[BC] Cascading Algorithms

[Kevin Tekel]

Rich Wood wrote:
 > I ask "how long can you listen to CDs with good quality earbuds." The
 > answer from everyone, so far, is some variation of "I can listen for
 > hours."  The second question is "how long can you listen to your MP3
 > player with good quality earbuds.?" The last response I got was
 > "about an hour. It hurts my ears." No discussion of codecs is involved.

That's nothing.  Try asking how long they can listen to XM through
earbuds. In a noisy store when played through cheap demo speakers it
already sounds hideous.  Listening to XM in a quiet environment through
earbuds or headphones would be like torture to me, and would surely give
me a headache after only a few minutes.  Its horrendous SBR "synthetic
treble" has all the mellowness of a swarm of screeching cicadas.  And
since it also uses SBR, IBOC has this same exact problem, especially now
that most FM IBOC stations are multicasting and bitrate of their main
audio channel has been severely compromised.

Sirius may not be aural heaven either, since they're still using the
Lucent PAC codec whose use with IBOC was embarrassingly rejected by the
NRSC, but since PAC does not use "synthetic treble" I actually find Sirius
to be less painful to my ears than XM or IBOC.  Sirius also limits their
treble response to 12 kHz and blends all audio above 1.8 kHz to mono,
which helps to mask a lot of the codec artifacts.  After all, there is no
need to try to squeeze full "digital-quality" stereo audio through a codec
which just can't handle it gracefully.

Another reason why XM sounds so bad is that they are deliberately
transcoding their audio.  Last I heard, their music is stored on their
servers as 384 kbps MP2 files.  That format by itself already has plenty
of audible artifacts, and when you transcode it to ~32 kbps AACplus for
transmission over XM, it surely isn't going to sound any better!  And yet
the RIAA is still getting their knickers in a knot over the concept of XM
subscribers being able to make "CD-quality" digital recordings of the
music that comes over their favorite channels.  Add in another layer of
transcoding when the received XM audio is saved as, say, a 128 kbps MP3
file, and the result is probably so poor that it would be the aural
equivalent of what you see when you try to dub a MacroVision-protected
video tape.