Image Processing Reference
Caveats Regarding Surround Sound
Over-compression of the audio with normal compression tools will lose any embedded
surround-sound information. The consequence is that sound will come randomly from
any speaker and may jump around as the compression factor changes. You may also hear
strange bumps and whining noises from speakers driven by the surround circuits. This is
due to the severe corruption of the surround-sound signal. In some formats, this is an ana-
log-coded frequency or phase modulation that is outside of the normal range of human
hearing and therefore is outside the range that some audio codecs are optimized for.
The Fraunhofer Institute has developed some very innovative approaches to compression,
especially in the area of MPEG-4 and the related coding techniques for audio and video.
At the IBC 2004 conference, they showed how the coding of surround-sound audio could
be accomplished at 48 Kbps. This has profound implications.
The quality delivered is superior to that available from a matrixed analog audio sig-
nal. This announcement is innovative because it broadcasts a compatible signal that can
be played back on both stereo and mono systems.
It is based on coding a down-mix as a mono audio signal and then adding difference
information that can be cascaded onto that signal when it is decoded. You simply stop decod-
ing when you have extracted the number of channels you need. It's a very clever solution.
Fraunhofer has also developed an MP3 -based surround audio solution but while this
is interesting, the AAC -based techniques will likely deliver a longer-term solution.
Pitfalls and Gotchas
Audio is much simpler than video, but there are still some nasty surprises for the unwary.
CD Versus DVD Sample Rates
CD audio is sampled at a rate of 44.1 KHz. DVD audio is sampled at a rate of 48 KHz and
the difference is noticeable if the audio is clocked at the wrong rate. It also will not sync to
your video. Converting from one to the other introduces audible artifacts at a frequency
that is the difference between the two. A 3.9 KHz buzzing is very annoying if your audio
is not resampled properly.
You need to know what the target output platform is for your program mate-
rial. If you are planning to output to a DVD format, remember that the audio
sample rates are not the same as on CD audio discs.