And I really like your point about the third-octave analyser. Have you actually done this? What differences do you see, that is, do you lose frequency response on the low end, or the high end, or is the loss randomly distributed? How about at different encoding rates -- say AudioCatalyst VBR at the High quality setting?

I've seen the graph outputs of these tests, and they all show what our ears already hear: The differences get worse as you get closer to the high frequencies. They also show that as you get past 128kbps encoding rates, the differences start to decrease to negligible levels. Almost all MP3 encoders roll off the ultra-high frequencies above 16khz no matter what encoding rate you use, but as long as you're still in the non-canine range, the differences are negligible at rates of 160kbps and above. Honestly, though, trained human ears are infinitely more sensitive than one of those graphs. A good audiophile will hear differences that those graphs won't show.

I've never seen one of those graphs comparing VBR with CBR though. They always seem to do the comparisons as apples-to-apples constant bitrates between different encoders as compared to the original waveform.

I have gotten conflicting reports about whether an MP3 recorded at a high bitrate can be made indistinguishable from the original CD. The consensus of opinion from empeg users seems to be that it can be done with the empeg, which seems to have unusually high quality audio output.

Keep in mind that there is a difference between the sound output quality of your playback hardware, and the compression quality of your MP3 encoder. Those are two completely different and unrelated steps in getting the sound to your ears.

The Empeg has great audio output, but it's still at the mercy of your MP3 encoder. It's definitely a GIGO system.

The test graphs I refer to are completely independent of any audio output hardware, they only tested the encoder. How do they do this? By working only with the wave data directly on the hard disk. It goes like this:

- Original wave file is analyzed.
- Wave file is used as the source for the MP3 compression.
- Compressed MP3 file is "played to disk" using the WinAmp disk writer plug in. It goes through all the decompression steps, but the output is a new disk file instead of the sound card.
- The resulting "play to disk" wave file is analyzed the same way the original wave file was analyzed, and the results are compared graphically.

This way, you get an apples-to-apples comparison, so to speak, without the audio amplification hardware muddying the waters.

___________
Tony Fabris