limit.  But, your soundcard/driver is limited to 0bBFS, and if you export to a "normal" WAV file, it will be clipped at 0dB.    So depending on your volume setting, you may not hear clipping unless you make a WAV file and play it.  Or, you can set the hard-limiter effect to 0dB, or just below 0dB.

Ok so basically the answer for why it sounds the way it does is a lot of dynamic range compression?

(...) it involves turning down the loudest parts of the song so that all the parts are roughly the same volume, then boosting the overall volume level of everything to get the peaks close to the maximum allowed by the digital format. When you play this audio side-by-side with the original, it's louder, because it really is louder...the overall level is hovering very steadily within a few dB of, say, 3 dB below the digital maximum, rather than fluctuating much more wildly around, say, -10 dB below maximum.

(...) it involves turning down the loudest parts of the song so that all the parts are roughly the same volume, then boosting the overall volume level of everything to get the peaks close to the maximum allowed by the digital format. When you play this audio side-by-side with the original, it's louder, because it really is louder...the overall level is hovering very steadily within a few dB of, say, 3 dB below the digital maximum, rather than fluctuating much more wildly around, say, -10 dB below maximum.