Perceptual test results for the HDC codec have been reported on the NPR Labs website, and they do show that some listeners can tell a difference between 48kbps and 96. They also show that 96 is perceptually fairly close to CD.
www.nprlabs.org/public/research.php

However various other tests have not shown lossy formats to be close to transparent at either 48 or 96.
wiki.hydrogenaudio.org/index.php?title=Hydrogenaudio_Listening_Tests

A recent independent test was commisioned by NPR and was reported in an IEEE journal The IEEE Transactions on Broadcasting, Vol 52 #2, Results from subjective testing of the HD codec at 16-96 kbps. The results have not been posted by NPR labs. Could any kind person who has access to this journal, summarize the results. They are not summarized in the abstract. Or if there is a preprint online could anyone point out where it is located?

I'm not surprised that "other tests of lossy formats" have shown that listeners can tell a difference at low bitrates. More modern codecs like aac+ and it's cousin HDC are far more efficient than older ones like mp3, so they sound better at low bitrates. Still, purist formats like classical and other forms of acoustic music should probably forget multicasting, and concentrate on a pure 96kbps stream.

A good rule of thumb...ask yourself "what are people likely doing when they listen to our station". If it's driving, working at their desk, or listening in the background while reading, then by all means, multicast away. But if you have a "serious" music format, where people are listening intently to the music, that's a different story.

Unfortunately there have been few reports on AAC+, and no public tests of HDC. The big problem with rushing into HDC is that there may be a new idea for coding that makes 48k completely transparent, but that requires specific decoding. HDC might not be released for public use, so that it may not be possible to have public tests. This would of course cast doubts that it works as well as claimed. There might eventually be non professional implementations of HDC. But the big problem is that different implementations can sound quite different.

Incidentally "serious" music is considered to be less taxing on the formats than dense popular music. But people will listen to "serious" music more critically and may be less tolerant of artifacts. Actually the most taxing source seemed to be the human voice, and people speaking tended to be the biggest problem.

Unfortunately John that's always true. Another better technology may always be (and usually is) "just around the corner". But acting as if we believe that means NO NEW TECH PRODUCTS EVER, because we're always waiting for the "next thing", which undoubtedly will be better.

Don't despair, however...even if the receiving end is "locked", lots of improvements can be made to HD audio at the transmitting end with upgraded firmware, software, and hardware, improving things without consumers having to buy anything else.

Actually the easiest thing to detect artifacts on is usually voice with no music. Complexity in music actually makes artifacts LESS easy to hear, because there are more musical "strands" behind which they can hide. But a person speaking into a microphone in a dry acoustic...THAT gives artifacts nowhere to hide.

Remember that HD Radio really did come about by waiting for a better codec. We could have adopted Eureka-147 like the rest of the world ten years ago but that requires 192 Kbps to give good quality audio and 256 Kbps for excellent audio. They really did wait for better technology to come around and it paid off.

Of course HDC required the development of some new ideas in codecs. Fortunately these were incorporated in the final form. But now my original question: Does anyone have access to the IEEE journal and the recent testing? And would that someone be willing to discuss some of the results?

BTW most testing by NPR labs involved just testing the codec. But HD may come out better if it is tested against an actual FM transmission.