Notes by ff123
The most important thing you need to know when reviewing my listening results is how able am I at detecting artifacts. Closely tied in to that is how keen is my ability to hear very high frequencies. You can gauge my hearing ability for yourself (against your own) by visiting my "Test Your Hearing" pages:
Test your hearing cutoff
Test your hearing, Part 2
Test your hearing, Part 3
The bottom line is that I can tell you that I lack the ability to hear very high frequencies, based on listening comparisons using real music (not just tone sweeps) against other people. There are a couple of possible reasons for this: 1) I am getting older (age 39), and my hearing is starting to show age-related degradation, or 2) my audio equipment is limiting my ability to hear higher frequencies. Of these two reasons, I think that reason 1) is more likely because in listening to a frequency sweep, my six-year old son could hear almost up to 21kHz whereas I could hear only up to 16kHz using the same setup and volume.
But the reasons for my lack of ability in hearing high frequencies aren't important -- it's only important that you know of it. What it means is that there will be cases, especially at higher bitrates, where I won't be able to distinguish between coded and original sounds, even though differences may be apparent to someone with better hearing. While I believe I can readily pick up on midrange artifacts such as swishing and warbling, I won't be able to hear high-frequency anomalies, and it will definitely bias my results. For example, at my page: Test your hearing, Part 2, I demonstrate that I cannot tell the difference between a clip encoded with mp3enc at 128kbs and the original. Yet I know that it is possible to do so because somebody with better hearing than me described what he heard and identified the files correctly.
Now suppose there is a codec which is equal to mp3enc in every way except that it has a better high frequency response above 16kHz. I wouldn't be able to tell the difference. Conversely, if the frequencies above 16kHz are mucked up, I wouldn't be able to tell you anything about that either. Interestingly, the latter case may be exactly what's happening with FhG's older "Alternate" codec, for example as implemented by MusicMatch Jukebox 5 "very high" setting. On the clips I've encoded with it so far, everything sounds fine to me. But others say that it sounds horrible (see my Bug in older FhG "Alternate" Codec page).
So what does this all mean? I think that at 128kbs I'll be pretty close to the mark, since most codecs have significant midrange artifacts at this bitrate which I can pick up on. Plus the cutoff of most codecs at 128kbs is right around 16kHz. So the tradeoff between lack of midrange artifacting and wider frequency response will probably tilt mostly in favor of the former (i.e., lack of artifacting is better than wider frequency response) [see addendum below, however]. At the higher bitrates, my listening tests will mean less and less. At 192kbs and above, for example, I won't be able to properly tell you what codec sounds best, since most codecs are already removing the vast majority of midrange artifacts, making it mostly a contest of which codec has the best high-frequency response. For these listening tests, you'll have to rely upon somebody else.
Addendum (2-4-01): The listening experience with the Test For Ringing has taught me that even at 128 kbit/s, I may lack the high-frequency ability necessary to properly evaluate the quality of an mp3 codec.