I got home yesterday and felt compelled to test this out.
Test Scenario:
Computer loaded up and ready to go.
ASUS Digital HD Audio Adapter
RCA Amp 600 watts
8ft digital SPDIF audio cable (coaxial)
Some very thick speaker wire
Bose speakers w/ 15" sub in wooden floor enclosure
MP3 256kbps SBD of 3/8/09 (livephish)
FLAC SBD of 3/8/09 (livephish)
Ok, so I ran my sound out of the computer through digital out (coaxial) with my HD sound card. I plugged it into one of the digital coax ins on my amp and cranked the volume to about 75%. I set the amplifier for stereo (better for music) and ripped the FLAC version first. It was LOUD and I could hear everything really well. I listened to Twist > 2001 > Moma. Next, I loaded up the MP3s of the same songs and had a really LOUD descriptive listen as well. I hope I didn’t piss my neighbors too much. So, now that I’ve run ALL DIGITAL sound through a nice amp and bose speakers and listened to one format right after the other, here are the results!
Results:
FLAC Listen: Very few imperfections at all other than general mixing issues including that Mike and Fish are mixed a little low. think I may have heard a peak or two, but nothing really noticable. Overall it sounded great.
MP3 Listen: Everything really sounded great here too. I thought for sure that my argument on the board was sound and that the MP3 sounded great. That was up until the climax of 2001 happened. At sections where there was intense sections with a very heavy waveform, you could hear a tell-tale scratching and sizzling at the most dense sections of sound. I actually went back and put on the same time sequence on FLAC and listened for it and it just simply wasn’t there.
Conclusion:
When in normal listening scenarios, MP3 sounds good enough to be very convincing. In fact, I didn’t notice any problems at all until I had turned it up very loud and arrived at sections of the songs that were very aurally intense. I actually stand by some of my original statements that you cannot tell the difference, but now I only believe it is true in certain scenarios. During the quieter, more chill sections of songs, there was no discernible difference in quality, and no cracks or sizzles. However, at the peaks of the audio, the MP3 showed its true colors.
I concede that Will owned me yesterday ;p
Analysis:
I believe that with proper refining of compression algorithms these problems with peak cracking and sizzling could be avoided completely, or at least enough to where the human ear doesn’t notice at all. I think the problem can be defined easily using a little knowledge about sampling, and a little common sense. At points of the highest waveform density (the loudest, densest sound with the most going on at once) the algorithm is failing because the sampling is causing error much like rounding error. MP3 consists of thousands (if not millions) of short pieces of audio pulled from an original source. These pieces (samples) happen so fast that the majority of the audio remains unscathed. However, at the points where the audio is very intense, it seems that instead of accurately sampling, the algorithm is just translating the wave sample into a peak at that point where one doesn’t really exist. This in turn causes you to hear a scratch or click or sizzle type of sound. So…
nuff said…i lose…buy the FLACs or download AUDs