Here's a simple trick I found (but didn't originally discover) to compare the difference between a lossless file and a compressed version. This isn't just limited to MP3s - any audio compression that doesn't stretch or squash the audio (gaps at the beginning and end are fine) can be tested with this technique provided that you have the original.
What this does is cancel out everything except for the differences in the two files by inverting one and mixing them after they're both lined up. Anything that's not present or has been added during compression can be heard (not either or, as that's mathematically impossible). I used Adobe Audition to do this, but Audacity works fine as well if you can't afford decent software.
I used a slightly sped-up version of the song "The Bad Touch" because it was easy to line up the lossless and lossy versions, and it has both high and low frequencies. Overall a great example. The lossy version of the file was encoded at 192Kbps using Adobe Audition 3.0.1.
Additionally, I normalized the mixed track in order to allow it to be heard better, which as pointed out was not a very technical thing to do. But the purpose of this video is a tutorial so you can do this yourself and experiment more, allowing you to hear it at its original volume or whatever your heart desires.
Lastly (for now), someone pointed out that this is called null testing. I of course knew that my method wasn't new at all, but it was fun to come up with this on my own and figure out what can be done with it. I'm not an audiophile, and I don't claim to know everything about what I'm doing.
Addendum:
YouTube does use compression so this isn't a fair comparison, but is more intended as a demonstration of a technique you can try on your own. Still, just for fun, I did try comparing the two after YouTube compressed the video's audio.
Here's the original audio's spectrograph as encoded by YouTube : https://drive.google.com/file/d/0BwhT...
And the MP3 version, again as YouTube encoded it: https://drive.google.com/file/d/0BwhT...
And you can compare those to how it looks in the video. Obviously the high frequencies are cut off and there is some clipping that was introduced apparently, but there are still noticeable differences between the two. Probably not audible differences I might add.
Just in case anyone is wondering, get the video's audio to make those images, I used a service that downloads the direct audio stream from YouTube to an M4A file (no re-compression, just the straight audio data from YouTube ), then used NeroAAC to decode that into a .wav file. No transcoding used, let alone transcoding to MP3.
